1

Modul 9B

Numerical
Evaluation of
Dynamic Response
Patria Kusumaningrum, ST, PhD
SI 3213 Dinamika Struktur
Fakultas Teknik Sipil dan Lingkungan & Rekayasa Gempa
2
Materi Term 2
Mg. Topik Sub Topik
General Dynamic Loading Response of SDOF system excited by impulsive loading. Numerical
9
and the Numerical Evaluation solution including Duhamel’s & step by step integration method
Earthquake Resistant General concept of earthquake resistant buildings and the
10 importance of ductility factor
Building
11 Response Spectrum Analysis Concept and application of Response Spectrum (RS) analysis
Assembling of mass and stiffnes matrix, eigen vectors, eigen
12 MDOF values
MDOF: Modal Superposition
General concept of using modal superposition analysis to
13 Analysis and Normalization determine the responses of MDOF system
Mode of Vibration
General concept of using CQC and SRSS method, to determine the
14 MDOF: Maximum Response responses of MDOF system
General concept and application of Spectrum Response analysis
15 MDOF: RSA and ELF and Equivalent Lateral Force (ELF) procedure.
16 UAS
3

Numerical Methods
4

Numerical Methods

There are a variety of ways to solve the general loading • Fourier transform*
problem and all are carried out numerically on the • Duhamel integration*
computer. The Fourier transform is not particularly • Direct Integration
efficient (or easy to explain) and, hence, this is not
• Piecewise exact
covered here. Any text on structural dynamics will
• Newmark techniques
provide the required details.
The piecewise exact method is used primarily in the
• Central Differences
analysis of linear systems. The Newmark method is • Wilson -
useful for both linear and nonlinear systems. All techniques are
carried out numerically.
5
Why Numeric?

Analytical solution of the equation for a SDOF System is usually not

possible if the excitation – applied force p(t), or ground acceleration –

varies arbitrarily with time or if the system is nonlinear

6
Duhamel Integral
Duhamel’s Integral – Undamped System
• An impulsive loading is a load which is applied during a short duration of
time.
• The corresponding impulse of this type of load is defined as the product
of the force and the time of its duration.

Using Duhamel’s Integral, the

F(t) displacement response at time t
due to continuous F(t) is :
t
1
u t   F t  sin n t  t dt
F(t)∙dt

mn 0 (1)

t
t
t + dt
7

Direct Integration Method

8
Direct Integration of the Equations of Motion

Provides the response of system at discrete intervals of time (which

are usually equally spaced).

A process of marching along the time dimension, in which the

response parameters (i.e., acceleration, velocities & displacements at
a give time point) are evaluated from their known historic values.

Two of these equations are usually derived from

For a SDOF system, this requires assumptions regarding the manner in which the
three equations to determine three response parameters vary during a time step.
The third equation is the equation of motion written
unknowns. at a selected time point.
9
Direct Integration of the EoM
• When the selected point represents the current time (n), the method of
integration is referred to as Explicit method (e.g. Central difference
method).

• When the equation of motion is written at the next time point in the
future (n+1), the method is said to be Implicit method (e.g. Newmark’s β
method, Wilson – θ method).
10
Direct Integration of the EoM

.. .
For SDOF System mu + cu + ku = P(t) P(t)

Let ∆t = time interval Pn Pn+1

tn = n ∆t

. ..
un, un, un Displacement, velocity and
acceleration at time station ‘n’ ……
0 1 ∆t tn tn+1
Pn is the applied force at time tn
11
General Expression for Integration Methods
n n +1 . n +1 ..
u n +1   A u +B
l l l ul +Cu l l +R (1)
lnk lnk lnk

R is a remainder term representing the error given by Taylor Series

Em (m)
u () (n-k)∆t ≤ ξ ≤ (n+1) ∆t
m!

u(m) is the value of mth differential of u at t=ξ

Al, Bl and Cl are constants (some of which may be equal to zero)
12
Direct Integration
. ..
 Eq(1) relates un, un, un @ tn+1 to their values at the previous time stations n-k, n-
k+1, ………n
 Eq(1) has m = 5+3k undetermined constants A, B and C.
 The equation is employed to represent exactly a polynomial of order p-1, p
being smaller than m.
 Then (m-p) constants become available which can be assigned arbitrary
chosen values so as to improve stability or convergence characteristics of the
resulting formula.
 Formulas of type eq(1) for time integration can also be obtained from physical
considerations, such as, for example, an assumed variation in the acceleration
or from the finite difference approximations of the differentials.
13

Piecewise Exact Method

14

Piecewise Exact Method

Pi+1
dP
P (t ) P+
i
t
Pi
dt
Actual: P(t)
dF Interpolated: P (t )

dt
t
dt
15
Piecewise Exact Method
Initial conditions uo,0  0 uo,0  0
Determine “exact” solution for 1st time step
u1  u(t )

Establish new initial conditions

LOOP

Obtain exact solution for next time step

16

Piecewise Exact Method

Advantages:

• Exact if load increment is linear

• Very computationally efficient
Disadvantages:

• Not generally applicable for inelastic behavior

Note: NONLIN uses the piecewise exact method

for response spectrum calculations.
17
Piecewise Exact Method
18
Piecewise Exact Method

pi
pt   pi + t
ti
pi  pi +1  pi
19

Newmark’s β Method
20
Newmark’s β Method xn+1
xn

• In 1959, Nathan Newmark devised a series of numerical integration ∆t

formulas collectively known as Newmark’s β methods.

• The velocity expression is of the form (1)

• The displacement expression is of the form

(2)

To determine the constants, make equations (1) & (2) for u=1, u=t, u=t2, we get
a1=1, 2∆t = 2a2+2a3
b1=1, b2=∆t, 2b3+2b4=(∆t)2

Say a3=γ∆t & b4=β(∆t)2

21
Newmark’s β Method
Then equations (1) & (2) reduce to
(3)
(4)
Third relationship
(5)

..
Substituting eqn.(3) and (4) in eqn.(5), we get expression for un+1
. ..
To begin the time integration, we need to know the values of uo, uo and uo
at time t=0.
22
Constant Acceleration
γ=0, β=0

t
 0.636 (stable)
T

..
..
Acceleration,u

un

∆t ∆t Time, t
23
Average Acceleration

γ=1/2, β=1/4

.. .. ..
un+1 .. u n + u n +1
..
Acceleration,u

.. u=
un 2

∆t Time, t
24
Linear Acceleration
γ=1/2, β=1/6

t
 0.55 (stable)
T

..
un+1
..
Acceleration,u

.. .. .. t .. ..
un u = un + (un+1  u n )
t

t
∆t Time, t
25
Newmark’s β Method
26
Algorithm
Enter k, m, c, β, γ and P(t)

.
.. P(t  0) - c u 0  k u 0
u0 =
m

Select ∆t

c m
k^ k + +
βΔ t β(Δt) 2

m c m 
a + ; b  + Δt(  1)c
β Δt β 2β 2β
27
Algorithm (2) i=0
.
^=p+au .. i = i+1
∆p i i i + b ui

 p^i
∆ui =
^
k

.
 ui 
 u i
  u. i
+  t(1   )u
..
t. 2 i

. ..
.. u ui u
 u i   i 2  i
(t)   t 2

. . . .. .. ..
ui+1= ui+ ∆ui ; ui+1= ui+ ∆ui ; ui+1= ui+ ∆ui
28
Elastoplastic System: Newmark’s β Method
Enter k, m, c, Rt, Rc and P(t)

.
Set u0= 0, u0 = 0; Select ∆t

.. P(t  0) ; Rt
; uc = R c
u0 = ut =
m k k

Define key = 0 (elastic)

key = -1 (plastic behavior in compression
key = 1 (plastic behavior in tension)
29
i = i+1
i=0
Calculate ui and u.i

u.i > 0 y key = 0; ut= ui; uc= ui – (Rt – Rc)/k

R = Rt – (ut – ui) k

n
key = 0; uc= ui;
. y
ui < 0 ut= ui + (Rt – Rc)/k
R = Rt – (ut – ui) k

ui > u c y
R = Rt – (ut – ui) k
ui <u t
n
ui < u c y
key = -1; R=Rc
n
y .. .
ui >u t key = 1; R=R t ui+1 = (P(ti+1) – ci+1ui+1– R) /m
30

Central Difference Method

31
Central Difference Method
The method is based on finite difference approximations of the time derivatives of
displacement (velocity and acceleration) at selected time intervals

xn+1
Displacement, u

xn-1 θ
1
  ( x n +1  x n )
2h

(n-1)∆t (n+1)∆t Time, t

32
Central Difference Method
33
Central Difference Method
34
Algorithm
Enter k, m, c, and P(t)

.
.. P(t  0) - c u 0  k u 0
u0 =
m

. ..
u-1 = u0 - ∆t u0 + 0.5 ∆t2 u0

m c
k^ 
(t) 2 2t

m c 2m
a  ; b  k  ;
2 2
(t) 2t (t)
35
Algorithm (2)
i=0
i = i+1
p^i= pi – a ui-1 – b ui

p^i
ui+1 =
k^

. u i+ 1 - u i-1
ui 
2t

.. u i+ 1  2u i +u i-1
ui 
(t) 2
36

Wilson - θ Method
37
Wilson-θ Method
..
u..n+θ
un + 1
..
Acceleration, u
..

un

∆t Time, t
 ∆t
• This method is similar to the linear acceleration method and is based on the
assumption that the acceleration varies linearly over an extended interval θ∆t.

• θ, which is always greater than 1, is selected to give the desired characteristics

of accuracy and stability.
38
Algorithm
Enter k, m, c,  , ∆t and P(t)

Specify initial conditions

6 θΔt 3 6
a 1 ; a2  ; a3  ; a4 
(Δt)2 2 (  Δ t) (  Δ t)

n=0

pn+θ = pn (1- θ) + pn+1 θ

^k= a m + a c + k ; . .. . ..
1 3
a5 = a1un + a4un + 2un ; a6 = a3un + 2un + a2un
. A
39
Algorithm
^
un+θ= ( pn+θ+ma5 +ca6 ) /k

.. (θ Δt) 2 ..
un+θ = un }
3

.. .. .. ..
un+1 = un+ (un+θ – un) /θ

. . .. ..
un+1 = un+ (un + un+1) h/2 A

n = n+1
u n+1  u n
40

Selection of Numerical Integration Method:

Error, Stability and Time Step
41
Errors involved in the Numerical Integration
• Round off errors
Introduced by repeated computation using a small step size.
• Random in nature
• To reduce use higher precision
• Truncation errors
Involved in representing xn+1 and xn+1 by a finite number. of terms in the Taylor’s
series expansion.
• Represented by R in the previous slides
• Accumulated locally at each step.
• If integration method is stable, then truncation error indicates the accuracy.
• Propagated error
Introduced by replacing the differential equation by a finite difference equivalent.
42
Stability of the Integration method
• Effect of the error introduced at one step on the computations at the next
step determines the stability.
• If error grows, the solution becomes unbounded and meaningless.

• Spectral radius of a matrix ρ(A) = max of (magnitude of eigen values of A)

.
• A is ‘amplification matrix’
un+1  un 
 .   . 
 t un+1   [A] t un 
(t)2 ..u  (t)2 ..u 
 n+1   n 

ρ(A) > 1 Unstable

If θ ≥ 1.37 Wilson- θ is unconditionally stable

43
Attributes required
for good Direct Integration method

1. Unconditional stability when applied to linear problems

2. Not more than one set of implicit equation to be solved at each step

3. Second order accuracy

4. Controllable algorithmic dissipation in the higher modes

5. Self starting – Wilson-θ is reasonably good

* For MDOF systems, scalar equations of the SDOF systems become matrix
equations.
44

Spectral radii for α-methods, optimal collocation schemes and

Houbolt, Newmark, Park and Wilson methods
45
Selection of a numerical integration method

Period elongation vs. ∆t/T Amplitude decay vs. ∆t/T

* For the numerical integration of SDOF systems, the linear

acceleration method, which gives no amplitude decay and the
lowest period elongation, is the most suitable of the methods
presented
46
Selection of time step ∆t

 t must be small enough to get a good accuracy, and long enough to be

computationally efficient.
 p∆t < 1 i.e., ∆t/T ≤ 0.16 (arrived at from truncation errors for a free vibration
case)
 Typically, ∆t/T ≈ 0.1 is acceptable.
 Sampling of excitingfunction at intervalsequal to selected ∆t inspection of
the forcing function.
47

Problem and Solution

48
Recall
Interpolation of Excitation

pi
pt   pi + t
ti
pi  pi +1  pi
49
Problem B: Interpolation of Excitation
An SDOF system has the following properties :
12.00
• m = 4.500 kg-sec2/m 10.00 10.00
8.66 8.66
8.00
• k = 178.400 kgf/m

p(t)
6.00
5.00 5.00
4.00
• Tn = 1 sec (n = 6,283 rad/sec) 2.00
0.00 0.00 0.00
•  = 0,05. 0 0.1 0.2 0.3 0.4 0.5 0.6
t, sec

• Determine the response u(t) of this system to p(t) = 4,500 sin(pt/0,6) kgf,
defined by the half-cycle sine pulse by using piecewise linear
interpolation of p(t) with t = 0,1 sec.
50
Recall: Piecewise Linear Method
51
Solution 1: Piecewise Linear Method
ti pi C.pi D.pi+1 ůi B.ůi ui A.ui C'.pi D'.pi+1 A'.ui B'.ůi
0 0,00 0,0000 0,0008 0,0000 0,0000 0,0000 0,0000 0,0000 0,0237 0,0000 0,0000
0,1 2250,00 0,0016 0,0014 0,0237 0,0021 0,0008 0,0007 0,0216 0,0410 -0,0029 0,0179
0,2 3897,11 0,0027 0,0016 0,0777 0,0070 0,0058 0,0047 0,0375 0,0474 -0,0207 0,0586
0,3 4500,00 0,0031 0,0014 0,1228 0,0111 0,0160 0,0130 0,0433 0,0410 -0,0576 0,0927
0,4 3897,11 0,0027 0,0008 0,1194 0,0108 0,0287 0,0233 0,0375 0,0237 -0,1030 0,0902
0,5 2250,00 0,0016 0,0000 0,0483 0,0044 0,0376 0,0306 0,0216 0,0000 -0,1352 0,0365
0,6 0,00 0,0000 0,0000 -0,0771 -0,0070 0,0365 0,0296 0,0000 0,0000 -0,1312 -0,0582
0,7 0,00 0,0000 0,0000 -0,1894 -0,0172 0,0227 0,0184 0,0000 0,0000 -0,0814 -0,1430
0,8 0,00 0,0000 0,0000 -0,2244 -0,0203 0,0012 0,0010 0,0000 0,0000 -0,0044 -0,1694
0,9 0,00 0,0000 0,0000 -0,1739 -0,0158 -0,0193 -0,0157 0,0000 0,0000 0,0695 -0,1313
1 0,00 0,0000 0,0000 -0,0618 -0,0056 -0,0315 -0,0256 0,0000 0,0000 0,1131 -0,0467
ui in m
52
Recall: Central Difference Method

Central Difference Method

is stable if :
t 1

Tn p
53
Solution 2: Central Difference Method
Solve problem by the central difference method using t = 0,1 sec.

ti pi ui-1 ui pi' ui+1

0 0,00 0,0000 0,0000 0,0000 0,0000
0,1 2250,00 0,0000 0,0000 2250,0000 0,0048
0,2 3897,11 0,0000 0,0048 7395,2181 0,0159
0,3 4500,00 0,0048 0,0159 13884,5060 0,0299
0,4 3897,11 0,0159 0,0299 18538,8179 0,0399
0,5 2250,00 0,0299 0,0399 18033,8488 0,0389
0,6 0,00 0,0399 0,0389 10627,9866 0,0229
0,7 0,00 0,0389 0,0229 -411,7948 -0,0009
0,8 0,00 0,0229 -0,0009 -10620,7730 -0,0229
0,9 0,00 -0,0009 -0,0229 -16125,5428 -0,0347
1 0,00 -0,0229 -0,0347 -15096,8118 -0,0325

ui in m
54
Recall: Newmark Method

Newmark’s method is stable if :

t 1 1

Tn p 2   2
For  = ½ &  = ¼
t

Tn

For  = ½ &  = 1/6

t
 0,551
Tn
55
Solution 3: Newmark – Average Acceleration

ti pi üi pi pi' ui ůi üi ůi ui

0,0 0,0000 0,0000 2250,0000 2250,0000 0,0011 0,0221 0,4423 0,0000 0,0000
0,1 2250,00 0,4423 1647,1143 9733,0524 0,0048 0,0514 0,1441 0,0221 0,0011
0,2 3897,11 0,5864 602,8857 19534,2187 0,0096 0,0449 -0,2748 0,0735 0,0059
0,3 4500,00 0,3116 -602,8857 24190,5526 0,0119 0,0009 -0,6058 0,1184 0,0155
0,4 3897,11 -0,2942 -1647,1143 17855,1439 0,0088 -0,0631 -0,6742 0,1193 0,0274
0,5 2250,00 -0,9684 -2250,0000 -536,5761 -0,0003 -0,1176 -0,4156 0,0562 0,0361
0,6 0,0000 -1,3841 0,0000 -23865,0826 -0,0117 -0,1117 0,5351 -0,0614 0,0359
0,7 0,0000 -0,8490 0,0000 -39778,2274 -0,0195 -0,0447 0,8031 -0,1731 0,0242
0,8 0,0000 -0,0459 0,0000 -40857,8396 -0,0201 0,0341 0,7745 -0,2178 0,0046
0,9 0,0000 0,7286 0,0000 -27549,7397 -0,0135 0,0967 0,4760 -0,1837 -0,0155
1,0 0,0000 1,2046 0,0000 -5320,3417 -0,0026 0,1218 0,0271 -0,0871 -0,0290
56
Solution 3: Newmark – Linear Acceleration

ti pi üi pi pi' ui ůi üi ůi ui

0,0 0,0000 0,0000 2250,0000 2250,0000 0,0008 0,0228 0,4556 0,0000 0,0000
0,1 2250,00 0,4556 1647,1143 14205,5370 0,0048 0,0527 0,1429 0,0228 0,0008
0,2 3897,11 0,5984 602,8857 29786,5145 0,0101 0,0452 -0,2929 0,0755 0,0056
0,3 4500,00 0,3055 -602,8857 37171,5127 0,0125 -0,0010 -0,6307 0,1207 0,0156
0,4 3897,11 -0,3252 -1647,1143 27250,5424 0,0092 -0,0669 -0,6885 0,1197 0,0281
0,5 2250,00 -1,0137 -2250,0000 -1387,4573 -0,0005 -0,1216 -0,4050 0,0528 0,0373
0,6 0,0000 -1,4187 0,0000 -38531,8427 -0,0130 -0,1126 0,5862 -0,0689 0,0369
0,7 0,0000 -0,8325 0,0000 -61880,9015 -0,0209 -0,0406 0,8534 -0,1814 0,0239
0,8 0,0000 0,0209 0,0000 -61540,0822 -0,0208 0,0419 0,7970 -0,2220 0,0030
0,9 0,0000 0,8179 0,0000 -38988,7559 -0,0132 0,1046 0,4559 -0,1801 -0,0178
1,0 0,0000 1,2738 0,0000 -3644,6624 -0,0012 0,1259 -0,0303 -0,0755 -0,0309
57
Plot: Displacement Time History
58
Assignment
An SDOF system has the following properties :
• m = 4.500 kg-sec2/m
• k = 178.400 kgf/m
• Tn = 1 sec (n = 6,283 rad/sec)
•  = 0,05.
Determine the response u(t) of this system due to El-Centro 1940 N-S, using :
a. Interpolation
b. Central Difference
c. Newmark Average Acceleration
d. Newmark Linear Acceleration
Find the umax!
59