Professional Documents
Culture Documents
This document describes the free and forced response of single degree of
freedom (SDOF) systems. A single degree of freedom system is a spring-massdamper system in which the spring has no damping or mass, the mass has no
stiffness or damping, the damper has no stiffness or mass. Furthermore, the
mass is allowed to move in only one direction. The horizontal vibrations of a
single-story building can be conveniently modeled as a single degree of freedom
system. In part 1 of this document we examine some useful trigonometric
identities. In part 2 of this document we determine how damped SDOF
systems vibrate freely after being released from an initial displacement with
some initial velocity. In part 3 of this document we determine how damped
SDOF systems respond to a persistent sinusoidal forcing.
Consider the structural system shown in Figure 1, where:
f (t) = external excitation force
x(t) = displacement of the center of mass of the moving object
m = mass of the moving object, fI = dtd (mx(t))
= m
x(t)
c = linear viscous damping coefficient, fD = cx(t)
x(t)
m
k, c
11111111111
00000000000
00000000000
11111111111
f(t)
11
00
00
11
00
11
00
11
00
11
00
11
k
m
c
f(t)
x(t)
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
T (x, x)
=
(1)
(2)
(3)
T (x, x)
V (x)
+
+ p(x, x)
=0,
dt x
x
x
(4)
F = 0 : fI + fD + fS = f (t) .
(5)
Either way,
m
x(t) + cx(t)
+ kx(t) = f (t),
x(0) = do , x(0)
= vo
(6)
(7)
(8)
Proof:
X eit + X eit = (A + iB) (cos(t) + i sin(t)) +
(A iB) (cos(t) i sin(t))
(9)
(11)
= a cos(t) + b sin(t)
(12)
Comparing these forms, we see that a = 2A and b = 2B. Note that all of
the above expressions are exactly equivalent. Equation (7), is exactly the same
as equation (8). Equation (7) is easier to interpret as describing a sinusoidal
oscillation, however equation (8) is much easier to work with, mathematically.
We will endeavor to use both forms in this document, just to emphasize how
the two forms are one and the same.
Equations (7) and (8) describe a sinusoidal oscillation with a constant
of the oscillation x(t) can be found by adding
amplitude. The amplitude, X,
the magnitudes of the complex amplitudes X and X , or by solving x(
t) = 0
for t, and substituting into equation (7). Either way, the amplitude of the
oscillation is
= |X| + |X | = 2|X| = a2 + b2 ,
X
(13)
The oscillation attains its maximum and minimum values at times t, where
tan( t) = b/a. The oscillation attains a value of zero at times t where
CC BY-NC-ND H.P. Gavin
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
period, T
amplitude, |X|
response, x(t)
4
2
0
-2
-4
-6
0
10
time, t, sec
tan( t) = b/a. The oscillation can be expressed with a single sine term
with a phase-shift.
sin(t s )
x(t) = X
sin(t) cos(s ) X
cos(t) sin(s )
= X
(14)
(15)
cos(t c )
x(t) = X
cos(t) cos(c ) X
sin(t) sin(c )
= X
Setting x(t) = 0 results in tan( t) = cot(c ), and,
a
tan(c ) =
b
(17)
(18)
(19)
Note, again, that equations (7) and (8) are equivalent to one-another and
s ,
are also equivalent to equations (14) and (17) using the definitions for X,
and c given above.
1.2 Decaying Amplitude
which decays with time. Here we will use a real exponential, et , where < 0.
Multiplying equations (7) through (8) by et ,
x(t) = et (a cos(t) + b sin(t))
(20)
= et (Xeit + X eit )
(21)
= Xe(+i)t + X e(i)t
(22)
= Xet + X e
(23)
Again, note that all of the above equations are exactly equivalent. The exponent is complex, = + i and = i. If is negative, then these
equations describe an oscillation with exponentially decreasing amplitudes.
Note that in equation (20) the unknown constants are , , a, and b. Angular frequencies, , have units of radians per second. Circular frequencies,
f = /(2) have units of cycles per second, or Hertz. Periods, T = 2/,
have units of seconds.
In the next section we will find that for an un-forced vibration, and
are determined from the mass, damping, and stiffness of the system. We
will see that the constant a equals the initial displacement do , but that the
constant b depends on the initial displacement and velocity, as well mass,
damping, and stiffness.
6
response, x(t)
|X| e
period, T
2
0
-2
-4
-6
0
10
time, t, sec
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
x(0) = do , x(0)
= vo ,
(24)
(25)
(m2 + c + k)Xet = 0 ,
(26)
1,2
u
c
c
=
t
2m
2m
!2
k
.
m
(28)
1,2 = i k/m = in .
(29)
(31)
cc = 2 mk .
The ratio of the actual damping rate to the critical damping rate is called
the damping ratio, .
c
=
.
(32)
cc
The two roots of the quadratic equation are real and are repeated at
(33)
1 = 2 = c/(2m) = cc /(2m) = 2 mk/(2m) = n ,
and the two basic solutions are equal to each other, e1 t = e2 t . In order
to admit solutions for arbitrary initial displacements and velocities, the
solution in this case is
x(t) = x1 en t + x2 t en t .
(34)
where the real constants x1 and x2 are determined from the initial displacement, do , and the initial velocity, vo . Details regarding this special
case are in the next section.
Case 3 c > cc over-damped
If the damping is greater than the critical damping, then the roots, 1
and 2 are distinct and real. If the system is over-damped it will not
oscillate freely. The solution is
x(t) = x1 e1 t + x2 e2 t ,
(35)
which can also be expressed using hyperbolic sine and hyperbolic cosine
functions. The real constants x1 and x2 are determined from the initial
displacement, do , and the initial velocity, vo .
CC BY-NC-ND H.P. Gavin
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
x(t) = X et + X e t ,
(36)
We can re-write the dynamic equations of motion using the new dynamic
variables for natural frequency, n , and damping ratio, . Note that
v
u
c
k
1
c
k
c u
k
= c = = 2 t = 2n .
m
k m m
k m m
2 km m
m
x(t) + cx(t)
+ kx(t) = f (t),
c
k
1
x(t) + x(t)
+ x(t) =
f (t),
m
m
m
1
x(t) + 2n x(t)
+ n2 x(t) =
f (t),
m
(37)
(38)
(39)
(40)
The expression for the roots 1,2 , can also be written in terms of n and .
v
u
1,2
u
c
c
=
t
2m r 2m
!2
k
,
m
= n (n )2 n2 ,
q
= n n 2 1 .
(41)
(42)
(43)
= Im
1 x
1 + 2 = 2n
1 2 = 2 n 2 1
n2 = 14 (1 + 2 )2 14 (1 2 )2
n = 1 2
= (1 + 2 )/(2n )
2 x
= Re
The solution to a homogeneous second order ordinary differential equation requires two independent initial conditions: an initial displacement and
an initial velocity. These two initial conditions are used to determine the
coefficients of the two linearly independent solutions corresponding to 1 and
2 . If 1 = 2 , then the solutions e1 t and e2 t are not independent. In fact,
they are identical. In such a case, a new trial solution can be determined as
follows. Assume a new solution of the form
x(t) = u(t)x1 e1 t ,
(44)
1 t
x(t)
= u(t)x
+ u(t)1 x1 e1 t ,
1e
(45)
1 t
x(t) = u(t)x1 e1 t + 2u(t)
+ u(t)21 x1 e1 t
1 x1 e
(46)
=0
which is a first order ordinary differential equation for u(t).
The solution of
this ordinary differential equation is
u(t)
=C ,
CC BY-NC-ND H.P. Gavin
10
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
2 1 is imaginary, and
1,2 = n in | 2 1| = i.
(48)
d = n | 2 1| .
(49)
(50)
= Xet + X e t .
(51)
Now we can solve for X, (or, equivalently, A and B) in terms of the initial
conditions. At the initial point in time, t = 0, the position of the mass is
x(0) = do and the velocity of the mass is x(0)
= vo .
x(0) = do = Xe0 + X e
(52)
= X + X
(53)
= (A + iB) + (A iB) = 2A = a.
(54)
x(0)
= vo = Xe0 + X e 0 ,
(55)
= X + X ,
(56)
(57)
= A + id A + iB d B +
A id A iB d B,
(58)
= 2A 2B
(59)
= n do 2 d B,
(60)
CC BY-NC-ND H.P. Gavin
11
vo + n do
2d
vo + n do
d
(61)
(62)
= vo is
x(t) = e
n t
vo + n do
do cos d t +
sin d t .
d
!
(63)
vo
6
do
response, x(t)
|X| e
- n t
2
0
-2
-4
-6
0
10
time, t, sec
1,2 = n n 2 1 = d .
(64)
= vo into the
solution (equation (35)), and solving for the coefficients results in
vo + do (n + d )
,
2d
= do x1 .
x1 =
(65)
x2
(66)
CC BY-NC-ND H.P. Gavin
12
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
Substituting the hyperbolic sine and hyperbolic cosine expressions for the
exponentials results in
n t
x(t) = e
vo + n do
do cosh d t +
sinh d t
d
(67)
The undamped free response can be found as a special case of the underdamped free response. While special solutions exist for the critically damped
response, this response can also be found as limiting cases of the underdamped or over-damped responses.
vo
7
6
do
response, x(t)
5
4
3
=5.0
2
over damped
=1.5
=2.0
critically damped
0
-1
0
0.5
1.5
2
time, t, sec
2.5
3.5
(68)
(69)
(70)
CC BY-NC-ND H.P. Gavin
13
x1
en t1
en t1
2n/ 1 2
= (t +2n/ ) = t 2n / = e
,
d
n
d
x1+n
e n 1
e n 1e
which is independent of n and d .
log(x1 /x1+n )/n,
(71)
() =
2
1 2
(72)
and, inversely,
(73)
2
4 2 + 2
where the approximation is accurate to within 3% for < 0.2 and is accurate
to within 1.5% for < 0.1.
() =
2.6 Summary
k
m
c
c
= =
cc
2 mk
n =
d = n | 2 1|
= log(x1 /x1+n )/n
() = 2
2
4 + 2
14
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
If the SDOF system is dynamically forced with a sinusoidal forcing function, then f (t) = F cos(t), where is the frequency of the forcing, in radians
per second. If f (t) is persistent, then after several cycles the system will respond only at the frequency of the external forcing, . Lets suppose that
this steady-state response is described by the function
x(t) = a cos t + b sin t,
(74)
x(t)
(75)
(76)
then
and
(a sin t + b cos t) +
(a cos t + b sin t) = F cos t.
(77)
(78)
(m 2 b ca + kb) sin t = 0,
(79)
CC BY-NC-ND H.P. Gavin
15
which is a set of two equations for the two unknown constants, a and b,
a
F
k
m
c
=
,
(80)
2
b
0
c
k m
for which the solution is
c
(k m 2 )2 + (c)2
k m 2
b() =
(k m 2 )2 + (c)2
a() =
(81)
F .
(82)
X
= a2 + b2 . The angle c is the
The amplitude of this oscillation is X,
b sin t
X
t
a cos t
Figure 6. The amplitude of the sum of two oscillations in quadrature.
phase between the force f (t) and the response x(t), and
cw
a
(84)
tan c = =
b
k m 2
Note that c is negative, regardless of frequency, meaning that the response
to the forcing
always lags the force. The ratio of the response amplitude X
amplitude F is
X
X
1
= q
.
(85)
=
F
F
(k m 2 ) + (c)2
depends on the amplitude
This equation shows how the response amplitude X
of the forcing F and the frequency of the forcing , and has units of flexibility.
Lets re-derive this expression using complex exponential notation. The
equations of motion are
1
1
m
x(t) + cx(t)
+ kx(t) = F cos t = F eit + F eit .
(86)
2
2
CC BY-NC-ND H.P. Gavin
16
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
In a solution of the form, x(t) = Xeit + X eit , the coefficient X corresponds to the positive exponents (positive frequencies), and X corresponds
to negative exponents (negative frequencies). Positive exponent coefficients
and negative exponent coefficients may be found separately. Considering the
positive exponent solution, the forcing is expressed as F eit and the partial
solution Xeit is substituted into the forced equations of motion, resulting in
(m 2 + ci + k) X eit = F eit ,
(87)
from which
X
1
=
,
F
(k m 2 ) + i(c)
which is complex-valued. This complex function has a magnitude
X
F
1
,
(k m 2 )2 + (c)2
(88)
(89)
1
1
=
=
X
X
=
xst
1/k
,
+ i kc
1/k
m 2
k
2 !
+i
(90)
2 n
1/k
,
(1 2 ) + i (2)
F /k
q
(1 2 )2 + (2)2
1
(1 2 )2 + (2)2
,
(91)
(92)
,
(93)
(94)
where the frequency ratio is the ratio of the forcing frequency to the natural
frequency, = /n , and the static deflection xst is the response to a static
load F , F = kxst . This equation is called the dynamic amplification factor.
CC BY-NC-ND H.P. Gavin
17
=0.05
| X / Xst |
8
6
0.1
4
0.2
0.5
1.0
0
0
0.5
phase, degrees
1.5
2.5
2.5
=0.05
-45
-90
=1.0
-135
-180
0
0.5
1.5
frequency ratio, = / n
st , equation (94).
Figure 7. The dynamic amplification factor for external forcing X/x
It is the factor by which displacement responses are amplified due to the fact
that the external forcing is dynamic, not static. See Figure 7.
To summarize, the steady state response of a simple oscillator directly
excited by a harmonic force, f (t) = F cos t, may be expressed as
F /k
2
cos(t
+
)
,
tan
=
(95)
x(t) = q
c
c
1 2
(1 2 )2 + (2)2
or, equivalently,
x(t) =
F /k
[ (1 2 ) cos t + (2) sin t ] ,
2
2
2
(1 ) + (2)
(96)
where = /n .
3.2 Support Acceleration Excitation
When the dynamic loads are caused by motion of the supports (or the
ground) the forcing on the structure equals the mass of the structure times
the ground acceleration, f (t) = m
z (t).
CC BY-NC-ND H.P. Gavin
18
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
x(t)
111
000
000
111
000
111
11
00
00
11
00
11
00
11
00
11
00
11
00
11
00 z(t)
11
m
k, c
z(t)
k
m
c
x(t)
(97)
(98)
(99)
If the ground displacements are sinusoidal z(t) = Z cos t, then the ground
2 cos t, and f (t) = mZ
2 cos t. Using the
accelerations are z(t) = Z
complex exponential formulation, we can find the dynamic amplification factor as a function of the frequency of the ground motion, .
1
1
m
x(t) + cx(t)
+ kx(t) = mZ 2 cos t = mZ 2 eit + mZ 2 eit (100)
2
2
Again assuming a solution of the form x(t) = Xeit the dynamic amplification
factor is
X
m 2
=
,
Z
(k m 2 ) + i(c)
2
=
(1 2 ) + i(2)
(101)
and
X
Z
2
(1 2 )2 + (2)2
(102)
See Figure 9.
CC BY-NC-ND H.P. Gavin
19
=0.05
|X/Z|
8
6
0.1
4
0.2
0.5
1.0
0
0
0.5
phase, degrees
1.5
2.5
2.5
=0.05
-45
-90
=1.0
-135
-180
0
0.5
1.5
frequency ratio, = / n
Z,
equation (101).
Figure 9. The dynamic amplification factor for base-excitation X/
transimissibility: Z to (X+Z)
10
=0.05
| (X+Z) / Z |
8
6
0.1
4
0.2
2
0.5
=1.0
0
0
0.5
1.5
2.5
2.5
phase, degrees
=1.0
-45
-90
-135
=0.05
-180
0
0.5
1.5
frequency ratio, = / n
Figure 10. The transmissibility ratio |(X + Z)/Z| = Tr(, ), equation (104).
CC BY-NC-ND H.P. Gavin
20
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
Finally, lets consider the total motion of the mass x(t) + z(t).
X +Z
X
(1 2 ) + i(2) + 2
=
+1 =
Z
Z
(1 2 ) + i(2)
1 + i(2)
=
(1 2 ) + i(2)
and
X
+
Z
(103)
1 + (2)2
= Tr(, ).
(1 2 )2 + (2)2
(104)
ratio between the total response amplitude X + Z and the base motion Z.
See figure 10.
For systems that have a longer natural period (lower natural frequency)
than the period (frequency) of the support motion, (i.e., > 2), the transmissibility ratio is less than 1. In such systems the mass is isolated from
motion of the supports.
3.3 Eccentric-Mass Excitation
(106)
CC BY-NC-ND H.P. Gavin
21
x(t)
11
00
00
11
00
11
00
11
00
11
00
11
m
k, c
11111111111
00000000000
00000000000
11111111111
k
m
m
c
x(t)
when = 0 but increase with for > 2. This increasing effect is significant for > 0.2, as shown in Figure 12.
3.4
Finding the damping from the peak of the frequency response function
For lightly damped systems, the frequency ratio of the resonant peak,
the amplification of the resonant peak, and the width of the resonant peak
are functions to of the damping ratio only. Consider two frequency ratios
1 and 2 such that |H(1 , )|2 = |H(2 , )|2 = |H|2peak /2 where |H(, )|
is one of the frequency response functions described in earlier sections. The
frequency ratio corresponding to the peak of these functions peak , and the
value of the peak of these functions, |H|2peak are given in Table 1. Note that
the peak coordinate depends only upon the damping ratio, .
Since 22 21 = (2 1 )(2 + 1 ) and since 2 + 1 2,
2 1
2
(108)
CC BY-NC-ND H.P. Gavin
22
CEE 201L. Uncertainty, Design, and Optimization Duke University Spring 2012 H.P. Gavin
2
transmission: (Z to (X+Z))
2 | (X+Z) / Z |
10
=0.05
8
6
0.1
=1.0
0.2
0.5
2
0
0
0.5
1.5
2.5
2.5
phase, degrees
=1.0
-45
-90
-135
=0.05
-180
0
0.5
1.5
frequency ratio, = / n
which is called the half-power bandwidth formula for damping. For the
first, second, and fourth frequency response functions listed in Table 1 the
approximation is accurate to within 5% for < 0.20 and is accurate to within
1% for < 0.10.
23
i
(12 )+i(2)
4 2 (1 2 )
4 1 2
1
4 2
4 1 + 2
1
4 2 (1 2 )
4 1 2
18 2 (1 2 )
8 4
8 4 4 2 1+ 1+8 2
ouch.
1 2 2
1
22 21
2
(12 )+i(2)
1+i(2)
(12 )+i(2)
1
12 2
((1+8 2)1/21)
2
1/2