Raid Karoumi, KTH

Structural Dynamics
Two and multiple degrees-of-freedom systems
Modal analysis

Raid Karoumi
Structural Engineering & Bridges, KTH

Aim of lecture

After this lecture you should:

 Understand resonance & dynamic amplification

 Calculate natural frequencies and mode shapes

for multi-DOF system

 Understand the concept of mode superposition

 Calculate the response using mode

superposition

AF2011 Structural Dynamics 1

Raid Karoumi, KTH

Definitions

 Natural frequencies = eigenfrequecy=

eigenvalues (also fundamental frequency=f1)

 Mode shapes = eigenmode= eigenvectors

 Dynamic amplification factor = dyn. factor =

dyn. magnification factor = dyn. response factor

 Modal analysis = mode superposition

Dynamic response factor

p0 sin t   
u (t )  uc (t )  u p (t )  u0 (t ) sin  D t    
k  2 
2
  
2

1  2   4  
 n   n 

P(t)

After some while, the structure vibrates with the same frequency as the applied force.
It is the steady state response up(t).
The amplitude of these vibrations are now studied.
static deformation due to a static load po
po k
u p (t )  sin(t   )
1  (  )  n
2 2
 2 ( n ) 2

 The amplitude of the vibration is equal to the product of the static

deformation times a dimensionless dynamic factor Rd p. 76–82

AF2011 Structural Dynamics 2

Raid Karoumi, KTH

Dynamic response factor

Rd can be plotted as function of the ratio  /n
udyn 1 for different values of the damping coefficient 
Rd  
ustat  2 
2
  
2

1  2   4   5

 n   n  Rd
4.5

4  = 0.1
3.5
If the frequency of excitation is less than 3
~ 0.25 of the structures lowest eigen- 2.5
 = 0.2
frequency, inertia effects can be neglected 2
resulting in a quasi-static problem, 1.5  = 0.4
that is: 1

0.5
 / n  0.25  Rd  1 0
0 0.5 1 1.5 2 2.5 3

(  n )
m u  c u  k u  p0 sin(t   )
  n the amplitudes of vibrations
 u  p0 / k sin(t   ) become large : Resonance

Resonance

Resonance is reached for   r  n 1  2 2

1
For this value of  , the dynamic factor is Rd max 
2 1   2

1
if   0.1 then Rd max  and
2 Example
n  k m   D  n 1   2
 r  n 1  2 2   2%  Rd  25

The deformations at resonance are 25

times the static ones.

Stockholm 2010-03-23 6

AF2011 Structural Dynamics 3

Raid Karoumi, KTH

Dynamic response factors

p0
Displacement: u(t )  Rd  sint   
k

p0 1 1 k 1
Velocity: u (t )  Rv  cos t    k
 n  
k m

km
km

p0 1 1 k k 1
Acceleration: u(t )   Ra  sin t     n    
km 2 m
m km km m

2
  
Ra  Rv    Rd
n  n 

p. 80–82

Dynamic response factors

Chopra figure 3.2.7

p. 80–82

AF2011 Structural Dynamics 4

Raid Karoumi, KTH

System with 2 degrees of

freedom (DOF)

Two-degrees-of-freedom system

a) Reference axis 1 Reference axis 2

F1 p (t) F2 p (t)
1 1 2 2

kk31 c1 k2k 2 c2 kk33 c3

M11
m M22
m
D 3 D2 D3

1 2
b) k 2 ( 2 - 1)
k1 1 k3 2
M1 1 M2 2
Fp1 1 Fp22
m1u1 m2u2
cD1 1 1 cD3 3 2

Mass no. 1: cD2 2 ( 2 - 1)

 m1u1  c1u1  k1u1  k2 (u1  u2 )  c2 (u1  u2 )  0

Free vibration
Mass no. 2:
 m2u2  c3u2  k3u2  k2 (u2  u1 )  c2 (u2  u1 )  0
p. 345–349

10

AF2011 Structural Dynamics 5

Raid Karoumi, KTH

Two-degrees-of-freedom system

 m1u1  c1u1  k1u1  k2 (u1  u2 )  c2 (u1  u2 )  0

 m2u2  c3u2  k3u2  k2 (u2  u1 )  c2 (u2  u1 )  0

m1 0 u1  c1  c2  c2 u1  k1  k2  k2 u1   0 

 0 m u     c   
 2  2   2 c2  c3 u2    k2 k2  k3 u2   0 

M u
  C  u  K  u  0 The equation of motion in matrix form

11

Two-degrees-of-freedom system

M u
  C  u  K  u  0

 u1  u01 sin(n t   )
Assume (here): Cu  0 & Harmonic approach: 

u  u sin( t   )
 2 02 n

m1 0   n2u1  k1  k2  k2 u1   0 

 0 m  2     k  
 2  n u2   2 k2  k3 u2   0 

(K  n2M)u  0 Eigenvalue problem

p. 401–406
12

AF2011 Structural Dynamics 6

Raid Karoumi, KTH

Two-degrees-of-freedom system

For a simple system with two d.o.f. we can solve it by hand!

 k1mk2  n2  mk21  u1   0 

 1 k2     
  m2
k3  k 2
m2    u2   0  is simplified to
2
n

A 2  B  u1   0  of which the determinant is

     
 C D   2  u2   0 

 A D  A D 
2
12      AD  BC
 2  2 
A   D     BC  0
2
n
2
n


 2
 2 A D  A D 
Two d.o.f. – two frequencies!! 2  2  
 2 
  AD  BC

13

System with 3 DOF

Frame systems

1 M1 1 M1 1 1 F1
F1
k1 ( 1 - 2)
k'1
2 M2 2
M2 2 2 F2 A shear building
F2
k 2 ( 2 - 3)
k'2
3 M3 3
M3 3 3
F3 F3
k3' k3 3

As an example of a more complex system, we study a 3 story building

where the floor structure is assumed to be very stiff. Thus, the system
flexibility comes mainly from the columns.

We are interested in determining the Eigenfrequencies and modes

shapes for the structure.
p. (321–327)
345–357
(357–367)
14

AF2011 Structural Dynamics 7

Raid Karoumi, KTH

System with 3 DOF

Frame systems
F3 F2 F1
3 2 1

k3 k2 k1
m
M33 m22
M m
M11
c33
D c22
D c11
D

3 2 1
d) k2 ( 2- 3) k1 ( 1 - 2)
k3 3
m
M33 3 m2
M 2 m11
M 1
F3 F2 F1
Dc33 3

c22 ( 2 -
D 3) c11 (
D 1 - 2)

m1u1  k1 (u1  u2 )  F1 (t )  0
m2u2  k1 (u1  u2 )  k 2 (u2  u3 )  F2 (t )  0
m3u3  k 2 (u2  u3 )  k3u3  F3 (t )  0

15

System with 3 DOF

Frame systems
Rigid slabs and elastic columns

a) b) c) d) e)
M1 1 u1=1 1 k 12 = k 1 1
1 1 1
k11=k1
k'1 k'1 k1=2 k'1
M2 u2=1 k 23=k 2
2 2 2
2 2 2
k 22=k1+k2
k'2 k'2 k 2=2 k'2
M3 3 u3=1 3
k2
3 3 3 3
k 32= k 2 3
k3
k'3 k'3 k 3=2 k'3 k 33= k 2+k3

4 4 4 4 k 4 3 =k 3
4

12 EI 3EI
Fixed-fixed: k Fixed-pinned: k 
L3 L3

16

AF2011 Structural Dynamics 8

Raid Karoumi, KTH

System with 3 DOF

or in matrix form

m1 0 0   u1   k1  k1 0  u1   F1 (t ) 

0      
 m2 0  u2    k1 k1  k 2  k 2  u2    F2 (t )
 0 0 m3  u3   0  k2 k 2  k3  u3   F3 (t ) 

or with in a shorter style of writing

This is what you will develop
in A2 but for 4-story building.
M u
  K  u  f (t )

If we consider the damping, the equation will become

M u
  C  u  K  u  f (t )

17

System with 3 DOF

We look at the un-damped free vibration of the system to determine the Eigenfrequencies
and Eigenmodes of vibration

M u
  K  u  0

For the free vibration, we assume as always that the motion is harmonic, that is

u1  u01 sin(n t   ) u 2  u02 sin(n t   ) u3  u03 sin(n t   )

Insertion in the equation of motion will give us

   n u1   k1
2
m1 0 0  k1 0   u1  0
0 m2 0    2u    k

k1  k 2
   
 k 2  u2   0
  n 2 2   1
 0 0 m3   n u3   0  k2 k 2  k3  u3  0

18

AF2011 Structural Dynamics 9

Raid Karoumi, KTH

System with 3 DOF

or in matrix form

K   M u  0
2
n

This equation (which is even called the system characteristic equation) is a conventional
Eigenvalue problem and our job is to solve it in order to determine the Eigenvalues and
Eigenvectors.

The same equation is valid also for systems with more DOF than 3.

For large structural dynamic problems with many DOF, normally it is enough to determine
the lowest Eigenvalues.

19

System with 3 DOF

Mathematic software such as Matlab or Mathcad can be used to solve the Eigenvalue
problem.

Mathcad → A := genvecs(K,M) och B := genvals (K,M)

Matlab → [A,B] = eig (K,M)

 11  12  13          

           
The matrix A contain the Eigenvectors → 21  22  23   1   2   3 
   
 31  32  33          

ω12 0 0
 2 
The matrix B contain → 0 ω 2 0
 ω32 
0 0

20

AF2011 Structural Dynamics 10

Raid Karoumi, KTH

System with 3 DOF

Free vibration modes
Three d.o.f. = three natural frequencies and modes

U1=1 M 1=M U1=1 U1=1

1 1 1
k'1
3 M 2=2 M 3
U2= U2=
2 2 U2= 0 2 2
2
k'2
1 M 3=2 M 1
U3= U1=1 U3=
2 3 2 3
3
k'3
4 4 4

1 (Mod 1) 2 (Mod 2) 3 (Mod 3)

Se examples! 9.4–7 and 10.4–5

21

Multiple degrees-of-freedom
Free vibration modes

K   M u  02
Undamped free vibration!

• Solve the eigenvalue problem to determine modes & frequencies

• One unique mode for each natural frequency

• For n d.o.f. – n natural frequencies and modes

• Modes given in vector form n where n = 1, 2, 3, …..

• Scaling of modes = normalization

(mode shapes in previous slide are normalized for umax=1)

p. 401–418

22

AF2011 Structural Dynamics 11

Raid Karoumi, KTH

Example
A concrete foundation 4×1×2 m3 rests on concrete piles
with a cross-section of 0.3×0.3 m2.

Some piles are angled

(for simplicity in analysis these can
be substituted with horizontal and
vertical piles).

The modulus of elasticity of the

Concrete is 30 GPa. The mass moment of
inertia for this system is 31472 kgm2.

On the foundation is a machine placed.

Its mass is 1000 kg.

The centre of gravity for the foundation-machine system is

given in the figure on the next slide.

23

Example
a) Define coordinates that fully describes the plane movement of
the foundation.
b) Show the stiffness matrix.
c) Show the mass matrix.
d) Determine the natural frequencies.
e) Determine the free vibration modes.
5m 4m 5m
Study two cases:
0.5 m
x = 0 and x = 0.3 m!
PISTON

1 pile × 1 pile

2m 24

AF2011 Structural Dynamics 12

Raid Karoumi, KTH

Example solution

Suitable coordinates: ux = horizontal displacement

uv = vertical displacement
 = rotation (clockwise)

Mass matrix:

Mass moment of inertia: J = 31472 kgm2 (given)

Mass: m = 1000 + 2400(4·1·2) = 20200 kg

20200 0 0 
 
 
M  0 20200 0 
 
 
 0 0 31472

25

Example solution

Stiffness matrix:  k11 k12 k13  u x   p x 

    
    
k 21 k 22 k 23  u y    p y 
    
    
k 31 k 32 k 33     M 

Degree of freedom 1 uy = 0,  = 0
EA 2EA
→: px  2 ux  0 px  ux
L L

↑: py  0

EA 2EA
: M 2 ux  x  0 M x  ux
L L
26

AF2011 Structural Dynamics 13

Raid Karoumi, KTH

Example solution

Degree of freedom 2 ux = 0,  = 0

→: px  0

EA 4EA
py  4 uy  0 py  uy
↑: L L

: M 0

27

Example solution

Degree of freedom 3 ux = 0, uy = 0
EA 2 EA
px  2   x  0 px  x
→: L
L

EA EA py  0
↑: py  2   1  2   1  0
L L

:
EA EA 2 EA 2 4 EA
M 2   x  x  4   1 1  0 M x 
L L L L

28

AF2011 Structural Dynamics 14

Raid Karoumi, KTH

Example solution

In matrix form:
 2 EA 2 EA 
k13   L 0 x 
 k11 k12 L
   
   4 EA

k 21 k 22 k 23    0 0 
   L 
   
k 31 k 32 k 33   2 EA 2 EA 2 4 EA 
 x 0 x  
 L L L 

With EA / L = 30·109·0.09 / 5 = 5.4·108 is this stiffness matrix:

2 0 2x   10.8·108 0 10.8·108 x 
   
EA    
K 0 4 0  0 21.6·108 0 
L    
   
2x 0 2x 2  4 10.8·108 x 0 10.8·108 x 2  21.6·108 
29

Example solution
Natural frequencies:
 10.8·108 0 10.8·108 x 
 
  (K   2 M )u  0
K  0 21.6·108 0 
 
 
10.8·108 x 0 8
10.8·10 x  21.6·10 
8 2 to be solved with e.g the eig-command of

Matlab or by solving the system of three
equations.

20200 0 0  The commands for Matlab are:

 
  [fi,w2] = eig(K,M)
M  0
 20200 0  w = sqrt(w2)
 
  % the frequencies are
 0 0 31472 f = w/2/pi

% the mode shapes are

fi=fi./[max(abs(fi)); max(abs(fi); max(abs(fi))]

30

AF2011 Structural Dynamics 15

Raid Karoumi, KTH

Example solution
Frequencies
Alternative 1; x = 0 gives:

10.8·108 20200 2 0 0   0
   
   
 0 21.6·108 20200 2 0    0
   
   
  
 0 0 21.6·108 31472 2   0 

 1   231.2   f1   36.8 
       
       
with the solution      327.0  rad/s  f    52.0  Hz
equal to
  
2
   
2

       
    f   
 3   262.0   3   41.7 

31

Example solution
Frequencies
Alternative 2; x = 0.3 gives:

10.8·108  20200 2 0 3.24·108   0

   
   
 0 21.6·108  20200 2 0    0
   
   
 2  
 3.24·108 0 22.572·10  31472   0 
8

 1   216.40   f1   34.44 
       
       
with the solution      279.92  rad/s equal to  f 2    44.55  Hz
  
2
    
       
    f   
 3  327.00   3  52.04 

32

AF2011 Structural Dynamics 16

Raid Karoumi, KTH

Example solution
Mode shapes
For x = 0.3 m this will give:

 1   0.64  0
     
     
1   0  f1  34.44 2   0  f 2  44.55 3   1  f 3  52.04
     
  Hz    
Hz 0 Hz
  0.41  1   
   

Compare with the results obtained with x = 0:

1  0 0
     
     
   3   0  f 3  41.7
1  0
 
f1 36.8  2   1  f 2  52.0  
 
  Hz   Hz   Hz
 0  0 1
     

33

System with 3 DOF

Dynamic response
The general solution of the equation of motion for a 3 DOF system to a dynamic load
can be written as

u1 ( t )  11  12  13 

       
u2 ( t )  21  q1 (t)  22  q2 (t)  23  q3 (t)
u ( t )       
 3   31   32   33 

eigenmode
Eigenmode 11 eigenmode2 2
Eigenmode eigenmode3 3
Eigenmode

= q1(t ) + q2 (t ) + q3(t )

34

AF2011 Structural Dynamics 17

Raid Karoumi, KTH

Modal analysis

Dynamic response of multi-degree-of-freedom-systems

to external load

Displacement – the sum of modal contributions

Mode
N
u(t )  1q1  2 q2    N qN  n qn  Φq(t ) superposition
n 1 analysis

u = geometric coordinate vector (displacement)

q = modal (generalized) coordinate vector

 = modal matrix (mode vectors in columns)

p. 472–482

35

Modal analysis

Un-coupled equations of vibration

Equation of vibration for a m.d.o.f. system: (t) K u(t)  p(t)

Mu

With generalized mass, stiffness and load:  M q(t)  K q(t)  p(t)
T
n n
T
n n
T
n

One-degree-of-freedom equation (in generalized coordinates)

for mode n:

M n q(t )  K n q (t )  Pn (t )

36

AF2011 Structural Dynamics 18

Raid Karoumi, KTH

Modal analysis

Analysis using the dynamic response factor

Here for un-damped systems ( = 0)

Remember that
p0
1 u(t )  Rd  sint    (SDOF)
The amplification factor: Rd,n  k
2
1 2 u(t )  1q1  2 q2   (MDOF)
n

Pn
The modal amplitudes: qn  ustat  Rd ,n   Rd ,n
Kn

37

Example

On the foundation is a 1000 kg

machine placed.

The centre of gravity for the

foundation-machine system
is given in the figure.

The mass moment of inertia for

this system is 31472 kgm2.

The machine contains a rotating piston which gives a dynamic

force of p0=328989 N at a frequency of =104.7 rad/s. The
piston is fastened 0.5 m from the centre of gravity of the
foundation.

38

AF2011 Structural Dynamics 19

Raid Karoumi, KTH

Example

Determine the maximum

displacement of the foundation
when the machine operates.

Use the method with dynamic response

factor. Transform into generalized coordinates,
solve through mode superposition.

Study two cases: x = 0 and x = 0.3 m!

39

Example
Dynamic loading

Start with p  (K   M)u

2

 0   0
   
   
The load vector is: p   p0   328989  1  (the moment has a 0.5 m lever arm)
   
   
 0.5 p   0.5
 0  

1 1
The response factor is Rd   since  = 0
 2 
2
 
2 2
1  2   4  1
   n2
 n   n 

40

AF2011 Structural Dynamics 20

Raid Karoumi, KTH

Example

Generalized properties

The generalized expressions for stiffness and load are: K n  n K n

T

Pn  nT p

The eigenvectors are collected in a matrix: Φ  1 2 3 

The generalized expressions in vector form is (for e.g.. MATLAB):


k n  diag ΦT K Φ p n  ΦT p (size 3×1)

k n  ΦT KΦ I with I  1 1
T
or as 1

41

Example

Modal amplitudes and mode superposition

Pn
The modal amplitudes are given by: qn  ustat  Rd   Rd
Kn

which is the corresponding static displacement multiplied by the response factor.

In matrix form (MATLAB) is this written as: q n  pn . / k n . * R d (size 3×1)

Note: element-by-element operations

The maximum displacement is obtained through mode superposition:

3
umax  n qn  Φ qn
n 1

42

AF2011 Structural Dynamics 21

Raid Karoumi, KTH

Example

Alternative 1; with x = 0

Insertion of the results from Example 7 gives:

R d  1.258 1.114 q n  106 0 169.7

T T
1.190 and 90.6

 0 
 
 
and finally: umax   0.170 103 m
 
 
 0.091
 

43

Example

Alternative 2; with x = 0.3

Insertion of the results from Example 7 gives:

R d  1.306 1.114 q n  106  74.1 169.7

T T
1.163 and 61.2

  0.035
 
 
and finally: umax   0.170  10 m
3
 
 
 0.091 
 

44

AF2011 Structural Dynamics 22