Generalized

SDOFs
Giacomo Boffi
Introductory
Remarks

Multi Degrees of Freedom Systems

MDOFs

The Homogeneous
Problem
Modal Analysis
Examples

Giacomo Boffi
Dipartimento di Ingegneria Civile e Ambientale, Politecnico di Milano

April 29, 2014

Introductory Remarks

Generalized
SDOFs

Consider an undamped system with two masses and two degrees of freedom.

k1

p1 (t)

p2 (t)

m1

m2

k2

x1

p1

k3

k2 (x2 x1 )

p2
m2 x2

k3 x2

m2 x2 k2 x1 + (k2 + k3 )x2 = p2 (t )

Generalized
SDOFs
Giacomo Boffi
Introductory
Remarks
The Homogeneous
Problem
Modal Analysis
Examples

Generalized
SDOFs
Giacomo Boffi
Introductory
Remarks

An Example

An Example

The Equation of Motion

The Equation of Motion

Matrices are Linear

Operators

Matrices are Linear

Operators

An example

x2

m1 x1 + (k1 + k2 )x1 k2 x2 = p1 (t )

The equation of motion of a 2DOF system

Giacomo Boffi

Properties of Structural
Matrices

k2 (x1 x2 )

m1 x1

Introductory Remarks
An Example
The Equation of Motion, a System of Linear
Differential Equations
Matrices are Linear Operators
Properties of Structural Matrices
An example
The Homogeneous Problem
The Homogeneous Equation of Motion
Eigenvalues and Eigenvectors
Eigenvectors are Orthogonal
Modal Analysis
Eigenvectors are a base
EoM in Modal Coordinates
Initial Conditions
Examples
2 DOF System

Introductory
Remarks

We can separate the two masses, single out the spring forces and, using the
DAlembert Principle, the inertial forces and, finally, write an equation of
dynamic equilibrium for each mass.

k1 x1

Outline

The Homogeneous
Problem
Modal Analysis
Examples

With some little rearrangement we have a system of two

linear differential equations in two variables, x1 (t) and x2 (t):
(
m1 x1 + (k1 + k2 )x1 k2 x2 = p1 (t),
m2 x2 k2 x1 + (k2 + k3 )x2 = p2 (t).

Properties of Structural
Matrices
An example

The Homogeneous
Problem
Modal Analysis
Examples

The equation of motion of a 2DOF system

Generalized
SDOFs

fS = K x

Generalized
SDOFs

Giacomo Boffi
Introductory
Remarks
An Example
The Equation of Motion

Introducing the loading vector p, the vector of inertial

forces f I and the vector of elastic forces f S ,


 


p1 (t)
fI ,1
fS,1
p=
, fI =
, fS =
p2 (t)
fI ,2
fS,2

Matrices are Linear

Operators
Properties of Structural
Matrices
An example

The Homogeneous
Problem
Modal Analysis
Examples

we can write a vectorial equation of equilibrium:

Giacomo Boffi

It is possible to write the linear relationship between f S and


T
the vector of displacements x = x1 x2
in terms of a
matrix product.
In our example it is


k1 + k2
k2
fS =
x = Kx
k2
k2 + k3

Introductory
Remarks
An Example
The Equation of Motion
Matrices are Linear
Operators
Properties of Structural
Matrices
An example

The Homogeneous
Problem
Modal Analysis
Examples

introducing the stiffness matrix K.

The stiffness matrix K has a number of rows equal to the
number of elastic forces, i.e., one force for each DOF and a
number of columns equal to the number of the DOF.
The stiffness matrix K is hence a square matrix K

fI + fS = p(t).

ndofndof

f I = M x

Generalized
SDOFs

Matrix Equation

Giacomo Boffi

Giacomo Boffi

Introductory
Remarks
An Example
The Equation of Motion

Analogously, introducing the mass matrix M that, for our

example, is


m1 0
M=
0 m2
we can write

Matrices are Linear

Operators
Properties of Structural
Matrices
An example

Also the mass matrix M is a square matrix, with number of

rows and columns equal to the number of DOFs.

Introductory
Remarks

Finally it is possible to write the equation of motion in

matrix format:
M
x + K x = p(t).

An Example
The Equation of Motion
Matrices are Linear
Operators
Properties of Structural
Matrices
An example

The Homogeneous
Problem

The Homogeneous
Problem

Modal Analysis

Modal Analysis

Examples

fI = M
x.

Generalized
SDOFs

Of course, we can consider the damping forces too, taking into

introducing a damping matrix C
account the velocity vector x,
and writing
M x + C x + K x = p(t),
however it is now more productive to keep our attention on
undamped systems.

Examples

Properties of K

Generalized
SDOFs

Properties of M

Generalized
SDOFs

Giacomo Boffi

Giacomo Boffi

Introductory
Remarks

Introductory
Remarks

An Example

K is symmetrical, because the elastic force that acts on

mass i due to an unit displacement of mass j, fS,i = kij
is equal to the force on mass j due to unit diplacement
of mass i , fS,j = kji in virtue of Bettis theorem.
The strain energy V for a discrete system can be
written
1
1
V = xT f S = xT K x,
2
2
because the strain energy is positive it follows that K is
a positive definite matrix.

Generalisation of previous results

The Equation of Motion

Matrices are Linear
Operators
Properties of Structural
Matrices
An example

The Homogeneous
Problem
Modal Analysis
Examples

Generalized
SDOFs

An Example

Restricting our discussion to systems whose degrees of

freedom are the displacements of a set of discrete masses,
we have that the mass matrix is a diagonal matrix, with all
its diagonal elements greater than zero. Such a matrix is
symmetrical and definite positive, as well as the stiffness
matrix is symmetrical and definite positive.
En passant, take note that the kinetic energy for a discrete
system is
1

T = x T M x.
2

The problem
Graphical statement of the problem

Matrices are Linear

Operators

An example

k1

The Homogeneous
Problem
Modal Analysis
Examples

An example

The Homogeneous
Problem
Modal Analysis
Examples

k1 = 2k,

Introductory
Remarks
An Example

p(t)

The Equation of Motion

The findings in the previous two slides can be generalised to

the structural matrices of generic structural systems, with
one exception.
For a general structural system, M could be semi-definite
positive, that is for some particular displacement vector the
kinetic energy could be zero.

Properties of Structural
Matrices

Giacomo Boffi

An Example

Properties of Structural
Matrices

Matrices are Linear

Operators

Generalized
SDOFs

Giacomo Boffi
Introductory
Remarks

The Equation of Motion

The Equation of Motion

m1

k2

m2

x1
x2
k2 = k;
m1 = 2m, m2 = m;
p(t) = p0 sin t.
The equations of motion

m1 x1 + k1 x1 + k2 (x1 x2 ) = p0 sin t,
m2 x2 + k2 (x2 x1 ) = 0.

... but we prefer the matrix notation ...

Matrices are Linear

Operators
Properties of Structural
Matrices
An example

The Homogeneous
Problem
Modal Analysis
Examples

because using the matrix notation we can follow the same steps
we used to find the steady-state response of a SDOF system.
First, the equation of motion


 


3 1
1
2 0
x + k
x = p0
sin t
m
1 1
0
0 1
substituting x(t) = sin t and simplifying sin t, dividing by k,
with 02 = k/m, 2 = 2 /02 and st = p0 /k the above equation
can be written




 


3 1
3 2 2
1
1
2 2 0

=
2 = st
1 1
0 1
1
1
0

Generalized
SDOFs

Generalized
SDOFs

Giacomo Boffi

Giacomo Boffi

Introductory
Remarks

Introductory
Remarks

steady-state response for a 2 dof system, harmonic load

An Example

An Example

The Equation of Motion

The Equation of Motion

1/st
2/st

Matrices are Linear

Operators
Properties of Structural
Matrices

Matrices are Linear

Operators
Properties of Structural
Matrices

An example

The Homogeneous
Problem
Modal Analysis
Examples

solving for /st gives


 


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

=
=
.
1
1
st
( 2 2 )( 2 2)
( 2 2 )( 2 2)

Homogeneous equation of motion

The solution, graphically

An example

Normalized displacement

The steady state solution

The Homogeneous
Problem
Modal Analysis

Examples

0.5

5
2=2/2
o

Generalized
SDOFs

Eigenvalues

Giacomo Boffi

To understand the behaviour of a MDOF system, we start

writing the homogeneous equation of motion,
M
x + K x = 0.
The solution, in analogy with the SDOF case, can be
written in terms of a harmonic function of unknown
frequency and, using the concept of separation of variables,
of a constant vector, the so called shape vector :
x(t) = (A sin t + B cos t).
Substituting in the equation of motion, we have


K 2 M (A sin t + B cos t) = 0

Introductory
Remarks
The Homogeneous
Problem
The Homogeneous
Equation of Motion
Eigenvalues and
Eigenvectors
Eigenvectors are
Orthogonal

Modal Analysis
Examples

Generalized
SDOFs
Giacomo Boffi

The previous equation must hold for every value of t, so it

can be simplified removing the time dependency:


K 2 M = 0.
This is a homogeneous linear equation, with unknowns i
and the coefficients that depends on the parameter 2 .

Speaking of homogeneous systems, we know that there is

always a trivial solution, = 0, and that different non-zero
solutions are available when the determinant of the matrix
of coefficients is equal to zero,


det K 2 M = 0

The eigenvalues of the MDOF system are the values of 2

for which the above equation (the equation of frequencies)
is verified.

Introductory
Remarks
The Homogeneous
Problem
The Homogeneous
Equation of Motion
Eigenvalues and
Eigenvectors
Eigenvectors are
Orthogonal

Modal Analysis
Examples

Eigenvectors

Generalized
SDOFs

Initial Conditions

Generalized
SDOFs

Giacomo Boffi

For a system with

N degrees of freedom the expansion of
2
det K M is an algebraic polynomial of degree N in
2 , whose roots, i2 , i = 1, . . . , N are all real and greater
than zero if both K and M are positive definite matrices,
condition that is always satisfied by stable structural
systems.
Substituting one of the N roots i2 in the characteristic
equation,


K i2 M i = 0

Giacomo Boffi

Introductory
Remarks

Introductory
Remarks

The Homogeneous
Problem

The Homogeneous
Problem

The Homogeneous
Equation of Motion
Eigenvalues and
Eigenvectors

The most general expression (the general integral) for the

displacement of a homogeneous system is

Eigenvectors are
Orthogonal

Modal Analysis
Examples

The Homogeneous
Equation of Motion
Eigenvalues and
Eigenvectors
Eigenvectors are
Orthogonal

x(t) =

N
X

Modal Analysis

i (Ai sin i t + Bi cos i t).

Examples

i=1

In the general integral there are 2N unknown constants of

integration, that must be determined in terms of the initial
conditions.

the resulting system of N 1 linearly independent equations

can be solved (except for a scale factor) for i , the
eigenvector corresponding to the eigenvalue i2 .
A common choice for the normalisation of the eigenvectors
is normalisation with respect to the mass matrix,
T
i M i = 1

Initial Conditions

Generalized
SDOFs

Usually the initial conditions are expressed in terms of initial

displacements and initial velocities x0 and x 0 , so we start deriving
the expression of displacement with respect to time to obtain

x(t)
=

N
X
i=1

i i (Ai cos i t Bi sin i t)

x(0) =

N
X

i Bi = x0 ,

x(0)
=

N
X

i i Ai = x 0 .

The above equations are vector equations, each one

corresponding to a system of N equations, so we can compute
the 2N constants of integration solving the 2N equations
N
X
i=1

ji Bi = x0,j ,

Introductory
Remarks
The Homogeneous
Problem
The Homogeneous
Equation of Motion
Eigenvalues and
Eigenvectors

N
X
i=1

ji i Ai = x 0,j ,

Modal Analysis

Giacomo Boffi

r2

s2 ,

Take into consideration two distinct eigenvalues,

and
and write the characteristic equation for each eigenvalue:
K r = r2 M r
K s = s2 M s

Examples

i=1

i=1

Generalized
SDOFs

Giacomo Boffi

Eigenvectors are
Orthogonal

and evaluating the displacement and velocity for t = 0 it is

Orthogonality - 1

j = 1, . . . , N.

Introductory
Remarks
The Homogeneous
Problem
The Homogeneous
Equation of Motion
Eigenvalues and
Eigenvectors
Eigenvectors are
Orthogonal

Modal Analysis
Examples

premultiply each equation member by the transpose of the

other eigenvector
2 T
T
s K r = r s M r
2 T
T
r K s = s r M s

Orthogonality - 2

Generalized
SDOFs

Orthogonality - 3

Generalized
SDOFs

Giacomo Boffi
Introductory
Remarks
The Homogeneous
Problem

The term T
s K r is a scalar, hence
T
T
s K r = s K r

T

Giacomo Boffi

Substituting our last identities in the previous equations, we

have
2 T
T
r K s = r r M s

The Homogeneous
Equation of Motion

T
= T
r K s

but K is symmetrical, KT = K and we have

Eigenvalues and
Eigenvectors

2 T
T
r K s = s r M s

Eigenvectors are
Orthogonal

Modal Analysis
Examples

subtracting member by member we find that

T
T
s K r = r K s .

T
r M s = 0,

Orthogonality - 4

Generalized
SDOFs
Giacomo Boffi
Introductory
Remarks

The eigenvectors are orthogonal also with respect to the

stiffness matrix:

and

2
T
i K i = i Mi .

Eigenvalues and
Eigenvectors
Eigenvectors are
Orthogonal

Modal Analysis
Examples

for r 6= s.

x=

Eigenvalues and
Eigenvectors

Modal Analysis

Generalized
SDOFs

The eigenvectors are linearly independent, so for every

vector x we can write

The Homogeneous
Equation of Motion

Eigenvectors are
Orthogonal

for r 6= s.

Eigenvectors are a base

The Homogeneous
Problem

Examples

Mi = T
i M i

The Homogeneous
Equation of Motion

We started with the hypothesis that r2 6= s2 , so for every

r 6= s we have that the corresponding eigenvectors are
orthogonal with respect to the mass matrix

T
T
s M r = r M s .

By definition

The Homogeneous
Problem

(r2 s2 ) T
r M s = 0

By a similar derivation

2 T
T
s K r = r s M r = 0,

Introductory
Remarks

N
X

j qj .

j=1

The coefficients are readily given by premultiplication of x

by T
i M, because
T
i Mx

N
X

T
T
i M j qj = i M i qi = Mi qi

j=1

in virtue of the ortogonality of the eigenvectors with respect

to the mass matrix, and the above relationship gives
qj =

T
j Mx
.
Mj

Giacomo Boffi
Introductory
Remarks
The Homogeneous
Problem
Modal Analysis
Eigenvectors are a base
EoM in Modal Coordinates
Initial Conditions

Examples

Eigenvectors are a base

Generalized
SDOFs
Giacomo Boffi

Generalising our results for the displacement vector to the

acceleration vector, we can write
x(t) =

N
X

x(t) =

j qj (t),

j=1

xi (t) =

N
X

N
X

ij qj (t),

xi (t) =

j=1

Substituting the last two equations in the equation of

motion,

The Homogeneous
Problem

j qj (t),

Eigenvectors are a base

EoM in Modal Coordinates
Initial Conditions

Examples

ij qj (t).

Introducing q(t), the vector of modal coordinates and ,

the eigenvector matrix, whose columns are the eigenvectors,

The Homogeneous
Problem

+ K q = p(t)
Mq

Modal Analysis
Eigenvectors are a base

premultiplying by

EoM in Modal Coordinates

Initial Conditions

+ T K q = T p(t)
T M q

Initial Conditions Revisited

Generalized
SDOFs

Giacomo Boffi

mij?
kij?

=
=

Introductory
Remarks
The Homogeneous
Problem
Modal Analysis

T
i M j
T
i K j

ij Mi ,

i2 ij Mi .

where ij is the Kroneker symbol,


1 i =j
ij =
0

Substituting in the equation of motion, with pi? = T

i p(t)
we have a set of uncoupled equations
Mi qi + i2 Mi qi = pi? (t),

Eigenvectors are a base

EoM in Modal Coordinates
Initial Conditions

Examples

Giacomo Boffi

The initial displacements can be written in modal

coordinates,

Introductory
Remarks

and premultiplying both members by T M we have the

following relationship:

Modal Analysis

x0 = q0

M x0 = M q0 = M q0 .

i = 1, . . . , N

q0 = (M? )1 T M x0
and, analogously,
q i0 =

i T M x 0
Mi

qi0 =

The Homogeneous
Problem
Eigenvectors are a base
EoM in Modal Coordinates
Initial Conditions

Examples

Premultiplying by the inverse of M? and taking into

account that M? is diagonal,

i 6= j

Examples

where p? (t) = T p(t), and the scalar equation are

X
X
pi? =
mij? qj +
kij? qj .
Generalized
SDOFs

We must examine the structure of the starred symbols.

The generic element, with indexes i and j, of the starred
matrices can be expressed in terms of single eigenvectors,

Introductory
Remarks

+ K? q = p? (t)
M? q

(t).
x(t) = q

... are N independent equations!

Giacomo Boffi

introducing the so called starred matrices we can finally

write

j=1

x(t) = q(t),

Generalized
SDOFs

Modal Analysis

j=1

N
X

Introductory
Remarks

EoM in Modal Coordinates...

T
i M x0
Mi

2 DOF System

Generalized
SDOFs
Giacomo Boffi
Introductory
Remarks

k1

m1

k2

m2

The Homogeneous
Problem

p(t)

Modal Analysis
Examples
2 DOF System

k1 = k,

Generalized
SDOFs
Giacomo Boffi

The equation of frequencies is

2


2k
= 0.
K 2 M = 3k 2 m
2k
2k 2 m

Introductory
Remarks
The Homogeneous
Problem
Modal Analysis

Developing the determinant

2

(2m ) (7mk) + (2k ) = 0,

x1
x2
k2 = 2k;
m1 = 2m, m2 = m;
p(t) = p0 sin t.

Examples
2 DOF System

(1)

with 02 = k/m and 2 = 02 ,

 
 
x1
0
, p(t) =
sin t,
x=
x2
p0




2 0
3 2
M=m
, K=k
.
0 1
2 2

Eigenvectors

Equation of frequencies

22 702 + 204 = 0.

(2)

Solving the algebraic equation in 2

k 7 33
k 7 + 33
2
2
1 =
2 =
m
4
m
4
k
k
12 = 0.31386
22 = 3.18614
m
m
Generalized
SDOFs

Normalization

Generalized
SDOFs

Giacomo Boffi

Substituting 12 for 2 in the first of the characteristic

equations gives the ratio between the components of the
first eigenvector,
k (3 2 0.31386)11 2k21 = 0
while substituting 22 gives
k (3 2 3.18614)12 2k22 = 0.
Solving with the arbitrary assignment 21 = 22 = 1 gives
the unnormalized eigenvectors,




+0.84307
0.59307
1 =
, 2 =
.
+1.00000
+1.00000

Introductory
Remarks
The Homogeneous
Problem
Modal Analysis
Examples
2 DOF System

Giacomo Boffi

We compute first M1 and M2 ,

M1 =

Introductory
Remarks

T
1 M 1




2m 0
0.84307
= 0.84307, 1
0
m
1



0.84307
= 1.68614m, m
= 2.42153m
1


M2 = 1.70346m

the adimensional normalisation factors are

2 = 1.70346.
1 = 2.42153,
Applying the normalisation factors to
eigenvectors and collecting them in a
normalized eigenvectors

+0.54177
=
+0.64262

the respective unnormalised

matrix, we have the matrix of
0.45440
+0.76618

The Homogeneous
Problem
Modal Analysis
Examples
2 DOF System

Modal Loadings

Generalized
SDOFs

Modal EoM

Giacomo Boffi
Introductory
Remarks
The Homogeneous
Problem

The modal loading is

Modal Analysis
Examples

p (t) = p(t)

 
+0.54177 +0.64262
0
sin t
= p0
0.45440 +0.76618
1


+0.64262
sin t
= p0
+0.76618

Particular Integral

2 DOF System

1 = C1 sin t,

and substitute in the first modal EoM:

C1

12

solving for C1

C1 =

p?
sin t = 1 sin t
m

1
p1?
m 12 2

with 12 = K1 /m m = K1 /12 :
C1 =

p1?
12
1
(1)
= st
2
K1 1 2
1 12

(1)

with st =

p1?
p0

= 2.047
and 1 =
K1
k
1

of course
1
(2)
C2 = st
1 22

(2)

with st

p?
p0

= 2 = 0.2404
and 2 =
K2
k
2

Substituting its modal expansion for x into the equation of

motion and premultiplying by T we have the uncoupled
modal equation of motion
(
m
q1 + 0.31386k q1 = +0.64262 p0 sin t

Introductory
Remarks
The Homogeneous
Problem
Modal Analysis
Examples
2 DOF System

Note that all the terms are dimensionally correct. Dividing

by m both equations, we have

q1 + 12 q1 = +0.64262 0 sin t
m
p

q2 + 22 q2 = +0.76618 0 sin t
m

Giacomo Boffi

= 2 C1 sin t

Giacomo Boffi

m
q2 + 3.18614k q2 = +0.76618 p0 sin t

Generalized
SDOFs

We set

Generalized
SDOFs

Introductory
Remarks
The Homogeneous
Problem
Modal Analysis
Examples
2 DOF System

Integrals

Generalized
SDOFs

The integrals, for our loading, are thus

(1) sin t

q1 (t) = A1 sin 1 t + B1 cos 1 t + st 1 2

1

sin
t
(2)

q2 (t) = A2 sin 2 t + B2 cos 2 t + st

1 22
for a system initially at rest

(1)

q1 (t) = st

Giacomo Boffi
Introductory
Remarks
The Homogeneous
Problem
Modal Analysis
Examples
2 DOF System

1
(sin t 1 sin 1 t)
1 12
1

q2 (t) = (2)
(sin t 2 sin 2 t)
st
1 22

we are interested in structural degrees of freedom, too... disregarding

transient
!



(1)
(2)
st
st
1.10926
0.109271

x1 (t) = 11
+ 12
sin t =

1 12
1 22
1 12
1 22
!


(1)
(2)

1.31575
0.184245
x (t) = st + st

sin t =
+
21
22
2
1 12
1 22
1 12
1 22

p0
sin t
k
p0
sin t
k

The response in modal coordinates

Generalized
SDOFs

2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2

Introductory
Remarks
The Homogeneous
Problem
Modal Analysis

Giacomo Boffi

q2()/st

10

15

20

2 DOF System

25

= o t

In the graph above, the responses are plotted against an

adimensional time coordinate with = 0 t, while the
ordinates are adimensionalised with respect to st = pk0

30

Introductory
Remarks

Using the same normalisation factors, here are the response

functions in terms of x1 = 11 q1 + 12 q2 and
x2 = 21 q1 + 22 q2 :

Examples

q1()/st

Generalized
SDOFs

Giacomo Boffi

xi/st

qi/st

To have a feeling of the response in modal coordinates, lets say

that the frequency of the load is = 20 .
2.0
This implies that 1 = 1 = 0.31386
= 6.37226 and

2.0
2 = 2 = 3.18614 = 0.62771.

The response in structural coordinates

2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2

The Homogeneous
Problem
Modal Analysis
Examples

x1()/st

10

2 DOF System

x2()/st

15
= o t

20

25

30