Multiple-Degree-of-

Freedom Systems
 Introduction

• So far we’ve demonstrated how a structure can be

represented as a SDOF system the dynamic response of
which can be evaluated by the solution of a single
differential equation of motion

• If the physical properties of the system are such that its

motion can be described by a single coordinate and no other
motion is possible, then it actually is a SDOF system and the
solution of the equation provides the exact dynamic response
 Introduction

• In general, the dynamic response of a structure

cannot be described adequately by a SDOF model;
usually the response includes time variations of the
displacement shape as well as its amplitude.

• Such behaviour can be described only in terms of

more than one displacement coordinate

• We’ll first formulate the equations of motion for the

simplest possible MDOF system – a 2-DOF system
 TWO-DEGREE-OF-FREEDOM MODEL (UNDAMPED)

• Computer programs are ideally placed to solve for responses of

complex systems

• Two distinguishing phenomena between SDOF and MDOF

need to be stated:
• A two DOF system will have Two natural frequencies
• MDOF systems have a mode shape, NOT present in a
SDOF system

• A mode shape is a vector that describes the relative motion

between the two masses or between two degrees of freedom

• Mode shapes are intimately tied to the mathematical concepts

of eigenvalues and eigenvectors of computational matrix theory
(a) Simple 2DOF model consisting of two masses connected in series
by two springs.

(b) Single mass with 2 DOF (mass moves along both the x1 & x2

(c) Single mass with 1 translational DOF and 1 rotational DOF

 Equations of motion

Fig. 2: FBD of the two masses

Summing forces on each mass in the horizontal direction yields:

𝑚1 𝑥1ሷ = −𝑘1 𝑥1 + 𝑘2 𝑥2 − 𝑥1
Eqn. 1
𝑚2 𝑥2ሷ = −𝑘2 𝑥2 − 𝑥1

Re-arranging:

Eqn. 2
• Eqns 2 consist of two coupled 2nd Order D.E with constant
coefficients, each of which requires two initial conditions to solve.
• Hence these two coupled equations are subject to the four initial
conditions:

Eqn. 3
Where:

𝑥10
ሶ , 𝑥20
ሶ 𝑎𝑛𝑑 𝑥10 , 𝑥20

• represent the initial velocities and displacements of each of the

two masses
• These are assumed to be known or given and provide the 4 constants
of integration needed to solve the two 2nd D.E for the free response
of each mass.
 • To find the responses 𝑥1(𝑡) and 𝑥2(𝑡), neither equation can
be solved by itself because each equation contains both x1
and x2 (coupled eqn)
• This state implies the motion of x1 affects the motion of x2,
and vice versa
• A convenient method of solving this system is to use
vectors and matrices

• We define the vector 𝒙(𝑡) to be the column vector consisting

of the two responses of interest

Eqn. 4

This is called a displacement or response vector and is a 2 × 1

array of functions
Differentiation of a vector is defined here by differentiating each
element so that

Eqn. 4

These are; the velocity & the acceleration vectors, respectively

The mass of the system can also be represented in matrix form:

Eqn. 5

Here, the italic capital letters are used to denote matrices and bold
lowercase letters are used to denote vectors
A product of a mass matrix M and the acceleration vector 𝒙ሷ is
expressed as:

If you have a stiffness matrix K:

Eqn. 6

The product Kx (where x given in eqn 4) is given as:

Consider the vector equation for an undamped-free vibration:
Eqn. 7

Substitution of the value for M (from Eqn 5) and the value for K (from
Eqn. 6) into Eqn. 7 yields;
Adding the two vectors on the LHS of the equation yields:

Eqn. 8
• The mass and stiffness matrices, M and K, described previously have
the special property of being symmetric.

• A symmetric matrix is a matrix that is equal to its transpose.

• The mass matrix M is also called the inertia matrix

• The force vector 𝑀𝒙ሷ corresponds to the inertial forces in the

system of Fig. 1 (a)

• The force Kx represents the elastic restoring forces of the system

By making use of the Transpose Matrix, the initial conditions
expressed in eqn. 3 can be written in terms of vector as:

Where x0 and 𝐱ሶ 0 are the initial displacement vector and initial

velocity vector, respectively

Recall that the single-degree-of-freedom version of eqn. 7 was

solved by assuming a harmonic solution

In this case, a solution is assumed of the form

 Where
• u is a nonzero vector of constants to be determined
• ω is a constant to be determine
• d

Substitution of this assumed form of the solution into the vector

equation of motion (Eqn.7) yields;

Eqn. 8

Where
 Equation (8) yields the fact that ω and u must satisfy the vector
equation
Eqn. 9

This represents two algebraic equations in the three unknown scalars;

• For this homogeneous set of algebraic equations to have a nonzero

solution for the vector u, the inverse of the coefficient matrix (–Mω2+
K) must not exist.

• To see that this is the case, suppose that the inverse (–Mω2+ K) does
exist.

• Then multiplying both sides of eqn. 9 by (–Mω2+ K)-1 yields u = 0, a

trivial solution, as it implies no motion.
• Thus, the solution of eqn 6 depends in some way on the matrix
inverse

• Applying the condition of singularity to the coefficient matrix of

eqn. 9 yields the result that for a nonzero solution u to exist

• This yields one algebraic equation in one unknown (ω2)

• Substituting the values of the matrices M and K into this

expression yields
 Using the definition of the determinant yields that the unknown
quantity ω2 must satisfy

𝒎𝟏 𝒎𝟐 𝝎𝟒 − 𝒎𝟏 𝒌𝟐 + 𝒎𝟐 𝒌𝟏 + 𝒎𝟐 𝒌𝟐 𝝎𝟐 + 𝒌𝟏 𝒌𝟐 = 𝟎 Eqn. 10

• This expression is called the characteristic equation for

the system

• It’s used to determine the constants 𝜔 in the assumed form

of the solution given by equation 𝐱 𝑡 = 𝐮𝑒 𝑗𝜔𝑡 once the
values of the physical parameters 𝑚1, 𝑚2, 𝑘1, and 𝑘2 are
known
 EXAMPLE 1

Calculate the solutions for ω of the characteristic equation given by

EQN . 10 for the case that the physical parameters have the values m1
= 9 kg, m2= 1 kg, k1 = 24 N/m, and k2 = 3 N/m.

SOLUTION

Using the given value, the characteristic equation 10 becomes;

• Note that in the statement of eqn 9; 𝜔2 appears, not 𝜔.

• However, in proceeding to the solution in time, the frequency of

oscillation will become 𝜔 and the plus and minus signs on 𝜔 are
absorbed in changing the exponential into a trigonometric
function

• Once the value of 𝜔 in solution equation is established, the

value of the constant vector u can be found by solving eqn (9)
for u given each value of 𝜔2

• i.e , for each value of 𝜔2 (i.e., 𝜔12 and 𝜔22 ) there is a vector u
satisfying eqn 9
 Eqn. 11

Eqn. 12

These two expressions can be solved for the direction of

the vectors u1 and u2 but not for the magnitude
 Frequencies
• From the example, the two natural frequencies 𝜔1 and 𝜔2
of the two-degree-of-freedom system are not equal to
either of the natural frequencies of the two single-degree-
of-freedom systems constructed from the same
components. To see this,

𝑘1
• E.g = 1.63,
𝑚1

• 1.63 ≠ 𝜔1 (= 2) or 𝜔2 (=2)

𝑘2
• Similarly, = 1.732 ≠ ω1 and ω2
𝑚2
Multiple-Degree-of-
Freedom Systems

MODE SHAPES
 How to compute ‘u’
Recall that eqns 11 and 12 are:

−𝜔12 𝑀 + 𝐾 𝐮1 = 0 and −𝜔22 𝑀 + 𝐾 𝐮2 = 0

• These two expressions can be solved for the direction of the vectors
u1 and u2, but not for the magnitude

• To see that this is true, note that if u1 satisfies the characteristic eqn;
𝑚1 𝑚2 𝜔4 − 𝑚1 𝑘2 + 𝑚2 𝑘1 + 𝑚2 𝑘2 𝜔2 + 𝑘1 𝑘2 = 0
• so does the vector au1, where a is any nonzero number. Hence the
vectors satisfying the above equations are of arbitrary magnitude.

• The following example illustrates one way to compute u1 and u2 for

the values of
 EXAMPLE 2

Calculate the vectors u1 and u2 of eqns 10 and 11 for the values of ω,

K, and M from previous example where m1 = 9 kg, m2= 1 kg, k1 = 24
N/m, and k2 = 3 N/m.
Solution

and

From Example 1, we found:

From the vector equation 10;

The element for 𝐮𝟏 are: u11 and u21

Using 𝜔2 = 𝜔12 = 2; we the equation becomes:

 − 2 9 + 24 + 3 −3 𝑢11 0
𝑢21 =
−3 − 2 1 +3 0

9 −3 𝑢11 0
𝑢21 =
−3 1 0

These two equations are dependent and yield the same solution:
• Only the ratio of the elements is determined here

• Only the direction of the vector is determined by eqn.10 not its

magnitude
• As mentioned previously, this happens because if u satisfies eqn.10,
then so does au, where a is any nonzero number
• A numerical value for each element of the vector u may be obtained
by arbitrarily assigning one of the elements.

• For example, let u21 = 1; then the value of u1 is

This procedure is repeated using 𝜔22 = 4 to yield that the elements of
u2 must satisfy:

𝑢12 0
𝑢22 =
0

− 4 9 + 24 + 3 −3 𝑢12 0
𝑢22 =
−3 − 4 1 +3 0

−9 −3 𝑢12 0
𝑢22 =
−3 −1 0
The elements of u2 must satisfy:

• There are several other ways of fixing the magnitude of a vector

besides the one illustrated here.
• The solution of the vector equation subject to initial conditions 𝐱𝟎
and 𝐱ሶ 0 can be constructed in terms of the numbers ±ω1, ±ω2 and
the vectors u1 and u2 .

• This is similar to the construction of the solution of the single-

degree-of-freedom case.

• Since the equations to be solved are linear, the sum of any two
solutions is also a solution.

• From the preceding calculation, there are four solutions in the

form of equation; 𝐱 𝑡 = 𝐮𝑒 𝑗𝜔𝑡 made up of the four values of ω
and the two vectors
A general solution is the linear combination of these four:-

Where:
a, b, c, and d are arbitrary constants of integration to be determined
by the initial conditions

An alternative form of the solution is formulated using Euler

formulas (provided neither ωi is zero)

Eqn. 12
where
• the constants of integration are now in the form of two
amplitudes, A1 and A2, and two phase shifts, ϕ1 and ϕ2.
• These constants can be calculated from the initial conditions 𝐱𝟎
and 𝐱ሶ 0

• The equation,

is the two-degree-of-freedom analog of a single-degree-of-

freedom equation:
• The form of eqn. 12 gives physical meaning to the solution.
• It states that each mass in general oscillates at two frequencies: 𝜔1
and 𝜔2.
• These are called the natural frequencies of the system.
• Furthermore, suppose that the initial conditions are chosen such
that A2 = 0
• With such initial conditions, each mass oscillates at only one
frequency, 𝜔1, and the relative positions of the masses at any given
instant of time are determined by the elements of the vector u1

• Hence u1 is called the first mode shape of the system.

• Similarly, if the initial conditions are chosen such that A1 is zero,

both coordinates oscillate at frequency 𝜔2 with relative positions
given by the vector u2, called the second mode shape
• Mode shapes have become a standard in vibration
engineering and are used extensively in structural analysis

• The concepts of natural frequencies and mode shapes are

extremely important and form one of the major ideas used in
vibration studies
• It is assumed in the derivation that neither of the values of 𝜔
is zero.

• One or the other may have the value zero in some

applications, but then the solution takes on another form

• The value of u, however, cannot be zero

• A frequency can be zero, but a mode shape cannot be zero

• The zero frequency case corresponds to rigid body motion

 EXAMPLE 3

Using the system in Example 1, calculate the solution for the initial
conditions 𝑥1 (0)= 1 mm, 𝑥2 (0)= 0, and 𝑥1ሶ (0)= 𝑥2ሶ (0)=0

SOLUTION

Let’s express the solution eqn as:

In Matrix form, it becomes:

3a

Where:
 3b

3c

Differentiating eqn 3b and evaluating the resulting expression

at t = 0 yields

3d
• Eqn 3c and 3d represent four equations in the four unknown
constants of integrations A1, A2, ϕ1, and ϕ2.

• Writing out these four equations yields

i
ii
iii
iv

Adding the last two equations (iii and iv) would yield:

• Solving Eqn. (v) yields ϕ1 = π/2.

• Since ϕ1 = π/2, equation (iv) reduces to
vi

• Eqn. (vi) yields ϕ2 = π/2.

• Substitution of the values of ϕ1 and ϕ2 into equations (i) and (ii)
yields

Thus, the solution becomes:

Eqn. 13
These are plotted as shown:
• Note that in this case, the response of each mass contains both
frequencies of the system

• That is, the responses for both x1(t) and x2(t) are combinations of
signals containing the two frequencies 𝜔1 and 𝜔2 (i.e., the sum of
two harmonic signals).

• Note from the development of eqn. 13 that the mode shapes

determine the relative magnitude of these two harmonic signals