Lecture

11 Two Degree of Freedom Model

“And now we no longer operate in isolation since we are aware of others.”

Chapter outline

This lecture deals with the two-degree-of-freedom systems, which require two independent
coordinates to describe their motion. The coupled equations of motion of the system are derived
using Newton’s second law of motion. By expressing these equations in matrix form, the mas and
stiffness matrices of the system are identified. By assuming harmonic motion of two masses, the
eigenvalues or natural frequencies of vibration, the modal vectors and the free-vibration of the
undamped system are found. The method of incorporating the initial conditions is also outlined.

Learning Objectives

After completing this chapter, the reader should be able to do the following:

• Formulate the equations of motion of the two-degree-of-freedom systems.
• Identify the mass and stiffness matrices from the equations of motion.
• Compute the eigenvalues or natural frequencies of vibration and the modal vectors.
• Determine the free-vibration solution using the initial conditions.

INTRODUCTION

Consider the two degree-of-freedom linear model shown in Fig.11.1 that includes external forcing,
mass, damping, and stiffness elements. The sketch shows the assumed positive directions of
horizontal forces and displacements.

Fig.11.1 Two degree-of-freedom system subjected to external forces

In this section, we will solve the free vibration problem without damping. Referring to Fig.11.1
with 𝑐" = 0, 𝑐% = 0, 𝑐' = 0, 𝐹" 𝑡 = 0 and 𝐹% 𝑡 = 0, i.e. undamped and unforced scenario.

Fig.11.2 Undamped two degree-of-freedom system.


F.B.D. of mass 𝒎𝟏 F.B.D. of mass 𝒎𝟐

Fig.11.3 Free-body diagrams of the two degree-of-freedom system

Applying Newton’s second law of motion to each mass results in the following differential equations
in terms of the unknown displacement functions 𝑥" 𝑡 and 𝑥% 𝑡

𝑚" 𝑥" + 𝑘" + 𝑘% 𝑥" − 𝑘% 𝑥% = 0

𝑚% 𝑥% + 𝑘' + 𝑘% 𝑥% − 𝑘% 𝑥" = 0 11-1

These two equations of motion are coupled, meaning that the independent variables appear in both
equations, and they must be solved simultaneously. The coupling terms is −𝑘% 𝑥% in the first
equation and −𝑘% 𝑥" in the second equation. This mathematical coupling is due to the physical
coupling between the masses and stiffness element 𝑘% , that connect the two masses to each other.

It is customary and convenient to recast the equations in matrix-vector form,

𝑚" 0 𝑥" 𝑘 + 𝑘% −𝑘% 𝑥" 0

+ " 𝑥% = 0
11-2
0 𝑚% 𝑥% −𝑘% 𝑘% + 𝑘'

For simplicity and generality, we rewrite the stiffness matrix as

𝑘" + 𝑘% −𝑘% 𝑘 𝑘"%

= ""
−𝑘% 𝑘% + 𝑘' 𝑘%" 𝑘%%

Equation 11-2, or any degree-of-freedom system of equations, can be written in general matrix-
vector form as

𝑀 𝑥 + 𝐾 𝑥 = 0 11-3

where the matrices 𝑀 and 𝐾 are positive definite of dimension 𝑛 × 𝑛 and the vector of
displacements 𝑥 is of dimension 𝑛 × 1.The mass 𝑀 and stiffness 𝐾 matrices are symmetric:
𝑀 = [𝑀]: , 𝐾 = [𝐾]: with superscript 𝑇 denoting the matrix transpose. We can see that the
coupling occurs in the off-diagonal terms of the property matrices. Without coupling, the equations
of motion can be solved independently as single degree-of-freedom systems.

Note: Two points are important to help understand the solution of Equation 11-2. First,
the individual governing equations are homogeneous and therefore it is only possible to
obtain a solution of the displacement of each mass to within an arbitrary constant multiplier.
Thus, if 𝑥1 𝑡 and 𝑥% 𝑡 are solutions that satisfy the governing differential equations,
then so are 𝐴𝑥" 𝑡 and 𝐴𝑥% 𝑡 (the constant 𝐴 can be canceled since the right-hand
sides equal zero). This was also the case for the single degree- of-freedom free vibration
solution. Second, a harmonic solution can be assumed based on physical considerations in a
general solution.

 The governing equations for the displacement coordinates 𝑥" 𝑡 and 𝑥% 𝑡 are of the same
mathematical form, implying a synchronicity of motion. This means that 𝑥" 𝑡 and 𝑥% 𝑡
vary in a constant proportion to each other as a function of time. This motion of the two
masses is either in phase or 180 degrees out of phase.

We are interested in knowing whether 𝑚" and 𝑚% can oscillate harmonically with the same frequency
and phase angle but with different amplitudes. Assuming that it is possible to have harmonic motion
of 𝑚" and 𝑚% at the same frequency 𝜔 and the same phase angle 𝜙, we take the solutions of
Equation 11-1

𝑥" 𝑡 = 𝑋" 𝑐𝑜𝑠 𝜔𝑡 − 𝜙

𝑥% 𝑡 = 𝑋% 𝑐𝑜𝑠 𝜔𝑡 − 𝜙 11-4

where 𝑋" and 𝑋% are constants that denote the maximum amplitudes of 𝑥" 𝑡 and 𝑥% 𝑡 , and 𝜙
is the phase angle. Substituting Equation 11-4 into Equation 11-1

−𝑚" 𝜔% + 𝑘" + 𝑘% 𝑋" − 𝑘% 𝑋% 𝑐𝑜𝑠 𝜔𝑡 − 𝜙 = 0

−𝑚% 𝜔% + 𝑘' + 𝑘% 𝑋% − 𝑘% 𝑋" 𝑐𝑜𝑠 𝜔𝑡 − 𝜙 = 0 11-5

Since, Equation 11-5 must be satisfied for all values of the time t, the terms between brackets
must be zero. This yields

−𝑚" 𝜔% + 𝑘" + 𝑘% 𝑋" − 𝑘% 𝑋% = 0

−𝑚% 𝜔% + 𝑘' + 𝑘% 𝑋% − 𝑘% 𝑋" = 0 11-6

Equations 11-6 can be written in matrix- vector form,

−𝑚" 𝜔% + 𝑘" + 𝑘% −𝑘% 𝑋" 0

=
−𝑘% %
−𝑚% 𝜔 + 𝑘' + 𝑘% 𝑋% 0

𝑘"" − 𝑚" 𝜔% 𝑘"% 𝑋" 0

% 𝑋 = 11-7
𝑘%" 𝑘%% − 𝑚% 𝜔 % 0


The solution of Equation 11-7 for nontrivial values of 𝑋" and 𝑋% leads to parameter values of 𝜔
called the characteristic values or eigenvalues of the problem. The non- trivial solution can be found
only if the determinant of the characteristic matrix in Equation 11-7 equals zero.

Note: The trivial solutions are 𝑋" = 𝑋% = 0, for which there is no motion.

This requirement is due to the way in which a matrix equation is solved, as we see next. Let Equa-
tion 11-7 be written in shorthand as 𝐾 𝑋 = 0 .A trivial solution can be found by multiplying
both sides by the inverse of 𝐾 , that is, 𝐾 B" 𝐾 𝑋 = 0 . Then 𝑋 = 0 . A nontrivial
solution cannot be obtained if the inverse does not exist. The inverse is given by 𝐾 B" =
𝑎𝑑𝑗 𝐾 det 𝐾 , which exists only if det 𝐾 ≠ 0. Requiring that det 𝐾 = 0 implies that there
is no inverse and no trivial solution.

For nontrivial solutions, we require that 𝑑𝑒𝑡 = 0 in Equation 11-7

 𝑘"" − 𝑚" 𝜔% 𝑘"%
𝑑𝑒𝑡 = 0 11-8
𝑘%" 𝑘%% − 𝑚% 𝜔%

The left-hand side of Equation 11-8 is called the characteristic determinant. We can expand the
determinant to obtain the characteristic equation (also called eigenequation and frequency
equation),

%
𝑚" 𝑚% 𝜔K − 𝑚" 𝑘%% + 𝑚% 𝑘"" 𝜔% + 𝑘"" 𝑘%% − 𝑘"% = 0

with roots given by the quadratic formula,

% %
%
𝑚" 𝑘%% + 𝑚% 𝑘"" 𝑚" 𝑘%% + 𝑚% 𝑘"" 𝑘"" 𝑘%% − 𝑘"% 11-9
𝜔",% = ∓ −
2𝑚" 𝑚% 2𝑚" 𝑚% 𝑚" 𝑚%

The ∓ ordering of the roots follows the convention that the frequencies are numbered from lowest
%
to highest, with 𝜔" called the fundamental frequency. The roots 𝜔",% are known as the eigenvalues
or characteristic values of the system.

The two degree-of-freedom system has two (positive) natural frequencies of oscillation. Once 𝜔"
and 𝜔% have been calculated, either Equation 11-6 or 11-7 can be used to find the ratios 𝑋" 𝑋%
for each 𝜔 that solves the two equations. Since the equations are homogeneous, only relative values
of 𝑋" and 𝑋% can be found.

There is one ratio for each, and we will use the notation 𝑋% 𝑋" " 𝑜𝑟 𝑋%" 𝑋"" for 𝜔" and
𝑋% 𝑋" % 𝑜𝑟 𝑋%% 𝑋"% for 𝜔% , where the first subscript denotes the mass number and the second
denotes the frequency number. These ratios are known as the eigenvectors or characteristic vectors
of the system.

%
For the first eigenvalue 𝜔", , Equations 11-6 and 11-7 become

𝑘"" − 𝑚" 𝜔"% 𝑋"" + 𝑘"% 𝑋%" = 0 11-10

𝑘%% − 𝑚% 𝜔"% 𝑋%" + 𝑘"% 𝑋"" = 0 11-11

Equivalently, these equations can be written as

𝑘"" 𝑋"" + 𝑘"% 𝑋%" = 𝑚" 𝜔"% 𝑋""

𝑘%" 𝑋"" + 𝑘%% 𝑋%" = 𝑚% 𝜔"% 𝑋%"

or, in matrix-vector form,

𝐾 𝑋 " = 𝜔"% 𝑀 𝑋 " 11-12

where 𝑋 " = 𝑋"" 𝑋%" : .Either Equations 11-10 and 11-11 can be solved for the
eigenvectors,

𝑋%" 𝑘"" − 𝜔"" 𝑚" 𝑘"%

=− =− 11-13
𝑋"" 𝑘"% 𝑘%% − 𝜔"% 𝑚%

and similarly for the second eigenvalue 𝜔%% ; 𝐾 𝑋 % = 𝜔22 𝑀 𝑋 % with the eigenvectors
𝑋%% 𝑘"" − 𝜔%% 𝑚" 𝑘"%
=− =− 11-14
𝑋"% 𝑘"% 𝑘%% − 𝜔%% 𝑚%

where 𝜔"% and 𝜔%% are substituted from Equation 11-9.

Equations 11-13 and 11-14 represent the relative displacement magnitudes of each mass for the
frequencies 𝜔" and 𝜔% , respectively, and are known as the modal ratios. An equation exists for
each mode and the modal characteristics of a system include its frequency and mode shape, denoted
by 𝜔R 𝑋 R , 𝑖 = 1, 2 for this two degree-of- freedom case. The solution is incomplete until the
initial conditions are satisfied as we see next.


The motion of the system is the linear superposition of the two solutions (modes). From Equations
11-4, we have

𝑥" (𝑡) 𝑋 𝑋
= 𝐶" "" 𝑐𝑜𝑠 𝜔" 𝑡 − 𝜙" + 𝐶% "% 𝑐𝑜𝑠 𝜔% 𝑡 − 𝜙% 11-15
𝑥% (𝑡) 𝑋%" 𝑋%%

where four initial conditions, 𝑥" (0), 𝑥% (0), 𝑥" (0) and 𝑥% (0), are needed to evaluate the
constants, 𝐶" , 𝐶% , 𝜙" and 𝜙% . In order to substitute the modal ratios of Equations 11-13 and
11-14 into Equation 11-15, we can rewrite Equation 11-15 in the following equivalent form,

𝑥" (𝑡) 1 1
= 𝐶" 𝑐𝑜𝑠 𝜔" 𝑡 − 𝜙" + 𝐶% 𝑐𝑜𝑠 𝜔% 𝑡 − 𝜙% 11-16
𝑥% (𝑡) 𝑋21 𝑋11 𝑋22 𝑋12

where 𝐶" = 𝐶" 𝑋"" and 𝐶% = 𝐶% 𝑋"% are new constants.

Equation 11-16 indicates that the response of each mass is a sum of natural harmonics, where the
proportion contributed by each harmonic to the total response is given by the modal ratio. The
numerical values of 𝐶" , 𝐶% , 𝜙" and 𝜙% are determined by satisfying the four initial conditions,
that is, by simultaneously solving the following four equations,

𝑥" (0) 1 1
= 𝐶" 𝑐𝑜𝑠 −𝜙" + 𝐶% 𝑐𝑜𝑠 −𝜙% 11-17
𝑥% (0) 𝑋21 𝑋11 𝑋22 𝑋12

𝑥" (0) 1 1
= −𝐶" 𝜔" 𝑠𝑖𝑛 −𝜙" + 𝐶% 𝜔% 𝑠𝑖𝑛 −𝜙% 11-18
𝑥% (0) 𝑋 21 𝑋 11 𝑋 22 𝑋12

If for some particular initial conditions 𝐶%

= 0, then the two degree-of-freedom structure will
oscillate only in its first mode at its first natural frequency. Similarly, if 𝐶" = 0, the structure
will oscillate solely in its second mode at its second natural frequency.

To emphasize, a mode of oscillation is one in which all points on a system execute simple harmonic
motion at the same frequency and phase. This is known as synchronous motion. The amplitude
generally varies from point to point. Fig. 11.4 shows the two modes of a two degree-of-freedom
model in synchronous motion.

Fig. 11.4: The synchronous motion of the two modes of a two degree-of-freedom system. The
actual motion is a combination of the two modes of vibration.

Example: Two Degree of Freedom, Undamped

Consider the system of Fig. 11.1, simplified to the undamped, unforced system in Fig. 11.2. Let
𝑚" = 𝑚, 𝑚% = 2𝑚, 𝑘" = 𝑘, 𝑘% = 2𝑘, 𝑘' = 3𝑘, and obtain the natural frequencies and modes
of vibration.

Fig.11.5 Two degree of freedom system

Solution: The equations of motion are

𝑚𝑥" + 3𝑘𝑥" − 2𝑘𝑥% = 0

2𝑚𝑥% + 5𝑘𝑥% − 2𝑘𝑥" = 0

In matrix-vector form, the symmetries of the mass and stiffness matrices are evident,

𝑚 0 𝑥" 3𝑘 −2𝑘 𝑥" 0

+ = 11-19
0 2𝑚 𝑥% −2𝑘 5𝑘 𝑥% 0

Assuming harmonic responses of the form

𝑥" 𝑡 = 𝑋" 𝑐𝑜𝑠 𝜔𝑡 − 𝜙

𝑥% 𝑡 = 𝑋% 𝑐𝑜𝑠 𝜔𝑡 − 𝜙

differentiating twice, and substituting into the governing Equation 11-19 Combining terms, we
have
3𝑘 − 𝑚𝜔% 𝑘"% 𝑋" 0
% 𝑋 = 11-20
𝑘%" 5𝑘 − 2𝑚𝜔 % 0

Setting the determinant of the coefficient matrix to zero leads to the characteristic equation

2𝑚% 𝜔K − 11𝑚𝑘𝜔% + 11𝑘 % = 0,

with roots

%
%
11𝑚𝑘 11𝑚𝑘 22(𝑚𝑘)%
𝜔",% = ∓ −4
4𝑚% 4𝑚% 16𝑚K

] ] ]
= 2.75 ∓ 1.44 = 1.31 and 4.19
^ ^ ^

The two natural frequencies are then

𝑘 𝑟𝑎𝑑 𝑘 𝑟𝑎𝑑
𝜔" = 1.14 , 𝜔% = 2.05 ,
𝑚 𝑠 𝑚 𝑠

With the natural frequencies known, Equation 11-13 (or Equation 11-20) can be used to obtain
the respective modes of vibration,

𝑋%" 𝑘"" − 𝜔"% 𝑚" 3𝑘 − 1.31 𝑘 𝑚 𝑚

=− =− = 0.845
𝑋"" 𝑘"% −2𝑘

𝑋%% 𝑘"" − 𝜔%% 𝑚" 3𝑘 − 4.19 𝑘 𝑚 𝑚

=− =− = −0.595
𝑋"% 𝑘"% −2𝑘

The denominators cannot equal zero since the stiffness must be positive. Physically, zero stiffness
would imply that one mass is motionless while the other is vibrating. This can occur only if the
masses are not coupled. The modes are sketched in Fig.11.6.

Fig.11.6 Two modes of vibration: the first mode (top) and the second mode (bottom). Even though
the masses are only at location 1 and 2, dotted lines are drawn connecting these displacements

Modes generally have amplitudes of alternating signs and, therefore, the modal amplitude lines
(The lines are drawn for better visualization of the eigenvectors.) cross the zero ( − 1) times where
is the mode number. The crossing points are called nodes, indicative of no motion.

The complete response, following Equation 11-16, is

𝑥" (𝑡) 1 𝑘 1 𝑘
= 𝐶" 𝑐𝑜𝑠 1.14 𝑡 − 𝜙" + 𝐶% 𝑐𝑜𝑠 2.05 𝑡 − 𝜙%
𝑥% (𝑡) 0.845 𝑚 −0.595 𝑚

The second mode in the figure has one node. The response of each mass is a combination of two
harmonics that are weighted by the respective modal ratios.

Given the initial conditions 𝑥" (0), 𝑥% (0), 𝑥" (0) and 𝑥% (0), the constants 𝐶" , 𝐶% , 𝜙" and 𝜙%
can be found by solving the four simultaneous algebraic Equations 11-17 and 11-18

Harmonically Forced Vibration

Next, we consider the undamped vibration of a two degree-of-freedom system acted on by external
forces. We begin this case of forced vibration with the generic two degree-of-freedom matrix-vector
equation of motion, Equation 11-20, with 𝐹" (𝑡) and 𝐹% (𝑡)applied to the respective masses,

𝑚" 0 𝑥" 𝑘 𝑘"% 𝑥" 𝐹" (𝑡)

+ "" 𝑥% = 𝐹% (𝑡)
11-21
0 𝑚% 𝑥% 𝑘%" 𝑘%%

where, due to the symmetry of the property matrices 𝑘"% = 𝑘%" . First, we examine the response
due to harmonic 𝐹" (𝑡) only,

𝐹" 𝑡 = 𝑓" 𝑐𝑜𝑠𝜔𝑡

𝐹% 𝑡 = 0

where 𝜔 is the forcing frequency. Since there is no damping, the response will either be in phase
or 180b out of phase with the forcing.

The particular solution also oscillates at the forcing frequency,

𝑥" 𝑡 = 𝑋c" 𝑐𝑜𝑠𝜔𝑡

𝑥% 𝑡 = 𝑋c% 𝑐𝑜𝑠𝜔𝑡

where amplitudes 𝑋c" and 𝑋c% need to be determined. Here, we look for the particular solutions
only; the total solution must include both the homogeneous and the particular solutions.

Substituting 𝑥" 𝑡 and 𝑥% 𝑡 and their second derivatives into Equation 11-21 and canceling the
common co- sine term results in

𝑚" 0 𝑋c" 𝑘 𝑘"% 𝑋c" 𝑓

−𝜔% + "" = " 11-22
0 𝑚% 𝑋c% 𝑘%" 𝑘%% 𝑋c% 0

Combining terms, we find an expression that looks very similar to the eigenvalue problem, except
that the algebraic equations are nonhomogeneous,

𝑘"" − 𝑚" 𝜔% 𝑘"% 𝑋c" 𝑓

% 𝑋 = "
𝑘%" 𝑘%% − 𝑚% 𝜔 c% 0

Denoting the square matrix, which is the system characteristic matrix, by 𝑆 or 𝑆 𝑗𝜔 , we can
solve for 𝑋c" and 𝑋c% by taking the inverse of 𝑆 ,

𝑋c" B" 𝑓"

= 𝑆
𝑋c% 0

From matrix algebra, the inverse equals the adjoint matrix divided by the determinant,

B"
𝑎𝑑𝑗 𝑆
𝑆 =
𝑑𝑒𝑡 𝑆

and thus we can write

𝑋c" 1 𝑘%% − 𝑚% 𝜔% −𝑘"% 𝑓"

= 11-23
𝑋c% 𝑑𝑒𝑡 𝑆 −𝑘%" 𝑘"" − 𝑚" 𝜔% 0

where the response amplitudes, 𝑋c" and 𝑋c% are functions of the forcing frequency Expanding
the matrix equation, 𝑋c" and 𝑋c% are given by

𝑘%% − 𝑚% 𝜔%
𝑋c" = 𝑓"
𝑑𝑒𝑡 𝑆

−𝑘"%
𝑋c% = 𝑓
𝑑𝑒𝑡 𝑆 "
where

%
𝑑𝑒𝑡 𝑆 = 𝑘%% − 𝑚% 𝜔% 𝑘"" − 𝑚" 𝜔% − 𝑘"%

The determinant of 𝑆 , 𝑑𝑒𝑡 𝑆 is the characteristic polynomial with the squares of natural
frequencies as roots, equivalent to the characteristic polynomial obtained in Equation 11-8 and
roots in Equation 11-9. Therefore, at the natural frequencies, the equations for 𝑋c" and 𝑋c% will
have a zero in the denominator, resulting in an unbounded response. From the single degree-of-
freedom case, Equation 6-4, we already know that if the forcing frequency equals the natural
frequency, the response grows without bound for an undamped system. For a multi degree-of-
freedom system, such an instability can occur for each natural frequency.

Fig. 11.7 and 11.8 show the response amplitudes of the two degree-of-freedom system in Fig.
11.5 for parameter values 𝑚" = 𝑚% = 1 𝑘𝑔, 𝑘" = 𝑘% = 𝑘' = 1𝑁/𝑠, and 𝑓" = 1 𝑁, and 𝑓% = 1 𝑁. The
corresponding mass and stiffness matrices are

1 0 2 −1
𝑀 = 𝑎𝑛𝑑 𝐾 =
0 1 −1 2

The natural frequencies are 1.00 and 3 = 1 73 rad/s. These figures are analogous to Fig. 6.2
for the single degree-of-freedom system.


Fig. 11.7: Response amplitude 𝑿𝟏 as a function of driving frequency 𝝎 for 𝒇𝟏 = 𝟏 and 𝒇𝟐 = 𝟎.

The magnitude grows to infinity at the two natural frequencies: 𝝎 = 𝟏. 𝟎𝟎 and 1.73 rad/s.

Fig. 11.8: Response amplitude 𝑿𝟐 as a function of driving frequency 𝝎 for 𝒇𝟏 = 𝟏 and 𝒇𝟐 = 𝟎.

The magnitude grows to infinity at the two natural frequencies: 𝝎 = 𝟏. 𝟎𝟎 and 1.73 rad/s.