NPTEL – Mechanical Engineering – Nonlinear Vibration

References:
1. Pratiher, B., Dwivedy, S.K.: Nonlinear Dynamic of a Flexible Single –Link Cartesian
Manipulator, International Journal of Non-linear Mechanics 42, 1062-1073 (2007).
2. A. H. Nayfeh, Method of Normal Forms, John Wiley & Sons, INC, Canada 1993.
3. A. H. Nayfeh and B. Balachandran, Applied Nonlinear Dynamics-Analytical, Computational
and Experimental Methods, John Wiley & Sons, INC, Canada 1995.

Module 3 Lecture 10
INCREMENTAL HARMONIC BALANCE METHOD

Lau and Cheung (1981) developed incremental harmonic balance method. A practical weakness
of perturbation methods is that carrying out the expansion to higher order is very cumbersome,
especially for multiple degree of freedom systems. In practice it is difficult to go beyond the
third order unless the algebraic manipulations are performed by a computer (Cheung et al. 1990).
In Incremental Harmonic Balance Method (IHB) one can deal with strongly non linear systems
to any desired accuracy. This method is a combination of the incremental method (Newton-
Raphson procedure) with the harmonic balance method (Ritz and Galerkin’s averaging method).
It is exactly equivalent to a Galerkin procedure followed by a Newton-Raphson method.

The method possesses advantages in studying systems with severe nonlinearities and is easily
applied to systems with harmonic (or, more generally, periodic) excitation. Some insight into the
solution method is lost, however, since the problem of solving the original governing differential
equations is replaced with that of solving a second "simpler" set of equations involving
increments in the motion, exciting force and/or frequency of excitation. Ferri (1986) shown that
the IHB method is exactly equivalent to the Harmonic Balance Newton Raphson Method
(HBNR). Here this method is illustrated by taking the example of a multi degree of freedom
nonlinear system.

For a multi degree of freedom system with cubic non linearities, the non linear equations of
motion in general can be written as

n d 2q j n dq j n n n n
∑ M ij + ∑ Cij
dt 2 =j 1 dt =j 1
+ ∑ Kij q j + ∑∑∑ α ijkl q j qk ql (3.10.1)
=j 1 =j 1 =
k 1 =l 1
= fi cos(2m − 1)ω t, i= 1,2,..., n.

by substituting τ = ω t one may write (3.10.1) as

Joint initiative of IITs and IISc – Funded by MHRD Page 48 of 67

 NPTEL – Mechanical Engineering – Nonlinear Vibration

n n n n n n
ω ∑ M ij q j + ω ∑ Cij q j + ∑ Kij q j + ∑∑∑ α ijkl q j qk ql
2

=j 1 =j 1 =j 1 =j 1 =
k 1 =l 1
(3.10.2)
= fi cos(2m − 1)τ , =
i 1,2,..., n.

The q j are the unknowns of the system, the dots denote derivatives with respect to the
dimensionless time τ, and M ij , Cij , K ij , α ijkl , fi and ω are coefficients of the mass, damping, linear
stiffness, cubic stiffness, and excitation amplitude and excitation frequency respectively.
Equation (3.10.2) can be written in the matrix form as

ω 2 Mq + ω Cq + ( K + =
K n )q F cos(2m − 1)τ (3.10.3)

=
where q [ q= T
1, q2 ,....., qn ] , F [ f1, f 2 ,...., f n ]T , M,C and K are mass, damping, linear stiffness
matrices, with elements are denoted by M ij , C ij and K ij respectively, and K n is the cubic non-
linear stiffness matrix, its element α ijkl being taken in the form
n n
K nij = ∑∑ K ijkl qk ql (3.10.3)
=
k 1 =l 1

The first step of the IHB Method is a Newton-Raphson procedure. Let q jo , fio and ωo denote a
state of vibration; the neighboring state can be expressed by adding the corresponding
increments to them as follows:

q=
j q jo + ∆q j ,=
j 1, 2,...., n, (3.10.4)
f=
i fio + ∆fi ,=
i 1, 2,..., n, (3.10.5)
ω= ω0 + ∆ω. (3.10.6)
Substituting expansions (3.10.4)-(3.10.6) into equation (3.10.2) and neglecting small terms of
higher order, one obtains the following linearized incremental equation in matrix form:
ω02 M ∆q + ω0 C∆q + ( K + 3K n )∆q = R − (2ω0 M q0 + Cq0 )∆ω + cos(2m − 1)τ∆F , (3.10.7)
=R F0 cos(2m − 1)τ − (ω02 M q0 + ω0 Cq0 + Kq0 + K n q0 ), (3.10.8)

in which q0 , ∆q, F 0 , ∆F and K nij are given below.

q0 = [ q10 , q20 ,....., qn0 ]T , ∆q =[ ∆q1, ∆q2 ,....., ∆qn ]T , F0 = [ f10 , f 20 ,....., f n 0 ]T ,
n n
[ f1 , ∆f 2 ,....., ∆f n ] and K nij = ∑∑ α ijkl qko q10 .
∆F =∆ T

=
k 1 =l 1

R is a corrective vector which goes to zero when the solution is reached.

Joint initiative of IITs and IISc – Funded by MHRD Page 49 of 67

 NPTEL – Mechanical Engineering – Nonlinear Vibration

The second step of the IHB method is the Galerkin’s procedure. Because equation (3.10.2) is odd
and the excitation force is periodic, one can assume for steady state response,
Nc Ns

=
=q j0
k 1=k 1
∑ a jk cos(2k − 1)τ + ∑ b jk sin(2
= k − 1)τ Cs Aj , (3.10.9)
Nc Ns
∆q j = ∑ ∆a jk cos(2k − 1)τ + ∑ ∆b jk sin(2k − 1)τ = Cs ∆ Aj , (3.10.10)
=k 1=k 1
Where
Cs = [cos τ , cos 3τ ,......., cos(2 N c − 1)τ ,sin τ ,sin 3τ ,....,sin(2 N s − 1)τ ],
Aj = [a j1 , a j 2 ,...., a jNc , b j1 , b j 2 ,......, b jN s ]T ,
∆Aj =∆
[ a j1 , ∆a j 2 ,...., ∆a jNc , ∆b j1 , ∆b j 2 ,......, ∆b jN s ]T .

Hence the vectors of unknowns and their increments can be expressed by the Fourier coefficients
vector A and its increment ΔA as follows:
q0 = SA, (3.10.11)
∆q = S ∆A (3.10.12)

where S, A and ΔA are given as follows.

Cs 0
S =   , A = [ A , A ,......, A ]T , and ∆A =[∆A , ∆A ,......, ∆A ]T ,
 1 2 n 1 2 n

 0 Cs 
2π 2π 2π
 τ , C =  τ, K =
M= ∫ S T MSd ∫ S T CSd ∫S KSdτ ,
T

0 0 0
2π 2π 2π
K Sdτ , F ∫ S
∫ S= ∫S
(3)
K (3)
= T T
1)τ dτ , R f
F 0 cos(2m −= T
cos(2m − 1)τ dτ ,
0 0 0
Substituting equations (3.10.11) and (3.10.12) into equation (3.10.7) and using the Galerkin’s
procedure gives
2π

∫ δ (∆q) [ω
(3)
T 2
0 M ∆q + ω0 C∆q + ( K + 3K )∆q ]dτ
0
2π
(3.10.13)
= ∫ δ (∆q) [ R − (2ω M q + Cq0 )∆ω + cos(2m − 1)τ∆F ]dτ .
T
0 0
0

One can easily obtain a set of linear equations in terms of ∆A , ∆ω and ∆F ,

K mc ∆A = R − Rmc ∆ω + R f ∆F , (3.10.14)
in which
K mc = ω02 M + ω0C + 3K n (3.10.15)
R = F − (ω M + ω0C + K + K ) A,
2
0
(3)
(3.10.16)
Rmc (2ω0 M + C ) A,
= (3.10.17)

Joint initiative of IITs and IISc – Funded by MHRD Page 50 of 67

NPTEL – Mechanical Engineering – Nonlinear Vibration

It is worth mentioning that in equation (3.10.14) the number of incremental unknowns is greater
than the number of equations available due to the existence of ∆F and ∆ω . Since one is
primarily interested in the frequency-response curves of the system for a constant level, F is
fixed as a parameter vector, which implies ∆F = 0 . Hence equation (3.10.14) is reduced to
K mc ∆A = R − Rmc ∆ω. (3.10.18)
The solution process starts from a suggested solution (in general, from a corresponding known
linear solution), and then the non-linear amplitude frequency response is solved point by point by
incrementing frequency ω or incrementing one component of the amplitudes A. The Newton-
Raphson iteration can be applied within an incremental step. In the incremental process, an
increment which is prescribed a priori is called a control or active increment. If ∆ω is specified
as a control increment, then ω remains constant through the iterative process: i.e. ∆ω = 0 , while
other increments are solved from the equation
K mc ∆A = R (3.10.19)
The process is repeated until the magnitude of the corrected vector R is acceptably small-in
which case a solution is obtained. This process is called iteration. The value of ω is then
augmented an increment ∆ω artificially, and a new iteration is repeated with the new value of ω
until a new solution is obtained. The above process is called an augmentation. The whole
solution process is an alternative application of augmentation and iteration.
The above incremental process in which ∆ω is taken as active increment is called ω-
incrementation. Similarly, it is equally possible to have amplitude incrementation. In this case,
one component of ΔA, say ∆a jk , is specified as the control increment; then a jk remains
constant. ∆a jk =
0 through the iteration and one has to solve equation (3.10.18) to obtain other
increments of ΔA and ∆ω . After the amplitude of R has reached the desired accuracy, the
iteration is terminated and a new augmentation can be started by adding an increment on a jk .
This process is called a jk incrementation. In practice, the active increment is chosen as the one
that varies faster and therefore the ω-incrementation or the a jk incrementation can be adopted
along the response curves.
If one is interested in the forcing amplitude response curves of the system for a constant
frequency level, then ∆ω = 0 and ∆fi =0 , j ≠ i , and hence equation (3.10.14) is reduced to
K mc ∆A = R + R f ∆F (3.10.20)
Example 3.10.1:
Consider a two degree of freedom system consisting of two point masses and two springs with a
linear damper, under a harmonic excitation shown in Figure 3.10.1. Find the solution of the
system using incremental harmonic balance method.

Joint initiative of IITs and IISc – Funded by MHRD Page 51 of 67

 NPTEL – Mechanical Engineering – Nonlinear Vibration

Fig 3.10.1: Schematic diagram of a two degree of freedom system with cubic nonlinear spring.

Solution:
One of the springs is linear with the stiffness coefficient k10 and the other has a cubic non
linearity. Its restoring force is defined as
f12= k12 (q1 − q2 ) + µ (q1 − q2 )3 (3.10.21)

The differential equations of motion of the system can be written in non dimensional form as
q1 + k 2 q1 + γ (q1 − q2 ) + µγ l (q1 − q2 ) + µγ (q1 − q2 )3= p cos Ωt , (3.10.22)
q2 + (q2 − q1 ) − µ l (q1 − q2 ) − µγ (q1 − q2 )3 =
0, (3.10.23)
Where
γ = m1 / m2 , t = t (k12 / m2 ) , k 2 = k10γ / k12 , l = l (k12 / m2 ),
q = dq / dt , µ = µ / k12 , and p = pγ / k12 .
q1 , q2 are displacements of point masses, t is time and m1 , m2 , k10 , k12 , µ , l , Ω and p are
the masses of the system, coefficient of linear stiffness, coefficient of non linear stiffness,
coefficient of damping, excitation frequency and excitation amplitude respectively.

In the solution process, the number of harmonic terms is taken as N= c N=s 2:

q1 a11 cos τ + a12 cos 3τ + b11 sin τ + b12 sin =
= 3τ A11 cos(τ + φ11 ) + A12 cos(3τ + φ12 ), (3.10.24)

q2 a12 cos τ + a22 cos 3τ + b21 sin τ + b22 sin=

= 3τ A21 cos(τ + φ11 ) + A22 cos(3τ + φ22 ), (3.10.25)
Where
Aij = aij2 + bij2 , φij =tan −1 (−bij / aij ), i =1, 2, j =1, 2.
There exist two types of non trivial solutions:
a) Fundamental resonance only,= φ12 ), q2 A22 cos(3τ + φ22 ),
i.e. q1 A12 cos(3τ +=

b) Both fundamental resonance and sub harmonic resonance occur simultaneously: i.e. q1 and
q2 take the form of the equations (3.10.24) and (3.10.25).

Exercise problems:

Joint initiative of IITs and IISc – Funded by MHRD Page 52 of 67

 NPTEL – Mechanical Engineering – Nonlinear Vibration

1. Use incremental harmonic balance method to find the frequency response equations for the
Duffing equation with cubic nonlinearity.

2. Use incremental harmonic balance method to find the frequency response equations for the
Duffing equation with quadratic and cubic nonlinearity.

3. Use incremental harmonic balance method to find the frequency response equations for the
van der Pol’s equation.

4. The equation of motion of a bimaterial beam with alternating magnetic field and thermal loads
can be given by following equation. Use incremental harmonic balance method to solve this
equation (refer Wu, 2009).
∂ 2v
m 2 + Cd
∂t
∂v
∂t
∂ 4 v ∂c ∂  x
+ ( Et I t + El I l ) 4 + +  ∫ pd ξ
∂x ∂x ∂x  0 ( ) ∂v 
+
∂x  
 A γ
t t ( ∆T ) + A γ
l l ( ∆T )
∂ 2v
 ∂x 2 =
 0

The equation in its temporal form can be written as

d 2w dw
Ω 2 2 + 2Ω  k1 + k2 (1 + cos 2τ ) w2  + (1 − 2ϕ cos 2τ ) w =0
dτ dτ
References:

1. S.L. Lau and Y.K Cheung, Amplitude incremental variational principle for nonlinear vibration of
elastic systems, Journal of Applied Mechanics, 48, 959-964, 1981.
2. Y.K Cheung, S. H. Chen, S.L. Lau, Application of the incremental harmonic-balance method to
cubic nonlinearity systems. Journal of Sound and Vibration, 140(2), 273–286, 1990.
3. A.Y.T. Leung, S.K Chui, Nonlinear vibration of coupled Duffing oscillators by an improved
incremental harmonic balance method. Journal of Sound and Vibration, 181(4), 619–633, 1995.
4. A. Raghothama, S. Narayanan, Non-linear dynamics of a two-dimensional airfoil by incremental
harmonic balance method Journal of Sound and Vibration, 226(3), 493–517, 1999.
5. A. A. Ferri, On the Equivalence of the Incremental Harmonic Balance Method and the Harmonic
Balance-Newton Raphson Method,ASME, Journal of Applied Mechanics, 53, 455-457, 1986.
6. Y. Shen, S. Yang and X. Liu, Nonlinear dynamics of a spur gear pair with time-varying stiffness
and backlash based on incremental harmonic balance method. International Journal of
Mechanical Sciences, 48, 1256-1263, 2006.
7. G. Y. Wu, The analysis of dynamic instability of a bimaterial beam with alternating magnetic
fields and thermal loads, Journal of Sound and Vibration, 327, 197–210, 2009.

Joint initiative of IITs and IISc – Funded by MHRD Page 53 of 67