Professional Documents
Culture Documents
1. INTRODUCTION
There are innumerable engineering problems which can be modelled as a set of second
order, non-linear, ordinary differential equations. If such equations do not explicitly
contain the independent time variable, the system is called autonomous; otherwise, it is
called non-autonomous. Examples of autonomous systems are vortex-induced vibrations,
the galloping of tower guys or transmission lines and the free vibrations of a centrifugal
pendulum; and examples of non-autonomous systems are vibrations in machines and
oscillations in electrical networks which have sources. Although the autonomous systems
appear merely as a special case of more general non-autonomous ones, there exist
fundamental differences between these two systems. For example, autonomous systems
usually have equilibrium solutions, while non-autonomous systems do not. On the other
hand, non-autonomous systems more easily exhibit chaotic motions than autonomous
systems. In general, the analysis of a non-linear dynamic system involves three related
phases: (i) finding the equilibrium points and their stability; (ii) determining the possible
periodic and quasi-periodic dynamic motions which bifurcate from an equilibrium
state; and (iii) evaluating the stability of the dynamic solutions. For a non-autonomous
system, only the second and third phases are involved. The main purpose of this paper
is to focus upon the analysis of periodic and quasi-periodic motions. Although such a
system may exhibit chaos for certain parameter values, such motions are not investigated
here.
To quantitatively analyze the dynamic behavior of a non-linear system, many
approaches have been developed, among them are the LindstedtPoincare method,
multiple time scaling, harmonic balancing, time-averaging, the KrylovBogoliubovMitropolsky (KBM) method, and the Galerkin and Ritz procedures (see, e.g., reference [1], Part
II of reference [2] and Chapter 8 of reference [3]). These methods can be applied to both
413
0022460X/96/170413 + 26 $18.00/0
414
. .
415
method. The two methods will be shown rigorously to be equivalent for a general n-DOF,
non-linear system. Such a system may involve not only external but also internal
resonances, as well as a non-resonance.
A new technique has been developed, based on the KB and Galerkin methods, to find
explicit solutions and their stability conditions, especially for the internal resonances of
a multiple DOF, autonomous system [9, 10]. The CG method was modified, by using time
averaging, to produce time-averaged, differential equations rather than algebraic
equations. This modification is termed the enhanced method of Galerkin (EG method).
The proof of the equivalence of the KB and EG methods will be given here and includes
the stability analysis which was excluded from Newlands work. Although the KB and
EG methods will be shown to produce identical first order periodic and quasi-periodic
solutions as well as the same stability conditions, the KB method is preferred for
non-resonance as well as for combinations of the non-resonant and resonant cases in which
the number of resonant modes is fewer than, say, three. On the other hand, the EG method
may be utilized advantageously for resonances in which more than two modes are involved.
The next section is devoted first to proving that the KB and EG methods produce
identical first order solutions as well as stability conditions for a general n-DOF, weakly
non-linear system. Then the relations of the KB, EG and CG methods will be shown from
the results. Applications of the theory to the galloping problem are given subsequently.
Two typical resonant cases are detailed. Conclusions are drawn in the last section.
i, j = 1, 2, . . . , n,
(1)
where the dots indicate differentiations with respect to normalized time t, and hi and vi
are the ith normalized co-ordinate and eigenvalue, respectively, of the corresponding
(e = 0) linear system. The fi is the analogous normalized force, including all the non-linear
terms as well as all the damping terms and n is the frequency of the external
forcing function. The parameter e (01) represents a weak non-linearity. The internal
non-resonance and resonances of an autonomous system, which does not involve external
forcing functions, will be considered in subsections 2.12.4. Then the derivations will be
extended directly in subsection 2.5 to non-autonomous systems. The relations of the KB,
EG and CG methods are given in subsection 2.6.
2.1. -
It follows from equation (1) that, when e = 0, a response can be written in the form
hi (t) = Xi cos (vi t + fi ),
i = 1, 2, . . . , n,
(2)
where constants Xi and fi , respectively, represent the ith components amplitude and
phase, which are determined from the initial conditions. The solutions for e $ 0 can be
considered to be a perturbation of solution (2) and, therefore, may be expressed by
hi (t) = Xi (t) cos[vi t + fi (t)] 0 Xi (t) cos Ci (t),
i = 1, 2, . . . , n.
(3a)
Now, the Xi (t) and fi (t) are not constants but are slowly varying variables such that
their derivatives, Xi (t) and f i (t), are small (e order) terms. Equation (3a) belongs to the
. .
416
(3b)
i = 1, 2, . . . , n,
(4a)
where v*
i , unlike vi , is unknown. For a steady state solution, the amplitudes Xi (t) (or
X*
i . Moreover, the frequencies v i equal v*
i (t)) are constants, X
i for solution (4a), whereas
they equal vi + f i (t) for solution (3a). That is, for a steady state solution, f i (t) $ 0 but
f i* (t) = 0. An alternative expression for equation (4a) is
hi (t) = A*
i (t) cos v*
i t + B*
i (t) sin v*
i t.
(4b)
The choice between the KB solution (3a) or (3b) (or, alternatively, the EG solution (4a)
or (4b)) depends upon the ease of deriving explicit solutions by using the time-averaging
procedure. In fact, equation (3a) (equation (4a)) can be transformed to equation (3b)
(equation (4b)).
It will be shown in the following subsections that the two formulations (3a) and (4a)
(or (3b) and (4b)) give identical first order approximations.
2.2. -
This case implies that a ratio of any two natural frequencies is not close to a ratio of
two positive integers. Suppose that the periodic or quasi-periodic solutions are given by
equation (3a). Differentiate this equation with respect to t and let the amplitudes and
phases be chosen such that
h i (t) = vi Xi (t) sin Ci (t)
(5)
(6)
and
It should be noted that the last two equations cannot be obtained without assuming that
Xi (t) and fi (t) are varying slowly. Next, differentiate equation (5) with respect to t and
substitute the resulting equation and equation (2) into equation (1), to yield
Xi (t) sin Ci (t) + Xi (t)f i (t) cos Ci (t) =
01
e
f [X (t) cos Cj (t), vj Xj (t) sin Cj (t)].
vi i j
(7)
Solving equations (6) and (7) for Xi (t) and f i (t) and then employing time-averaging leads
to
Xi =
01
e
1
lim
vi T:a T
(8)
417
and
f i =
0 1
e
1
lim
vi Xi T:a T
(9)
where t has been suppressed from Xi (t) and Ci (t) for brevity.
Following the same procedure, the analogous equations from form (4a) are
0 1
0 1
Xi* =
1
1
lim
v*
T
T:a
i
e
1
lim
v*
T:a T
i
T
2
sin C*
vi2 )hi + efi (X*
i
i [(v*
j cos C*
j , v*
j X*
j sin C*
j )] dt
sin C*
i fi (X*
j cos C*
j , v*
j X*
j sin C*
j ) dt
(10)
and
f i* =
0 1
1
1
lim
v*
T:a T
i X*
i
T
2
cos C*
vi2 )hi + efi (X*
i
i [(v*
j cos C*
j , v*
j X*
j sin C*
j )] dt
0 1
v*2 vi2
e
1
= i
lim
2v*
v*
X*
T
T:a
i
i
i
cos C*
i fi (X*
j cos C*
j , v*
j X*
j sin C*
j ) dt.
(11)
Generally, explicit integrals cannot be obtained from equations (8)(11) because the Xi
(X*
i ) and Ci (C*
i ) are functions of t. However, they may be approximated by assuming
that variables Xi (X*
i ) and Ci (C*
i ) are constants in a period (which implies that the ith
component of the motion is periodic) because these variables are varying slowly. These
approximate, explicit integrals can be obtained by expanding the fi into multi-variable
Fourier series. It should be noted that non-zero residuals arise from those terms of the
Fourier series containing cos Ci (cos C*
i ) and sin Ci (sin C*
i ) (i.e., j = i) because they
correspond to the non-resonant case.
2.2.1. Steady state solutions
First, consider the KB solution (3a). The amplitude of the steady state solution, Xi , is
determined by setting Xi = 0 in equation (8). This yields
1
T:a T
lim
(12)
0 1
e
1
lim
vi Xi T:a T
(13)
The last equation is in the form v i = vi + O(e), where O(e) represents an e order term,
as expected.
. .
418
Next, the steady states for the EG solution (4a) can be found similarly, but from
equations (10) and (11), by setting Xi* = f i* = 0. This results in
1
T:a T
lim
sin v*
i tfi (X*
j cos v*
j t, v*
j X*
j sin v*
j t) dt = 0
(14)
and
0 1
2
v*
vi2
e
1
i
=
lim
v*
T
2v*
X*
T:a
i
i
i
cos v*
i tfi (X*
j cos v*
j t, v*
j X*
j sin v*
j t) dt.
(15)
sin v*
i tfi (X*
j cos v*
j t, vj X*
j sin v*
j t) dt + O(e) = 0
(16)
and
v*
i = vi
0 1$
e
vi X*
i
1+
O(e)
2vi
%$
1
1
T:a T
lim
cos v*
i t
fi (X*
j cos v*
j t, vj X*
j sin v*
j t ) dt + O(e)
or
v*
i = vi
0 1
e
1
lim
vi X*
T
T:a
i
T
2
cos v*
i tfi (X*
j cos v*
j t, vj X*
j sin v*
j t ) dt + O(e ).
(17)
It can be seen from equations (16) and (17) that their leading (first) order approximate
forms are identical to equations (12) and (13), respectively. Therefore, the KB and EG
methods produce the same first order amplitudes and frequencies for the steady state
motions. In particular,
X*
i Xi = O(e)
and
v*
i = O(e 2 ).
i v
(18)
419
01
e
1
lim
vi T:a T
(19)
and
Xi* =
0 1
01
e
1
lim
v*
T:a T
i
T
2
sin v*
i tfi (X*
j cos v*
j t, vj X*
j sin v*
j t) dt + O(e ),
or
Xi* =
e
1
lim
vi T:a T
(20)
(Note that equations (12) and (16)(18) have been used in the derivation.) The last two
equations clearly indicate that the formulations of the first order approximate differential
equations are actually identical for the amplitudes.
2.3.
For simplicity, suppose that the ratios of the frequencies, vi , are close to
p1 : p2 : p3 : : pn , where pi are positive integers. This implies that
v j pj
= O(e),
v i pi
for i, j = 1, 2, . . . , n.
(21)
(22)
>
v = s vj s pj ,
j=k
and
l $ (k, k + 1, . . . , n).
(23)
j=k
The v is a known constant. It will be demonstrated later that the choice of v which satisfies
equation (23) leads to the same first order results. (As a matter of fact, other forms of v
may be defined provided that (vj pj v), i = 1, 2, . . . , n, are order e.) By using equation
(23), equation (1) can be rewitten as
h i + (pi v)2hi = [(pi v)2 vi2 ]hi + efi (hj , h j ),
i, j = 1, 2, . . . , n.
(24)
To apply time-averaging, the term (pi v)2 vi2 appearing on the right of equation (24) must
be of order e. Indeed, it can be shown, by using equations (21) and (23), that
(pi v)2 vi2 = O(e). Furthermore, it can be proved that the difference of any two v defined
by equation (23) is also of order e.
. .
420
Based on equation (24), the periodic or quasi-periodic solutions which bifurcate from
an equilibrium point can be described by
hi (t) = Xi (t) cos [(pi v)t + fi (t)] 0 Xi (t) cos Ci (t),
i = 1, 2, . . . , n.
(25)
This form is similar to the KB solution (3a) because the frequencies pi v are known.
Equations similar to equations (24) and (25) can be written for the EG method as
h i + (pi v*)2hi = [(pi v*)2 vi2 ]hi + efi (hj , h j ),
i, j = 1, 2, . . . , n
(26)
and
hi (t) = X*
i (t) cos [(pi v*)t + f*
i (t)] 0 X*
i (t) cos C*
i (t),
i = 1, 2, . . . , n,
(27)
respectively. The GCD frequency, v*, is unknown and the (pi v*)2 vi2 , i = 1, 2, . . . , n,
are of order e.
The KB procedure, which was used to obtain equations (8) and (9) for the non-resonant
case, can be applied again to produce the time averaged, differential equations
Xi =
0 1g
e
2pi p
2p/v
(28)
and
e
(p v) vi
f i = i
2pi v
2pi pXi
2
1g
2p/v
(29)
As in the non-resonant case, the above two integrals can be approximated by assuming
that Xi and fi are constant in the period, 2p/v. However, any term appearing in the
Fourier series of the fi which contains cos Cj or sin Cj contributes a non-zero residual
because this situation corresponds to a resonant case. The extra terms (for j $ i) include
not only the amplitudes Xj but also the phase differences fi 2 mj fj 2 2 mk fk which
satisfy pi 2 mj pj 2 2 mk pk = 0. There are only (n 1) phase differences that are linearly
independent. Without loss of generality, let ci1 = p1 fi pi f1 , i = 2, 3, . . . , n; the other
phase differences can be expressed as linear combinations of ci1 . Consequently,
c i1 = p1 f i pi f 1 can be found from equation (29). Similarly, the time-averaged differential
equations for the EG approach can be written as
Xi* =
0 1g
e
2pi p
2p/v*
(p v*) vi
e
f i* = i
2pi v*
2pi pX*
i
2
sin C*
i fi (X*
j cos C*
j , pj v*X*
j sin C*
j ) dt,
(30)
1g
2p/v*
cos C*
(31)
i fi (X*
j cos C*
j , pj v*X*
j sin C*
j ) dt,
421
2p/v
i = 1, 2, . . . , n,
(32)
and
0 1g
e
p Xi
2
i
2p/v
p
v
0 1g
e
p X1
2
1
2p/v
$0 1 0 1 %
vi
pi
v1
p1
(i = 2, 3, . . . , n).
(33)
Note that (p/v)[(vi /pi )2 (v1 /p1 )2 ] is of e order due to equation (21). The frequency of
the ith component, Xi , is given by
v i = pi v + f i
or
v i =
1g
2pi pXi
2pi v
2p/v
(34)
Equations (32)(34) suggest that any v chosen from equation (23) leads to identical first
order solutions because the difference between any two such v is of order e. Furthermore,
note that
pi v j pj v i = pi (pj v + f j ) pj (pi v + f i ) = [pi (c j1 + pj f 1 ) pj (c i1 + pi f 1 )]/p1 = 0,
for i, j = 1, 2, . . . , n, so that v 1 : v 2 : : v n = p1 : p2 : : pn . The last ratios imply that
the solution is, indeed, a periodic motion, as expected. Therefore, a GCD frequency v can
be defined such that v i = pi v .
Next, consider the form of the solutions for the EG approach. Then the steady state
solutions can be obtained from equations (30) and (31) by setting Xi* = f i* = 0. This
results in 2n independent, non-linear algebraic equations for the 2n independent unknowns
X*
i , i = 1, 2, . . . , n, c*
i1 , i = 2, 3, . . . , n, and v*. Specifically, the 2n equations can be
written as
2p/v*
sin C*
i fi (X*
j cos C*
j , v*
j X*
j sin C*
j ) dt = 0
(35)
. .
422
and
2
v*
vi2
e
i
=
2v*
2pi pX*
i
i
1g
2p/v*
cos C*
i fi (X*
j cos C*
j , v*
j X*
j sin C*
j ) dt,
(36)
where v*
i* = 0.
i = pi v* is the frequency of the ith amplitude component, Xi , because f
It can be seen for the resonant cases that the two approaches do not lead to a too
different number of governing equations for the steady state solutions. (One produces
(2n 1) equations and the other gives 2n equations.) However, it will be shown next that
a particular choice between the two approaches may make it much easier to find explicit,
steady state solutions for certain internal resonances.
2.3.2. First order equivalence of the steady state, internally resonant solutions
To prove the equivalence of the two steady state solutions, first note that
Ci = pi vt + fi 0 v i t and pi v = vi + O(e). Rewrite equation (32) as
2p/v
2p/(1 + O(e))v
or
2p/v
so that
2p
(37)
Similarly, by using C*
i = pi v*t and v*
i 0 pi v* = vi + O(e), equations (35), (34) and (36)
can be rewritten as
2p
(38)
(39)
2
cos pi ufi (X*
j cos pj u, vj X*
j sin pj u) dt + O(e ),
(40)
v i = vi
1g
2p
1g
2p
e
2pvi Xi
and
v*
i = vi
e
2pvi X*
i
423
respectively. A comparison of equations (37) and (38) as well as equations (39) and (40)
leads to two sets of identical forms of first order equations. This implies that
X*
i Xi = O(e)
and
v*
i = O(e 2 )
i v
(41)
(42)
and the KB and EG methods indeed produce identical first order amplitude components,
phase differences and frequencies.
2.3.3. First order equivalence of the stability conditions
The stability conditions for the two resonant, steady state solutions can be determined
from the eigenvalues of a (2n 1) (2n 1) Jacobian matrix. This matrix is derived from
the differential equations of Xi and c i1 for the KB form (25) and from the differential
equations of Xi* and c i*1 for the EG form (27). A periodic solution is stable if all the real
parts of the non-zero eigenvalues of the Jacobian matrix are negative; otherwise, it is
unstable. By following the procedure demonstrating the equivalence of the stability
conditions for non-resonance, it can be shown that the two first-order (2n 1) (2n 1)
Jacobian matrices are identical. Hence, the first order stability conditions derived from the
KB and EG methods are identical.
2.4. -
When one or more natural frequencies correspond to internal resonances while the
remaining natural frequencies are non-resonant, the differential equation (1) can be
rearranged into two subsystems. One subsystem involves only resonant frequencies and
the other incorporates solely non-resonant frequencies. Then the procedures employed
previously for the non-resonant and resonant cases can be used again. However, the two
subsystems are coupled because of non-linear terms. (On the other hand, the governing,
time-averaged equations are coupled only by the amplitudes of motion.) The derivation
of the steady state solutions and proof of their first order equivalence are similar to those
given in subsections 2.2 and 2.3, so they are omitted. The procedure for handling
combinations of resonant and non-resonant cases will be given in the next section for a
3-DOF oscillator.
2.5.
Such a system can be handled by employing the procedure used in the previous
subsection. In addition to the two subsystems which correspond to internal non-resonance
and resonances, a third subsystem may be found. This external resonant subsystem consists
of the systems natural frequencies which are resonant with the external forcing frequency,
n. However, they cannot be included in the internal resonant subsystem. The solution
forms given for the internal resonances can be applied here, provided that the known
external frequency, n, is used instead of the GCD frequency, v or v*, and the external
resonant natural frequencies are expressed in terms of n. This implies that the KB solution
forms (3a) and (3b) are identical to the EG forms (4a) and (4b), respectively, because
v = v* = n. Then, the demonstration of the equivalence of the KB and EG methods for
this subsystem is merely the proof of the equivalence of forms (3a) and (3b). It will be
shown in the following subsection that these two forms produce identical approximate
. .
424
solutions up to any order. Therefore, by using the procedure described in subsections 2.2
and 2.3, the steady state solutions and their stability conditions can be obtained, and their
first order equivalence can be proved similarly for the whole system. Details are omitted
for brevity.
2.6. ,
2.6.1. The KB and EG methods
The relationship between the KB and EG methods is illustrated in Figure 1, in which
ODE, ABE, T.A. and E.Q. are abbreviations for ordinary differential equation, algebraic
equation, time-averaged and equivalence, respectively. Although the equivalence of the KB
and EG methods has been proved explicitly only for solution forms (3a) and (4a), it can
be deduced from Figure 1 that another three equivalences are also true: viz., equations
(3b) \ (4b), (3b) \ (4a) as well as (3a) \ (4b). This is because the relation between the
solution forms (3a) and (3b), or between the solution forms (4a) and (4b), is merely a
co-ordinate transformation, given by
Ai = Xi cos fi ,
Bi = Xi sin fi ,
Xi = zAi2 + Bi2 ,
(43a)
B*
i = X*
i sin f*
i ,
2
2
X*
+ B*
i
i ,
i = zA*
(43b)
425
T 1
A comparison of the EG and CG methods
Method
Frequency
Solution
form
Governing
equation
Stability
analysis
EG
Unknown
(4a) or (4b)
Differential
equation
Available
CG
~ Known
_Unknown
(3a) or (3b)
(4a) or (4b)
Algebraic
equation
Not
available
for equations (4a) and (4b). Therefore, the results obtained by using solution form (3a)
(or (4a)) are actually identical to those produced by employing solution form (3b) (or (4b)).
Note that the most frequently used equivalence is the last one; i.e., (3a) \ (4b).
2.6.2. The EG and CG methods
The EG method can be used for a stability analysis, unlike the CG method. However,
the solution forms of the CG method can have either a known or an unknown frequency,
while that of the EG method can have only an unknown frequency. A comparison of the
two methods is shown in Table 1. The proof of the equivalence of the two methods is given
next.
First, assume that an unknown frequency is given in the solution form of the CG
method. Consider an alternative set of equations for solving the steady state solutions from
the EG approach by using solution form (4a). They are found, by setting Xi* = f i* = 0
in equations (10) and (11), to be
1
T
T:a
1
T:a T
lim
T
2
sin (v*
vi2 )hi + efi (hi , h i )] dt = 0,
i
i t + f*
i )[(v*
i = 1, 2, . . . , n,
(44a)
2
cos (v*
vi2 )hi + efi (hi , h i )] dt = 0,
i
i t + f*
i )[(v*
i = 1, 2, . . . , n.
(44b)
and
lim
It should be noted that equations (44a) and (44b) apply not only to non-resonance but
also to internal resonance and combinations of the two cases. Multiplying equation (44a)
by sin f*
and equation (44b) by cos f*
i
i , and adding the resulting equations, leads to
1
T:a T
lim
T
2
sin v*
vi2 )hi + efi (hi , h i )] dt = 0,
i
i t[(v*
i = 1, 2, . . . , n,
(45a)
i = 1, 2, . . . , n,
(45b)
because the f*
i , i = 1, 2, . . . , n, are constants. Similarly,
1
T:a T
lim
2
cos v*
vi2 )hi + efi (hi , h i )] dt = 0,
i
i t[(v*
. .
426
can be found from equations (44a) and (44b). In fact, the relationships between equations
(44a) and (44b) to equations (45a) and (45b) can be represented by the simple
transformation matrix
cos f*
i
sin f*
i
sin f*
i
.
cos f*
i
(46)
Therefore, the steady state solutions derived from the EG and CG methods are identical
if an unknown frequency is assumed in the CGs solution form. The same relation (46)
can be obtained if solution form (4b) is used.
Next, suppose that a known frequency is used in the solution form of the CG method.
It is easy to see that this case is merely the KB method without the stability analysis, but
expressed in the forms (45a) and (45b), where the unknown frequency, v*
i , is now replaced
by the known frequency, vi . The same relationship given by equation (46) can be derived
if fi is substituted for f*
i . Thus, the steady state solutions derived from the CG method
with a known frequency are identical to those obtained by using the KB method.
Therefore, the first order equivalence of the KB and EG methods implies that the EG and
CG methods give equivalent first order steady state solutions if a known frequency is used
in the CG method. The relationship between the EG and CG methods is also shown in
Figure 1.
2.6.3. The KB and CG methods
From the above discussion of the relationship between the EG and CG methods, it can
be deduced directly that the KB and CG methods produce identical or equivalent first
order steady state solutions when a known or an unknown frequency is assumed in the
CG method. The relationship between the KB and CG methods is also given in Figure 1.
Also indicated in Figure 1 are the relationships derived by Newland [15], who
demonstrated the equivalence of the KB and CG methods between: (1) equations (3a) and
(4a) for the amplitudefrequency relation of the non-resonance of a 1-DOF system, and
(2) equations (3a) and (3b) for a 1 : 1 external resonance.
3. APPLICATIONS
The KB and EG methods will be employed next to analyze a 3-DOF, eccentric model
of galloping.
3.1. 3-
A continuous multi-span, iced transmission line can be reduced to the 3-DOF, discrete
oscillator model [9, 10] illustrated in Figure 2, where the overall centre of mass, G, deviates
427
from the centre of the bare conductor due to ice accretion. The motions can be described
by a dimensionless form
[M]{q } + [C]{q } + [K]{q} = {F},
(47)
where a dot superscript indicates a derivative with respect to time, t, vector {q} = (y z u),
in which y and z represent the dynamic displacements in the vertical and horizontal
directions, respectively, while u is the rotation about the centre of the bare line. [M], [C]
and [K] are the mass, damping the stiffness matrices, respectively. {F} is the non-linear
aerodynamic force vector, details of which can be found, together with the other matrices,
in reference [9]. It should be noted that, although equation (47) is derived for an electrical
transmission line, the results given in the following sections can be applied to other weakly
non-linear systems which are described by equation (47).
The first step in investigating the feasibility of galloping is to find whether the initial
equilibrium state (IES) of equation (47), given by {q} = 0, is stable. If the IES is unstable,
the next step is to find solutions bifurcating from the IES and to determine their stabilities.
The stability conditions for the IES can be found from the linearized equation of system
(47). For a dynamic analysis, on the other hand, equation (47) needs to be transformed
to canonical form (1). This transformation is accomplished by first finding the natural
frequencies of system (47), which may be obtained from the associated free vibration
system [M]{q } + [KL ]{q} = {0}, where the linear parts of the aerodynamic forces have
been included in [KL ] to incorporate the effect of wind. Having determined the wind-on
natural frequencies, vi , introduce the linear transformation {q} = [F]{h} into equation
(47), to obtain
T
h i + vi2 hi = Ri (hj , h j ),
i = 1, 2, 3,
(48)
where {h(t)}T = [h1 h2 h3 ] are the principal co-ordinates. [F] is a 3 3 matrix having
column vectors, {fi }, found from [[KL ] vi2 [M]]{fi } = {0}. The Ri are elements of vector
{R} which consist of only damping and aerodynamic forces that are small (e order)
compared to the vi . These forces are usually approximated by polynomials up to third
order so that closed form solutions can be obtained. Thus,
{R} = [P]{a a 2 a 3}T [j]{h },
(49)
a = d s (f3i hi + ri h i ).
(50)
i=1
Constant matrices [P] and [j] and constants ri can be found in reference [9].
The KB and EG methods described in the previous section will be used to evaluate
two representative internal resonant cases, i.e., 1:3:0 and 1:1:1, where 0 indicates a
non-resonant mode.
3.2. 1:3:0
This case describes not only the 1:3:0 but also the 3:1:0, 1:0:3, 3:0:1, 0:1:3 and 0:3:1
resonances. A 1:3:0 resonance implies that the ratio of the first two natural frequencies,
v1 and v2 , is close to 1/3. Specifically, the difference 3v1 v2 is of order e. Both the KB
and EG methods and the technique developed in references [9, 10] have been used to derive
explicit solutions for this resonance. It has been shown that the KB method finally
produces a second and a seventh degree polynomial (details will be given later), while the
EG approach ultimately yields a 13th degree polynomial. Thus, the solutions derived by
. .
428
utilizing the EG approach are more sensitive to small variations in the parameter values,
so that the EG approach is not recommended for this case. Indeed it has been found for
some parameter values that although the KB method produces accurate approximations,
compared to numerically integrated results, the EG approach is intractable. Therefore,
only the KB procedure is detailed here. Apply equation (23) to let
v = 14 (v1 + v2 ),
d1 = v 2 v12 ,
d2 = (3v)2 v22 ,
(51)
h 2 + (3v)2h2 = R2 + d2 h2 ,
h 3 + v32 h3 = R3 .
(52)
By combining forms (3a) and (25), the first order approximation of the motion can be
written as
h1 = X1 (t) cos [vt + f1 (t)] 0 X1 (t) cos C1 (t),
h2 = X2 (t) cos [3vt + f2 (t)] 0 X2 (t) cos C2 (t),
h3 = X3 (t) cos [v3 t + f3 (t)] 0 X3 (t) cos C3 (t),
(53)
where the Xi and fi , respectively, represent the ith amplitude and phase. Next, by following
the procedure described in subsections 2.2 and 2.3, equation (53) can be applied to
equation (52) to produce the governing bifurcation equations
2X1 = X1 [j11 r1 (p11 + p13 W1)] + p13 (C1 cos c C2 sin c)X12 X2 ,
2X2 = X2 [j22 r2 (p21 + p23 W2)] + 19 p23 (D1 cos c D2 sin c)X13 ,
1
2c = d + [3f31 (p11 + p13 W1 ) 13 f32 (p21 + p23 W2 )]
v
01
0 1%
X2
X1
+ 13 p23 (D2 cos c + D1 sin c)
X1
X2
C 3 = v3
and
X12 ,
f33
(p + p33 W3 ),
2v3 31
(54)
where
v = 14 (v1 + v2 ),
c = f2 3f1 ,
d=
1
[3(2v 2 + v12 ) v22 ],
3v
1
[f (f 2 r12 v 2 ) 6r1 r2 f31 v 2 ],
v 32 31
2
r12 v 2 ) 2r1 f31 f32 ,
C1 = 3r2 (f31
C2 =
2
r12 v 2 ),
D1 = r1 (3f31
D2 =
f31 2
(f31 3r12 v 2 ).
v
(55)
c is the phase difference between X2 and X1 , which are the amplitude components of the
periodic motion having frequency v. On the other hand, W1 and W2 are described by
Wi = W (f3i2 + ri2 v 2 )Xi2
(i = 1, 3),
W2 = W (f + 9r v )X ,
2
32
2
2
2
2
2
+ 9r22 v 2 )X22 + s (f3j2 + rj2 vj2 )Xj2 .
W = 2 (f32
j = 1,3
(56)
429
The steady states which bifurcate from the IES can be found by setting Xi = c = 0,
i = 1, 2, 3 in equation (54). Note, however, that the equation describing X3 is decoupled
from the first three equations so that only X1 , X2 and c need be solved simultaneously from
the three equations given by X1 = X2 = c = 0. The resulting three, non-linear algebraic
equations are coupled so that no general method exists to find analytical expressions and,
hence, numerical iterations are usually employed. A specific analytical procedure has been
proposed in references [9, 10], on the other hand, to find explicit solutions in order to
save substantial computational time. This saving is achieved by first deducing a polynomial
the coefficients of which are expressed explicitly in terms of known system parameters.
Then the Xi and c can be found from analytical expressions involving the roots of the
polynomial.
The particular polynomial can be found by eliminating two of the three variables, X1 ,
X2 and c, from the three equations obtained by setting X1 = X2 = c = 0. This is achieved
by first employing the first two equations (given by X1 = X2 = 0) to express cos c and sin c
as
cos c =
01
01
A
B
2
2
(f31
3r12 v 2 )f31 X12 + 3
[f31
f32 (r1 f32 4r2 f31 )r1 v 2 ]X22 ,
D
D
01
sin c = v
01
A
B
2
(3f31
r12 v 2 )r1 X12 + 2v
[f31 (r1 f32 r2 f31 ) + r12 r2 v 2 ]X22 ,
D
D
(57)
where
A=
j11 + r1 v[(1 4m)Y + 2(1 2m)Z],
B=
j22 + 2r2 v[2(1 2m)Y + (1 4m)Z],
(58)
j11 and
j22 are detailed in Table 2. m = 0 for periodic solutions and 1 for quasi-periodic
solutions. Consequently, the 1:3:0 resonance may have a family of periodic solutions
consisting of Xi $ 0, i = 1, 2, X3 = 0, and quasi-periodic motions lying on a two
dimensional torus in which Xi $ 0, i = 1, 2, 3.
Next, use sin2 c + cos2 c = 1 and the third equation, given by c = 0, to yield the second
degree polynomial
g3(Z) g6 (Z)Z 0 g3 g6 Z = 0
(59)
(60)
where
2
Y = (f31
+ r12 v 2 )X12 ,
2
Z = (f32
+ 9r22 v 2 )X22 .
(61)
. .
430
T 2
The coefficients, Ai and Bi , used for the 1 : 3 : 0 resonance
A0 = m1 r1 s2 (m1r1 + m2 pr r2 )v 2
A1 = r1 s2 (2m1 m2 r1 + m3 pr r2 )v 2 s42
A2 = m2 r1 s2 (m2 r1 + m1 pr r2 )v 2
A4 = v[m1 r1 (2j
11 + pr
j22 /3) + m2 r2 pr
j11 ]s2
A6 = 3v[m2 r1
j22 + m1 r2 (3j
11 + 2pr
j22 )]s3
+3v[m1 r1
j22 + m2 r2 (3j
11 + 2pr
j22 )]s3
[4d
m1 r2 v 2/p13 + (3m2 f31 m1 pr f32 )j
22 ]
22 ]
+[4d
m2 r2 v /p13 + (3m1 f31 m2 pr f32 )j
2
A7 =
j11 (j
11 + pr
j22 /3)s2
A8 =
j22 (3j
11 + pr
j22 )s3 4d
vj
22 s4 /3p13
B0 = m1 r1 p23 s2 /12
B1 = m2 r1 p23 s2 /9
B2 = 3m2 r2 p13 s1
B3 = 27m1 r2 p13 s1
B4 = p23
j11 s2 /9v
j11 p23
j22 /3)s3 /v +
ds4
B5 = (3p13
B6 = 9p13
j22 s1
m2 = 2(1 2m)
2
s1 = f31
+ r12 v 2
s2 = f + 9r v
2
32
2
2
m3 = m12 + m22
ti = ri pi3 ,
i = 1, 2
t3 = f31 p13 /v
p4 = p23 /3p13
jii = v[jii ri (pi1 + 2mpi3
j33 )]/p13 ,
i = 1, 2
j33 = (p31 j33 /r3 )/p33
d = d + (3p11 f31 p21 f32 /3)/v
+2m(3p13 f31 p23 f32 /3)j
33 /v
1
(A Z + A4 ),
A0 1
gi (Z) =
1
(A Z 2 + Ai + 3 Z + Ai + 5 ),
A0 i
gi (Z) =
1
(B Z + Bi ),
B0 i 3
i = 4, 5, 6.
i = 2, 3,
(62)
431
Having found Z from equations (59) and (60) (by using, for example, the IMSL program
RPOLY [16]), Y is determined from
Y=
(63)
1
m
(j r3 p31 ) 2(Y + Z) .
2
+ r32 v32 r3 p33 33
f33
X32 =
(64)
The first order approximation of the periodic and quasi-periodic solutions is described by
hi = Xi cos v i t,
i = 1, 2, 3,
(65)
where X3 is given by equation (64). The frequency, v 3 , for the quasi-periodic solution is
given by the fifth equation of (54) as
v 3 = v3
f33
(p + p33 W3 ),
2v3 31
(66)
whereas
3v 1 = v 2 = 3[v d1 f31 (p11 + p13 W1 ) + p13 (C2 cos c + C1 sin c)X1 X2 ]
(67)
i = 1, 2, 3,
(68)
where v* is the unknown frequency of the assumed periodic motion. Next, the first order
approximation is assumed to take the form
hi = Ai (t) cos v*t + Bi (t) sin v*t,
i = 1, 2, 3,
(69)
Ai = lim
1
T:a T
Bi = lim
1
[R + (v*2 vi2 )hi ] sin v*t dt,
v* i
1
[R +(v*2 vi2 )hi ] cos v*t dt,
v* i
i = 1, 2, 3.
(70)
. .
432
With the aid of equations (4)(6), as well as equations (68) and (70), the following six
governing equations can be found:
3
i = 1, 2, 3,
(71)
j=1
where
3
E = s (f3j Aj + v*rj Bj ),
F = s (f3j Bj v*rj Aj ).
j=1
(72)
j=1
Periodic solutions can be derived from the non-linear equations obtained by setting
Ai = Bi = 0, i = 1, 2, 3, in equations (71). The resulting six equations need to be solved
simultaneously to find the periodic solutions. Frequency v* is an appropriate candidate
for the variable of the polynomial which most reduces computational effort. After
algebraic manipulation, a fifth degree polynomial results, which takes the general form:
[r1 v*2 (j11 f31 + r1 v12 )](J 2 + K 2 ) j12 f32 (L 2 + G 2 ) j13 f33 (M 2 + N 2 )
+[r2 v*2 (j11 f32 + j12 f31 + r2 v12 )](LJ + GK) + [r3 v*2
(j11 f33 + j13 f31 + r3 v12 )](NJ + MK) [v*2(f33 + r1 j13 r3 j11 )
f33 v12 ]
(GJ LK)
(MJ NK)
[v*2(f32 + r1 j12 r2 j11 ) f32 v12 ]
v*
v*
1
(r j r3 j12 )(GN LM) = 0,
v* 2 13
(73)
(J 2 + K 2 )2 2
(E + F 2 ),
P2 + Q2
A22 + B22 =
L2 + G2
(A12 + B12 ),
J2 + K2
M2 + N2
(A12 + B12 ),
J2 + K2
(74)
p11 2v*
+
[j (J 2 + K 2 ) j12 (GK + LJ) j13 (MK + NJ)],
p13 p13 Q 11
(75)
A32 + B32 =
where E 2 + F 2, P and Q are given by
E2 + F2 =
(76)
433
T 3
Polynomial functions used for the 1 :1 : 1 resonance
Polynomial
Degree
in v*
Expression
G(v*)
G1 J2 G3 J3 K1 L2 + K3 L1
K(v*)
G2 J2 G3 J2 K2 L3 + K3 L2
M(v*)
G2 J1 G1 J2 K2 L1 + K1 L2
L(v*)
G1 L3 G3 L1 K3 J1 + K1 J3
J(v*)
G2 L3 G3 L2 + K2 J3 K3 J2
N(v*)
G2 L1 G1 L2 + K2 J2 K1 J1
i = 1, 2, 3
Ji (v*),
i = 1, 2, 3
i = 1, 2
Ki (v*)
Li (v*),
K3
i = 1, 3
(1)
(i + 1)/2
L2
and
Q = v*[r1 (J 2 + K 2 ) r2 (LJ + GK) r3 (NJ + MK)] + f32 (GJ LK) + f33 (MJ NK),
(77)
respectively.
The first order approximation of the periodic solution can be obtained by assuming
B1 = 0 because the system is autonomous. Therefore,
h1 = A1 cos v*t, h2 = A2 cos v*t + B2 sin v*t,
(78)
where
A1 = (J 2 + K 2 )
A2 = A1
A3 = A1
E2 + F2
P2 + Q2
1/2
B2 = A1
GJ LH
,
J2 + H2
B3 = A1
MJ NH
.
J2 + H2
GH + LJ
,
J2 + H2
MH + NJ
,
J2 + H2
(79)
3.4.
Both periodic and quasi-periodic solutions are feasible for the 1 : 3 : 0 resonance, but
only a periodic solution is possible for the 1:1:1 resonance. Stability conditions can be
determined straightforwardly from the Jacobian matrices of the time-averaged equations
(54) and (71) for the 1:3:0 and 1:1:1 cases, respectively. A periodic or quasi-periodic
434
. .
solution is stable if all the real parts of the non-zero eigenvalues of the Jacobian matrix
are negative. It should be noted, however, that the 6 6 Jacobian matrix of the 1:1:1
resonance has one zero eigenvalue [17], because it is evaluated on the periodic solution
which actually has only five independent variables.
3.5.
3.5.1. Example 1: bluff angle section
The angle section, which is shown in Figure 3(a), is actually a 2 DOF, co-centric system.
This particular example was studied analytically by Blevins and Iwan [12] for the
non-resonant case as well as for the 1:1 and 1:3 resonances. However, explicit solutions
and stability conditions were given only for the non-resonance. Dynamic responses in the
resonances were obtained numerically from the time-averaged equations. Later, the 1:1
resonance was re-examined and explicit solutions were derived by using the EG method
[8]. Recently, a typical angle section model was experimentally considered in detail by
Modi et al. [18]. This model had 1 1 in (254 254 cm) and 3 3 in (762 762 cm)
cross-sectional dimensions with uniform leg thicknesses of 1/16 in (16 cm) and 1/2 in
(127 cm), respectively. The aerodynamic lift, drag and moment coefficients of the model
were measured from wind tunnel tests and dynamic responses of the model were also
obtained, especially for the case of vu /vy = 292 (close to 1:3 resonance). However, unlike
reference [12] and this paper, in which we are particularly interested in the dynamic
behavior of the model with respect to the variation of the frequency ratio (vu /vy ) near
the 1:3 resonance, Modi et al. [18] investigated the dynamics of the model with respect
to the wind speed (U) at a fixed frequency ratio (292). Moreover, the results presented
in reference [18] were obtained from a one-degree-of-freedom (either plunge or torsion)
model under appropriate wind speeds. Therefore, the results are actually obtained for
non-resonant cases, and the word resonance used in reference [18] (e.g., plunging
resonance or torsional resonance) means that the corresponding maximum plunge
(torsion) response is with respect to a particular wind speed, rather than the usual
definition [810, 12] indicating the relationship between simultaneous plunge and
torsional motions. Although the main purpose of this paper is to compare the two
435
approximation methods in resonant cases, and thus only the results regarding the variation
of vu /vy are presented, the behavior of the model under different wind speeds is very
important and should be studied. In fact, an interesting phenomenon can be observed from
the results presented in reference [18]: galloping is only associated with certain values of
wind speeds in resonant cases. This observation has indeed been revealed analytically in
our early results [9, 10].
The form of the explicit solution for the 1:3:0 resonance given in the previous
subsections can also be applied for the 1:3 situation. The structural and aerodynamic
properties of the bluff angle section are well documented in reference [12], and they are
used to yield
[M] = diag [1
1],
[K] = diag [1
000513v],
1 v 2 ],
v = vu /vy ,
F006024(0656a + 783a3 )J
h.
{F} = g00
f045550(0105a + 934a3 )j
(80)
(81)
It should be noted that the horizontal component of the aerodynamic forces is zero,
which results, in turn, in a zero horizontal response so that the system really has 2 DOF.
The normalized wind speed Uy = Uz /vy d, on the other hand, equals 45.
The IES, {q} = {0}, was found to be stable for the specified parameters over the range
of structural frequency ratios, 11114 Q v Q 12815. A dynamic analysis was performed
for the initially unstable regions. A stable periodic solution was obtained by using the
formulae derived for the 1:3:0 resonance over the range 28 Q v Q 32. The resulting
vertical, Ay , and torsional, Au , amplitudes of the motion and that obtained by using
numerical integration are shown in Figure 4, where the results given by Blevins and Iwan
[12] are also indicated. It can be seen from this figure that the vertical amplitude is quite
smooth near v = 3 but the torsional amplitude has a peak at v = 3. Also, it can be noted
that, for the vertical motion, the present results and those of Blevins agree with the
numerically integrated data. For the torsional motion, however, our results are in excellent
agreement with the numerically integrated solution for a large interval of vu /vy
(30 2 02), but the results of Blevins cover a very narrow interval (approximately for
vu /vy = 30 2 0015, while the results outside this region are given by the non-resonant
analysis).
3.5.2. Example 3: D-section
A D-section transmission line, which is depicted in Figure 3(b), has been analyzed
previously in order to consider a 1:1:1 galloping motion having strong coupling between
the vertical, horizontal and torsional components [9, 10]. Although a detailed study has
been published, the dynamic responses were limited to fixed parameter values, although
the steady wind speed was changed. The main interest here, however, lies in the dynamic
behavior near resonance, so that this example will be reassessed.
The dimensionless data are given by
[M] =
&
100000
000000 000520
000000
103000 001780
000520 001780
800000
'
. .
436
&
'
000150v1
054400v2 ],
v1 = vz /vy , v2 = vu /vy ,
(82)
and
&
'
(83)
Figure 4. The dimensionless response for the 1 : 3 : 0 resonance; Uy = 45. e, Blevins numerical integration
[12]; q, our numerical integration; - - - -, Blevins asymptotic solution [12]; , our asymptotic solution.
437
Figure 5. The dimensionless response for the 1 : 1 : 1 resonance; Uy = 258. (a) vertical amplitude; (b) horizontal
amplitude; (c) torsional amplitude. The left subfigures are obtained from the analytical formulation; the right
subfigures represent the numerical results.
reference numerically integrated data, shown in the right column of this figure. However,
the computation of each integrated point consumed hours of CPU time on a SUN work
station.
4. CONCLUSIONS
Two seemingly different, time averaging procedures are shown analytically to produce
identical first order, approximate solutions and stability conditions for an n-DOF
non-linear, discrete system. It is suggested that, depending upon the non-resonant or
resonant case under consideration, the application of one or the other procedure can be
more advantageous. The two procedures are used to analyze galloping for which periodic
and quasi-periodic solutions are derived explicitly.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial support of the Canadian Electrical
Association (CEA Project No. 321 T 672) through the auspices of Manitoba Hydro. Partial
438
. .
support was also received from the Natural Sciences and Engineering Research Council
of Canada.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.