Lagrange Housner Hudson
Lagrange Housner Hudson
Lagrange Housner Hudson
The quantities ax, 6y and 62 are small displacements, and they have
been written as deltas to emphasize that they may have arbitrary
values, consistent with the physical restraints, and hence are not
differentials.* The quantities ax, 6y and 62 are virtual displacements,
in the sense that this term is used in statics, and Equation (9.2)
is equivalent to Equation (7.30) previously obtained by a combination
of D'Alembertls Principle and the Principle of Virtual Displacements.
We now transform the coordinates in Equation (9.2) from the
(x, y, z) coordinates to the (ql, 92, q3) system, using the general trans-
formation Equations (9.1) above. From Equation (9.1) the dis-
placements 6x, 6y, and 62 can be written in terms of the q's as:
(9.4)
The left side of Equation (9.4) has a simple physical meaning:
it is the work done by the external forces of the system during the
displacement 641. We shall equate this work to the expression
(Q18ql) and thus define the generalized force Q1 as:
a3 ax
so that: - - -; etc.
agl aql
LAGRANGE'S EQUATIONS FOR A PARTICLE 297
We may also put:
Substituting this into the right hand side of Equation (9.4)along with
similar expressions for y and z , we obtain:
from which :
To find the generalized force Q,, imagine that the spring is stretched
a positive amount x, and consider the work done by all of the forces
of the system during an additional displacement 6x. The gravity
force on the pendulum, mg, does no work, and the spring force does
the work ( - kx)6x. Equating the total work to Q,Sx we obtain:
Qh= - -a v- - - -
a [$kx2 + mgl(1 - cos $)I = - rngl sin 4
a+ a+
Now, using Lagrange's equations (Equations (9.8) or (9.10)) we
obtain, for the x-coordinate :
d aT d
&) = a (mx + mZQ cos +) = mx + mZ d; cos 4 - mZq9 sin 4
PROBLEMS
9.1. By means of Lagrange's equations, derive the differential equations
of motion for a particle in spherical coordinates r , 0, 4.
9.2 A spherical pendulum consists of a particle of mass m supported by
a massless string of length 1. Using the angles 0 and 4 as generalized
coordinates, derive the two differential equations of motion by means of
Lagyange's equations. Show that these equations reduce to previously
known results when 0 and 4 are successively held constant.
9.3. .A particle moves in a plane and is attracted toward the origin of a
coordinate system by a force which is inversely proportional to the square
of the distance from the origin. Find, using Lagrange's equations, the
differential equations of motion of the system in plane polar coordinates.
LAGRANGE'S EQUATIONS FOR A SYSTEM OF PARTICLES 303
9.4. A particle of mass m rests on a smooth horizontal table whose
surface lies in the x, y plane. The particle is connected by a massless
string of length 1 to a point which moves along the x-axis according to the
law x = f (t),starting from a position in which the string is parallel to the
y-axis as shown in the figure. (a) Using the coordinates x, 4 as shown, de-
rive by the use of Lagange's equations the differential equations of motion
of the particle. (b) If x = f (t) = vt, solve the equations of part (a),
determining the motion of the particle and the tension in the string.
Note that since we are measuring the coordinates from the position
of static equilibrium there are initial forces mg in the springs, which
lead to terms in the potential energy expressions for the springs which
cancel the potential energy change of the mass due to the gravity
force (See Example 1, Section 3.6.) The kinetic energy of the
system is:
T = 9142 + imx2
306 ADVANCED METHODS IN DYNAMICS
Substituting these expressions directly into Lagrange's equations,
we obtain:
mx + k2x = kzr4
I4 + +
( k ~ k2)r24 = kzrx
A method of finding the solution of such simultaneous differential
equations will be discussed in the next section.
PROBLEMS
9.6. A rigid body oscillates about a horizontal axis as a compound
pendulum. The moment of inertia of the body about its axis of rotation
is I, and the distance from the axis to the center of mass of the body is a.
Derive the differential equation of motion of the system by Lagrange's
equations, and find the period of small oscillations.
9.7. A double pendulum consists of two equal masses and two strings
of equal length and of negligible mass. Using as coordinates the angles
between the strings and the vertical, as shown in the figure, find the
differential equations of motion for the system, by means of Lagrange's
equations. Show also the simplified form assumed by these equations for
small oscillations.
9.8. A rigid straight uniform bar of length 1 and mass m is pinned at one
end and is supported at distance a from the pinned end by a spring having
a spring constant k . Find by the use of Lagrange's equations the dif-
ferential equation of motion describing small oscillations of the bar about
the position of static equilibrium, and find the frequency of the motion.
LAGRANGE'S EQUATIONS FOR A SYSTEM O F PARTICLES 307
9.9. A solid homogeneous disk of radius R and mass m is constrained to
roll, without slipping, in its own plane. Two massless springs, each of
spring constant k, are attached to the disk in such a way that the position
of static equilibrium, with the springs in an unstretched position, is as
shown in the figure. Derive the equations of motion of the system, using
Lagrange's equations, and find the natural period of the motion.
9.10. A rope of negligible mass passes over a fixed pulley of moment of
inertia I l , mass ml, and radius rl, and supports a movable pulley of moment
of inertia I z , mass mz, and radius Y Z as shown in the diagram. A con-
centrated mass is attached to one end of the
rope, and the other end is fixed, with the rope
sections vertical as shown. Assuming no slip-
ping of the rope on the pulleys, find by means
of Lagrange's equations the differential equation
describing the motion of the mass ms, as the
system moves under the action of gravity.
9.11. The string of a simple pendulum is
assumed to be elastic with a spring constant k,
as indicated in the diagram. Taking as general-
ized coordinates of the mass a displacement x in
the direction of the spring, and the angle 4
between the spring and the vertical, find the
differential equations describing small oscilla- 1
tions of the system.
9.12. A block of mass M has a cylindrical PROB.9.11
groove of radius R , as shown in the diagram. A
M
small cylinder of radius r and mass - rolls without slipping in the groove
2
under the action of gravity. The contact between M and the supporting
3 08 ADVANCED METHODS I N DYNAMICS
horizontal surface is frictionless. Find, using Lagrange's equations, the
two differential equations of motion of the system.
xe = A2 sin wt
k
Substituting wlZ = -, we obtain:
m
hence, the two masses move in phase with equal amplitudes. Sub-
k
stituting wz2 = 3 - we obtain:
m
and the two masses move with equal amplitudes but in opposite
directions, that is, with a 180" phase difference.
The physical significance of these two motions can easily be seen.
If A1 = A2 the two masses move together with equal amplitudes
and the center spring connecting them is neither extended nor
compressed; hence it cannot affect the motion. The frequency of
vibration should thus be the same as for one of the masses restrained
k
by only one of the springs, that is, w2 =-.m For the motion in which
the two masses always move in opposite directions, we note that the
312 ADVANCED METHODS I N DYNAMICS
displacements are symmetrical and that the center point of the center
spring can be considered as fixed. We can thus represent the
system as shown in Fig. 9.6, which is a single degree of freedom
system having a spring constant of k + 2k = 3k and hence a fre-
quency
The above analysis started with the assumption that the solution
was of the form x = A sin wt. We could just as well have started
with a solution of the form x = A cos wt, in which case the same
algebraic equations for the A's would have been obtained. Since
both sine and cosine terms will satisfy the original equations, and
since superposition of solutions will be valid for the linear differential
equations involved, the general solution can be made up of a sum of
sines and cosines as:
where the subscripts on the A's and B's refer to the coordinates, and
the superscripts refer to the frequencies. We have already seen that
the A's and hence the B's, which must satisfy the same equations
as the A's, are not arbitrary, but must satisfy the conditions:
PRINCIPAL MODES OF VIBRATION 313
so the general solution becomes:
and each coordinate involves only one frequency and one mode of
vibration. Coordinates which have this property are called normal
coordinates, and the modes of vibration corresponding to them are
called the principal modes, or the normal modes of the system.
Once the normal coordinates of a system have been determined,
the problem has been essentially reduced to a study of a set of inde-
pendent single degree of freedom systems. This can be illustrated
314 ADVANCED METHODS I N DYNAMICS
by the system of Fig. 9.5 with a force F = FOsin ot applied to the
left mass, as shown in Fig. 9.7. Since we have already seen that the
coordinates ( I , ( 2 describe the principal modes of the system, we shall
use these normal coordinates as the generalized coordinates for the
forced oscillation problem. Substituting xl = + 6 2 ; x 2 = 6 1 - 6 2
v = k(Ei2 + 3 ( z 2 )
To find the generalized force Q associated with the applied external
force F, we note that the work done by F during the displacement
6x1 = 651 + 8 6 2 is:
Note that some of the forces acting on the masses have been included
in the potential energy, while some have been left as generalized
forces. This is often a convenient procedure in problems in which
the forces fall naturally into two different categories as in the above
example. I t is necessary, however, to be careful that all of the forces
are included, and that no forces are included twice.
PRINCIPAL MODES O F VIBRATION 315
Substituting the expressions for T, V and Q into Lagrange's
equations, we obtain :
Fo sin
m%z + 3k& = mt
PROBLEMS
9.16. Derive the differential equations of motion for the system of
Fig. 9.5 by a direct application of Newton's second law in the form F = ma.
9.17. Find the natural frequencies of vibration of the two-mass system
shown in the figure.
9.18. Two simple pendulums having equal lengths I and equal con-
centrated masses m are connected by a spring of spring constant k as
shown in the figure. Find the frequencies of vibration for small oscil-
lations, and the principal modes of vibration of the system. What initial
conditions would be necessary to obtain free oscillations of the system in
the first mode without exciting the second mode, and vice versa?
9.19. For the purpose of studying
torsional oscillations, a particular
engine-pump system is idealized as two
equal disks of mass m and moment of
inertia I about the geometric axis,
- equally spaced on a shaft one end of
which is built in. The torsional spring
constant kt lb ftlrad of the two portions
of the shaft are equal. Find the natural
PROB. 9.19 frequencies and mode shapes of tor-
sional oscillations of the system.
9.20. Find the natural frequencies of small oscillations of the double
pendulum of Prob. 9.7, and determine the shapes of the corresponding
modes of vibration.
9.21. In the expressions given above for the general solution x l and x2
for the two degree of freedom system, determine the four constants A l c l ) ,
A I ( ~ B) ,l ( l ) and B1(2)in terms of the initial displacements x l o , x2o and the
initial velocities xlO, i 2 0 .
9.22. (a) Starting with the expressions given above for the general
solution x l and x2 for the two degree of freedom system in terms of the
PRINCIPAL MODES O F VIBRATION 317
four constants A 1 ( l ) , A I W , B 1 ( l ) and B I ( ~show
) that if when t = 0,
xl = x2 = 0 and i1= xlo, x2 = x20, the displacement xl is given by:
xl = Cl sin wlt + C2 sin w2t
(b) Using the result of part (a), show that if wl is nearly the same as
w2 SO that (wl - w2) = Aw, X I varies between the values (C1 Cg) and+
(C1 - C2) with a frequency Aw, as shown in the figure. This modulation
.I
a2v
Since -= - the term involving the mixed derivative can
aqi 841 841 84;
be written :
All the terms in the expression for V will thus have the same form,
and a summation convention can be used as follows:
The derivatives in this expression, since they are all evaluated a t the
origin, are constants, which we may write in the form:
The terms k2j may be called the generalized spring constants for
the system. In some simple systems these k's will be the same as
the familiar spring constants of a spring element, but more generally
they are combinations of such spring constants, and may relate to
torsional restoring forces, gravitational restoring forces, etc.
9.8 The Kinetic Energy Function. The total kinetic energy of
a system can be expressed in terms of the orthogonal coordinates
( x i , yt, zt) of each particle as
322 ADVANCED METHODS I N DYNAMICS
We shall now express T in terms of the generalized coordinates. We
have
xr = + 1 ( q 1 , 9 2 , 9,)
Yt = $2(41, 9 2 , ' qn)
zt = +3(ql, 921 ' ' ' qn)
and thus:
T = 12 x m & j j
i=l j=l
where, for example:
We shall now make use of one of our basic assumptions that the
velocities (dl, q2, . . . qn) remain small. If any of the terms in the
above series expansion for mtj beyond the constant term ( m t j ) ~
were to be retained, the expression for T would have third order
terms in it, since the mtj is multiplied by (jagj). We thus see that
for small oscillation problems the mi, terms are constant, and we
have finally for the kinetic energy expression :
n n
The constants mif are called the inertia constants of the system.
For simple systems they may be simple masses or moments of inertia,
but more commonly they are combinations of such quantities
depending on the coordinate system used.
It will be noted that both the potential energy and the kinetic
energy are similar quadratic forms. This fact has made it possible
to apply directly to small oscillation problems many of the mathe-
matical techniques which have been developed for the study of
quadratic forms.
9.9 The General Equations of Free Oscillations. Since the
kinetic energy expression, Equation (9.16), does not involve the q's
aT
directly, but only the q's, the term - in Lagrange's equations will
a~,
vanish, and we have:
3 24 ADVANCED METHODS I N DYNAMICS
The terms involving the ktj cross-products are referred to as the
static coupling terms, and those containing the mrj cross-products as
dynamic coupling terms. Note that the word "coupling" refers to
the coordinates and not to the systems, so that the kinds of coupling
present depend upon the coordinates chosen rather than upon the
characteristics of the system itself. I t can be shown in general that
it is always possible to choose the generalized coordinates in such a
way that all the static coupling terms, or all the dynamic coupling
terms, or all the static and dynamic coupling terms together, will be
zero.
We shall suppose, in the following sections, that the coordinates
have been selected in such a way that all of the dynamic coupling
terms are zero. This simplification is for convenience only, and does
not represent any limitation of the basic theory.
The expression for the kinetic and potential energies for this
statically coupled system are:
where we no longer need a double subscript for the mtj terms, since
there are no cross-product terms. Substituting these expressions
for T and V into Lagrange's equations, Equation (9.17) above, we
obtain :
mlql +kllql + k12q2 + ' ' i- k2nqn = 0
+
m2q2 h l q l + A2242 + . . + k2nqn = 0
A similar expression involving a cos wrt instead of the sin w,t could
be added if necessary to satisfy the initial conditions of a particular
problem.
9.10 Orthogonality of the Principal Modes. In the present
section a basic property of the principal modes of vibration will be
developed which will be of considerable use for the calculation of
natural frequencies and mode shapes, as well as for further extensions
of the theory.
Consider two particular principal oscillations having different fre-
quencies w , and w,. From Equations (9.20) we may write:
In the left side of the second equation the subscript i may be written
as j and the subscript j may be written as i without affecting the
NATURAL FREQUENCIES AND MODE SHAPES 327
value of the term, since krj = kjc and in each case we are only
summing from 1 to n on the subscripts. We thus see that the left
sides of the two equations are identical and hence if the second is
subtracted from the first there results:
or, putting
L Equilibrium position
First mode
Second mode
Third mode
So :
similarly
PROBLEMS
9.30. In the expression for the kinetic energy T, Equation (9.16), write
the general term mtj in the form of a summation, and show that mtj = mjc.
9.31. Refer to Example 2 , Section 9.2 above, and show that for small
oscillations the equations for kinetic and potential energy derived there
in terms of the coordinates x and 4 reduce to the form of Equations (9.15)
and (9.16). Find the mcj's and the ktj1s for this particular example.
What kind of coupling exists between the (x, 4) coordinates ?
9.32. Referring to the two masses on a stretched string of Prob. 9.24
above, write the kinetic and potential energies in terms of suitable
coordinates, and determine the ktj's and the mij's for the problem. Cal-
culate the natural frequencies of vibration of the system by a direct sub-
stitution of the k's and m's into the frequency determinant Equation (9.21).
9.33. The figure shows three different coordinate systems suitable for
the description of small oscillations of a double pendulum consisting of
two equal masses and two massless strings of equal length. Write expres-
sions for the kinetic and potential energy of the system in each of the three
coordinate systems, and state whether static coupling or dynamic coupling
is present in each.
9.34. Three equal simple pendulums of length 1 and concentrated mass
m are coupled by two equal springs of spring constant k located a distance
a from the point of support as shown in the figure. (a) Using as co-
ordinates the three angles 41, c&, 4 3 and considering small oscillations only,
write the expressions for the potential and kinetic energy of the system,
NATURAL FREQUENCIES AND MODE SHAPES 331
and find the values of the kills and the mu's for this problem. (b) Find
the natural frequencies of vibration of the system. (c) Find the mode
shapes for the system.
9.35. An approximate dynamic model of a three story steel-frame
building can be formed from three concentrated masses and three massless
springs as shown in the figure. During lateral vibrations of the building,
as for example, during an earthquake, it is assumed that the floors move
parallel to each other, so that the action is primarily that of shear. The
masses of the three floors and the shearing spring constants are as shown
in the figure. Show that the frequency equation for small lateral
oscillations is :
mw2
2f3- 13f2+ 20f- 6 = 0 ; f = -
k
332 ADVANCED METHODS IN DYNAMICS
9.12 Forced Oscillations. We shall now suppose that the
statically coupled conservative system which has been considered
above is acted upon by a system of periodic exciting forces Fz sin wt
so that the basic equations of motion become :
$(%) + = F , sin wt
Following the same procedure that was successful for the single
degree of freedom system of Chapter 5, we shall assume that the
steady state solution will be harmonic and of the same frequency as
the exciting force, In what follows we shall consider steady state
motion only. We suppose that there is a small amount of damping
present which after a time will eliminate the transient terms, but
which is not large enough to change appreciably the steady state
forced amplitudes.
It cannot be expected that the solution will involve only a single
mode of vibration, since all of the modes may be excited simul-
taneously. We accordingly write a trial solution in the form :
Introducing this into the last expression above, we can write Equation
(9.27) as:
334 ADVANCED METHODS I N DYNAMICS
Note that for the single degree of freedom case Equation (9.29)
reduces to the form :
which checks the conclusions reached in Chapter 5 for the steady state
amplitude of the single degree of freedom system with no damping.
For the multiple degree of freedom system this calculation is not so
simple, because the amplitude coefficients a(r) occur in Equation
(9.29) as the coefficients in a series. These constants can be deter-
mined, however, by expanding the exciting forces Fr into a series of
the same normal functions A t @ ) that appear with the amplitude
coefficients. We will thus write the exciting force as:
and hence the only term left on the right is the single term involving
f ( r ) , and we find :
The final solution for the steady state vibrations is then given by
Equation (9.28) and (9.26):
sin
sin
j- 1
Thus we have :
THE CALCULUS OF VARIATIONS 337
The values of the A's have already been determined in the example
of Section 9.11 above, where we obtained :
A 1 ( l )= 1 Al(2)= 1 A1(3)= 1
A2(l) = 1.802 A2(2) = 0.445 A 2 ( 3 )= - 1.247
A3(1) = 2.245 A3(2) = - 0.802 A3(s)= 0.555
which gives directly the final answer:
XI = -
k
0.241
(0.9 - w2)
k
-
( ~ 3 -
0.435 I
2 w2)
k
sin wt
PROBLEMS
9.36. In the example of the three masses and the three springs worked
out above, suppose that the force F sin wt acts on the right hand mass,
and determine the steady state motion of the left hand mass.
9.37. Refer to the triple pendulum system of Prob. 9.34, and apply a
sinusoidal horizontal force F sin wt
to the left hand pendulum. Calcu-
fia(t;:
late the steady state motion of the
middle pendulum.
9.38. Suppose that during an
earthquake the ground supporting
a shear-type building moves with a
h o r i z o n t a l acceleration a ( t ) .
(a) Show that if xl, x z and xs are
the lateral displacements of the
floors relative to the moving ground,
the equations of forced oscillations (a) (b)
of the building are the same as if PROB. 9.38
the ground were fixed and external
exciting forces as shown in diagram (b) were applied.
(b) If a = a0 sin wt, find the displacement of the first story relative to
the ground, in terms of the applied motions, the natural frequencies, the
masses, and the mode shape numbers A&').
9.13 The Calculus of Variations. Many problems of classical
mechanics are most naturally formulated in terms of maximum or
minimum statements. Consider, for example, the famous problem
of the Brachistochrone, as formulated b y John Bernoulli in 1696.
338 ADVANCED METHODS IN DYNAMICS
"A particle, under the action of gravity, slides along a smooth curve
which lies in a vertical plane. Find the form of the curve for which
the time required for the particle to move between two given points
on the curve is a minimum." As a second example, consider the
statics problem of the determination of the equilibrium form of a
flexible chain, supported a t its ends and hanging in a vertical plane.
This problem can be solved by finding the shape of the chain which
makes the potential energy of the system a minimum.
Such problems have aroused the interest of some of the greatest
investigators. The Brachistochrone problem was first solved
independently by John Bernoulli, Newton,
and Leibnitz and the development of the basic
A
theory of such problems was carried out by
Euler and Lagrange. An idea of the funda-
mental mathematical problem may be obtained
y " f ( ~ ) from the analytical statement of the Brachi-
stochrone problem. In Fig. 9.10, the particle
of mass m slides down the vertical curve AB
Y under the action of the gravity force mg, and
FIG.9.10 we wish to determine the shape of the curve
y = f ( x ) such that the time from A to B is a
minimum. The time required by the particle to travel an arc length
ds is dslv, and thus the total time from A to B is:
From the energy principle, the velocity of the particle will depend
upon the vertical distance traveled :
we also have ds =
[1 + B)'1 ' dx and hence the final expression
for the time becomes:
and we wish to find the values of the x's for which the function has an
extreme value. The necessary conditions for the solution of this
problem are given by the equations:
I = Ixa
Xl
/ (x,y, $) dx, where y = C(x) (9.34)
y =
=
C f (y, y', x) dx, where
d~ find y = +(x) so that I has a stationary value.
&), and y' = -,
dx
340 ADVANCED METHODS IN DYNAMICS
Referring to Fig. 9.11, we note that if the curve shown there as a
solid line is actually the correct +(x) to give the minimum value to I,
then the integral evaluated over some slightly different curve, such
as the dotted curve, will have a slightly larger value. This slightly
different curve, which has the same end points A and B as the exact
curve, is called the varied path. If we imagine that at a particular
value of the independent variable x we move to the varied curve, the
ordinate of the varied curve a t that point will be y + Sy which can
be written as:
y + sy = + ( X I + E~(x) (9.35)
B where Sy is called the variation in y
and the small quantity E indicates
that the varied path is slightly
different from the minimum path.
A We use the variation symbol S to
distinguish Sy from the differential
Y
x dy. Note that dy is the infinitesimal
FIG.9.11 change in the given function $(x)
caused by an infinitesimal change dx
in the independent variable x, whereas Sy is a small arbitrary change
in y equal to &(x).
As the path in Fig. 9.11 is changed from the original to the varied
path, the value of the integral I will change by a small amount
61, called the variation of the integral. If F(y,y', x) is the modified
integrand corresponding t o the varied path, we have :
d d
and thus 6y' = y' +-
dx
(6y) - y ' = - (6y), which indicates that the
dx
processes of variation and differentiation can be interchanged. Equa-
tion (9.37) can now be written as:
Since the varied curve coincides with the exact curve at the end-
points A and B, 6y = 0 at A and B and the term
multiply through this equation by y', and add and subtract the term
y" ,-?fobtaining:
ay
(:
the first three terms in this expression are equal to - y1 g) and the
; -&)
(Y.
df
- -& = 0 =
d [(Y'
& g1)f ]-
y !-
I a - f = constant
ayl
thus giving a general first integral of Euler's equation for this special
case.
EXAMPLE 1. Find the form of the plane curve y = +(x) joining
two points AB with the shortest length.
The function f in the integrand is of the form f (y') and hence we can
use the integrated form of Euler's equation described in Case 1
above :
-a !-- constant
ayl
thus
from which :
this becomes :
1 + cos 0 - 1 - 2c2gr + 2c2gr cos 8
1 - cos 8 2c2gr(l - cos 8)
If we put 2c2gr = 4,this equation is identically satisfied, so that the
Brachistochrone is a cycloid of radius r = 1/4c2g.
It should be noted that in some problems a simplification can be
introduced by an interchange of the independent and dependent
variables. I n the above example of the Brachistochrone, suppose
that we take y as the independent variable and x as the dependent
variable so that the integral becomes:
EULER'S DIFFERENTIAL EQUATION 345
and Euler's Differential Equation becomes
PROBLEMS
9.39. A flexible cable of length I hangs under its own weight from two
supports on the same level a distance d apart. The shape which the cable
assumes is such that the potential energy is a minimum. Set up an
integral expression for the potential energy of the system, and minimize
this integral to find the equation for the cable.
9.40. A particle moves in the x-y plane from a point A to a point B.
The velocity v of the particle is variable, but is a function of y only.
Show that for the path which makes the time from A to B a minimum, the
angle between the tangent to the curve and the y-axis obeys the law:
sin 4
-
v
= constant
Note that this problem is the mechanical version of the optical problem
346 ADVANCED METHODS IN DYNAMICS
of determining the path of a ray of light through a medium which has an
index of refraction which varies in one direction. In the optical case the
above expression is called Snell's Law.
9.41. A container of height h has a circular top and bottom, each of
radius R. The sides are formed of an axially symmetrical surface of
revolution. What should be the shape of the sides to give a minimum
surface area ?
9.42. In studying the resistance of bodies of revolution in a flowing
fluid, Newton made assumptions equivalent to supposing that the pressure
(b) Show that the shape of the curve which minimizes the integral of
part (a) must satisfy the equation
HAMILTON'S PRINCIPLE 347
(c) Show that the equation of the surface of minimum drag in para-
metric form is:
Note that the basic assumption described above applies to only a few
very special cases, and that for ordinary flow conditions the results are
not even a rough approximation.
a a2
writing 6y = - (6y) and 6y" = - (6y) we have :
at ax2
jotEIy " -
ax2
a2
(6y)dx =
a
a
1 IOz
E Iy " - ( 6 ~ )
-
a
EIy" - (6y)dx
ax
= EIy" ,
a lo 1
(6y) - EIy" Sy lo +
1
~ o l E I y zSy
v dr
In this expression, the first term is zero, since a t both 0 and 1 the
bending moment in the beam, which is proportional to y", is zero,
because the beam is supported by frictionless pins a t the ends.
The second term is zero since 6y is zero a t the two ends of the beam.
Thus, the variational equation becomes :
350 ADVANCED METHODS IN DYNAMICS
Since 6y is arbitrary, and can be taken as different from zero, the
condition that the integral should vanish requires that the term in
the parentheses should be zero. This gives the required partial
differential equation of vibration of the beam:
Note that the above variational procedure not only developed the
differential equation, but that the boundary conditions as well
were automatically involved.
PROBLEMS
9.43. A uniform flexible string of mass per unit length p is fastened at
each end to rigid supports a distance I apart. The string is subjected to a
tension force F which may be assumed to be constant for small lateral
motions of the string.
(a) By considering the change of length of the string as it is deflected,
show that the potential energy of the system is:
(a) Assuming small displacements of the pipe, show that the x-com-
ay
ponent of the fluid velocity is v, and that the y-component is - + v ar
at -a
ax
(b) Show that the kinetic energy of the system is:
(c) Show that the potential energy of the system (strain energy of
bending in the pipe) is: