MEISEL

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Lagrvnge', equation S9

This is the second equilibrium equation obtained after substituting Eqs. (2-22) and
(2-23).
Obtaining the equilibrium equations by the energy method seems to be a lengthy
job. For a problem as simple as this one, the equilibrium equations could be written
by inspection. The major use of the energy formulation is for systems involving
certain interaction force terms. These force terms are often difficult to postulate by
inspection, and the energy formulation becomes a more useful tool. Consider this
example and the one preceding it to be illustrations of a new technique and not a
suggestion for complicating a problem you could have solved quite simply by using
your previous experience.

2-4 DEGREES Of fREEDOM AND GENERALIZED COORDINATES


Before generalizing the restricted form of Lagrange's equation, as given
by Eq. (2-17), we must carefully examine certain basic coordinate require­
ments. For a mechanical system composed of n lumped mass and spring
elements completely unattached, a total of n coordinates would be required
to completely specify the configuration of each element. For example, a
set of n position coordinates XI(t), x:(t), . . . ,x..(t) could be used for the
n elements. If the defining characteristic curve for each element were
available, the associated momentum and forces could be determined. The
velocities for the elements are obtained by differentiating these position
coordinates.
If the n elements are now connected together in a particular configura.­
tion, the original n coordinates are no longer free to vary independently.
Connecting the lumped elements together establishes equations of con­
straint between the coordinates. If the constraint equations involve only
relationships between the coordinates, then these are called holonomic
constraints. If the constraint equations involve nonintegrable relation­
ships between coordinate differentials, then the system has nonholonomic
constraints. The number of the degrees of freedom possessed by a system
is defined as the number of coordinates minus the number of constraints
between the coordinates. For a system having n coordinates and m con­
straint equations, there are (n - m) degrees of freedom.
Repeating the restricted form of Lagrange's equation given by Eq.
(2-17), we have

\.+ OV(XI,X:,.
i !i IOT'(XI,X2,
dt

OXt
,x,,)
ox"
• ,x,,)
=
Q
t
for k = 1, 2, . . . ,n (2-26)

It Upon expanding Eq. (2-26) we would have a total of n equilibrium equa­


tions describing the motion of the system. Now suppose the system con­
nections between the lumped elements impose m additional constraint
I
t
60 Principle. of electromechanical-energy conven;on

equations. We would then have a total of (n + m) equations with only


n unknowns. All of the (n + m) equations cannot be independent.
In order for Lagrange's equation to give a complete set of equilibrium
equations which are sufficient to describe the motion of the system, the
coordinates used must be mutually independent. Therefore, the coordi~
nates must be so selected that no constraint equations involving the coor­
dinates can be written. Such coordinates are known as generalized coor­
dinates. 1 The number of generalized coordinates required to describe the
configuration of a system is always equal to the number of degrees of
freedom possessed by the system. If we have n generalized coordinates,
then the system has n degrees of freedom, since the number of constraint
eq~ations between the generalized coordinates must be zero.
The state functions can be formulated in any convenient coordinate
system. However, before these state functions are used in Lagrange's
equation, a set of generalized coordinates must be substituted for the
original set of coordinates. Examples 2~1 and 2-2 in the preceding section
illustrate this procedure.
Suppose we have a system with ,OJ degrees of freedom, but the par­
ticular coordinate system we have chosen describes the system with a
total of six coordinates. Therefore, one constraint equation between the
six coordinates must exist. If the constraint equation were holonomic
(i.e., a relationship between the coordinates and not their differentials),
then the constraint equation could be solved for one of the coordinates
in terms of the other five. By use of this relationship, the selected coor­
dinate could be eliminated in all of the state functions. The state func­
tions can now be used in Lagrange's equation, since the five remaining
coordinates form a generalized set.
If the constraint equations were nonholonomic, then relationships
between the coordinates could not be obtained because the differential
constraint equations could not be integrated. The state functions would
therefore continue to be functions of the six nongeneralized coordinates
and Lagrange's equation could not be used to obtain the equilibrium equa­
tions for the connected system. In the next section we examine the effect
on Lagrange's equation of a general transformation of coordinates.

2-5 A COMPLETE FORMULATION OF LAGRANGE'S EQUATION


FOR CONSERVATIVE MECHANICAL SYSTEMS

The formulation of the energy-state functions required for Lagrange's


equation is very often more convenient in terms of a set of nongeneralized
I The term "generalized coordinates" is somewhat of a misnomer, since the generalized

set of coordinates is in reality a very specific set.


Lagrange's equation 61

coordinates, or coordinates for which constraint equations can be written.


Examples 2-1 and 2-2 have energy-state functions formulated in this man­
ner. In Sec. 2-4, however, we found that the coordinate set used in the
state functions must first be transformed to a generalized set of coordi­
nates before the state functions can be used in Lagrange's equation. A
transformation from a sct of nongeneralized coordinates to a set of gen­
eralized coordinates can usually be accomplished only when the constraints
are holonomic (Le., the equations of constraint are relationships between
the coordinates and not their differentials). In this section we shall
propose a vcry general coordinate transformation and study its effect upon
the restricted form of Lagrange's equation formulated in Sec. 2-3.
Suppose that we decide to change from the original XI, X2, • • • , Xn
coordinates used in Eq. (2-17) to a new set of coordinates denoted by

XI = XI(~I,~2, • • • ,~m,t)
X2 X2(~I,~2, • • • ,~ ... ,t)
(2-27)

where ~ is the symbol used for these m new coordinates and the possibility
of an explicit dependence on the time t is included. I The total time deriva­
tive of anyone of the old X coordinatcs is given, in general, by

(2-28)

Equation (2-28) shows that the old velocities Xt are functions of not only
the new velocities E; but also the new position coordinates ~;, where i is
equal to any integer, from 1 to m. 011 taking the partial derivative of t,he
old velocities Xt with respect to one of the new velocities Ei, we have

(2-29)

This equation reduces simply to

(2-30)

since E. appears only in the term with aXk!a~i as a coefficient.


Taking the restricted form of Lagrange's equation as given by (2-17)
or (2-26) and mUltiplying through by aXk!aEi, which is equivalent to

I The letter ~ is a general coordinate which, for our electromechanical systems, could
be charge, flux linkage, position, or momentum, both angular and translational.
62 Principles of electromechanical-energy conversion

aXk/iH;" gives us
iJXk ~ [aT'(XI,X2, .~ .. ,X,,)] + aXk [aV(XI'X2" . ,X,,)] = aFt Qk
a~, dt aXk a~, ax" a~,
for k 1, 2, . . . ,n (2-31)

The total differential of the product of two functions can be expanded as


follows:

(2-32)
I
Also, the total time derivative of the last term in Eq. (2-32) is equivalent
to

(2-33)

If the old velocities X/o are not a function of the new position coordinates
~i, then the last term in Eq. (2-32) will reduce to zero.
Observe that the first term on the right-hand side of Eq. (2-32) is
identical with the first term in the equilibrium equation (2-31). Substi­
tuting Eq. (2-32) into Eq. (2-31) with the identity of Eq. (2-33) yields
!
for k = 1,2, .. ,n
(2-34)
II
For each value of k one equation similar to Eq. (2-34) can be written.
Adding all of these n equations together gives us as a result j
!
I
i
i
!
Lagronge', equation 63

By using Eqs. (2-27) and (2-28), the state functions T' and V can be
expressed in terms of the m new coordinates and the time t. After making
the substitution of coordinates, the following identities can be written:

,x,,) aXk
a~.
(2-36)
aT'al'~2' ... ,~... ,tl,t2' ,~ ... ,t)
at.
n
" aT' (XI,X2. . . . ,xn ) aXk (2-37)
L,
• -1
aXk at•
aV(~I,t2' ... ,tm,t) aV(XI,X2, .. ,Xn ) aXk
(2-38)
at. aXk at.

Also let us define

(2-39)

The djdt and summation operations in the first term of Eq. (2-35) are
mutually exclusive and can be interchanged. Therefore, after substitut­
ing Eqs. (2-36) to (2-39) into Eq. (2-35), we have

~ [aT'al'~2' ... ,~... ,the2' ... ,tm,t)]


dt a~.
aT'<EI'~2' I ,~ ... ,tJ,t2' ,t....t) + aV(ehe2, ... ,em,t) = Q.
ae; at.'
for i 1, 2, . . . ,m (2-40)

One equation can be written for each value of i = 1, 2, . . . ,m, giving


a total of m equilibrium equations in terms of the m new coordinates.
If these m coordinates are selected to be a generalized set, Eq. (2-40)
represents a complete form of Lagrange's equation.
Notice in Eq. (2-40) that the kinetic coenergy has become a function
of not only the new velodties but also the position coordinates and pos­
siblyeven the time explicitly. The potential function, on the other hand,
is never a function of the velocities, being only a function of the coordinates
and the time. If the kinetic-coenergy function were not actually depend­
ent on the position coordinates, then Eq. (2-40) would reduce to the
original Eq. (2-17) with the exception of the explicit time dependence in
T' and V.
64 Principle. of electromechanical-energy converdon

In order to further systematize this energy formulation, let us define


a new function
.eal'~2, ... '~""~h~2, ... ,~... ,t)
= T'al,~2, ... ,~... ,~l'~" ... ,~... ,t) - V(tl,h, . . . ,~ ...,t) (2-41)
as the difference between the kinetic coenergy and the potential energy.
This new function .e is called the LOIJrangian, and, in general, it is a func­
tion of the ~'s, the ~'s, and t. In terms of the Lagrangian function, Eq.
(2-40) becomes simply
!!: [a.ea,~,t)]
dt a~i
_ a.ea,l;,t)
a~i
= Qt' f
or t
.
=
1 2
, , ... ,m
(2-42)

whete the abbreviated notation .e(tl;,t) is used to indicate the complete


functional dependence of the Lagrangian as given by (2-41). Equation
(2-42) is equivalent to (2-40), since
a.ea,I;,Q = aT'(~,I;,t) _ 0 (2-43)
a~i a~i
and
(2-44)

as seen from the definition of the Lagrangian.


The set of equations given by (2-42) is the usual form of Lagrange's
equation, and it represents a set of m simultaneous differential equations
in terms of the ~ coordinates and the time t. These are the equilibrium
equations for the system, the solution of which yields all of the m coordi­
nates as functions of time and thus describes the complete operation of the
system. Keep in mind that these equilibrium equations are identical
with those obtained from a usual Newtonian force formulation. If these
equilibrium equations can be formulated directly, then Lagrange's equa­
tion offers no further information. Fo'r many systems, however, these
operating equations are difficult to formulate by summing forces. In
these cases the proper energy-state functions, and thus the Lagrangian,
can often be developed. Then by a straightforward application of
Lagrange's equation, the desired equilibrium equations are obtained.
The purpose of this entire Lagrangian development is to aid in the analytic
description of lumped physical systems. The problems in modeling the
actual system and solving the reSUlting equilibrium equations are in no
way simplified. Only the formulation is greatly systematized.

Example 2·3
A classical problem illustrating the real power of this method is the double pendulum
shown in Fig. 2-7. A mass M, is hung by means of a massless rod of length r, from &

You might also like