1m
4 Solution of Electrostatic Problems
where
a-4. 7)
‘The total charge on the sphere is
Q=Q+Q1+Q.+
=ofis+
The series in Hq, (4~78) usually converges rapidly (x < 1/2). Now since the charge
pairs (—Qo, Qs), (—Q1, Qa). yield a zero potential on the sphere, only the original
Qo contributes to the potential of the sphere, which is
(4-78)
wy
Hence the capacitance between the sphere and the conducting plane is, from Eqs.
(4-78) and (4-79),
Q
otal itt —
which is larger than the capacitance of an isolated sphere of radius a, as expected,
‘The potential and clectrc field distributions between the sphere and the conducting
plane can also be obtained from the image point charges.
‘5. Boundary-Value Problems in Cartesian Coordinates
‘We saw in the preceding section that the method of images is very useful in solving
certain types of electrostatic problems involving free charges near conducting bound-
aries that are geometrically simple. However, if the problem consists of a system of
conductors maintained at specified potentials and with no isolated free charges,
cannot be solved by the method of images. This type of problem requires the solu-
tion of Laplace’s equation Fxample 41 (p. 154) was such a problem where the
electric potential was a function of only one coordinate. Of course, Laplace's equation
applied to three dimensions is a partial differential equation, where the potential is,
in general, a function of all three coordinates. We will now develop a method for
solving three-dimensional problems where the boundaries, over which the potential
or its normal derivative is specified, coincide with the coordinate surfaces of an or-
thogonal, curvilinear coordinate system. In such cases, the solution can be expressed
as a product of three one-dimensional functions, each depending separately on one
coordinate variable only. The procedure is called the method of separation of
‘variables.
Problems (electromagnetic or otherwise) governed by partial differential equa-
tions with prescribed boundary conditions are called boundary-calue problems.4-5 Boundary-Value Problems in Cartesian Coordinates 15
Boundary-value problems for potential functions can be classified into three types
(1) Dirichlet problems, in which the value of the potential is specified everywhere on
the boundaries; (2) Newmann problens, in which the normal derivative of the poten-
tial is specified everywhere on the boundaries; (3) Mixed boundary-value problems,
in which the potential is specified over some boundaries and the normal derivative of
the potential is specified over the remaining ones. Different specified boundary con-
ditions will require the choice of diferent potential functions, but the procedure of
solving these types of problems—that is, by the method of separation of variables—
for the three types of problems is the same. The solutions of Laplace's equation are
often called harmonic functions.
Laplace's equation for scalar electric potential V in Cartesian coordinates is,
ev ev av
ae tartar
(4-81)
To apply the method of separation of variables, we assume that the solution V(x, y, 2)
can be expressed as a product in the following form:
V(x, 352) = XOYOIZ), (4-82)
where X(x), ¥(y), and Z(2) are functions of only x, y, and 2, respectively. Substituting
Eq, (4-82) in Eq, (4-81), we have
#X(x) #Y0)
veya) pa + Xe2e) FP” + XI¥O)
which, when divided through by the product X(x)¥(y)Z(2), yields
1 @xe) , 1 yy), 1 WZe)
YO) dy * Ze) dz*
Note that each of the three terms on the left side of Eq. (4-83) is a function of only
cone coordinate variable and that only ordinary derivatives are involved. In order for
Eq, (4-83) to be satisfied for all values of x, y, z, each of the three terms must be @
constant. For instance, if we differentiate Eq. (4-83) with respect to x, we have
af 1 d?x(x)
ila ar |-* we
#70)
a?
(4-83)
since the other two terms are independent of x. Equation (4-84) requires that
1_@Xx(x)
Xe) ae?
Ke (4-85)
where k2 is a constant of integration to be determined from the boundary conditions
of the problem. The negative sign on the right side of Eq. (4-85) is arbitrary, just as
the square sign on k, is arbitrary. The separation constant k, can be a real or an
imaginary number. If k, is imaginary, &2 is a negative real number, making —KZ a176
4) Solution of Electrostatic Problems
TABLE 4-1
Possible Solutions of X”(x) + K3X(z)
Kk xe) Exponential forms! ot x(x)
0 0 Agx+By
+k Ay sin kx + By 008 kx Cet + De
= jk Ay sinh kee + B, cosh ex Cie + Dye
"The exponential forms of Xe) are related to the trigonometric and hyperbolic forms
Tiida etd cofuna by the flowing foemioe
ke
22% = cos sin x, cos kx = eM +69, Bee
et cosh kx sinh kx, cosh ex e+e), sinh ke = He
positive real number. It is convenient to rewrite Eq. (4-85) as
(4-86
Jar manner, we bave
#Y0) | yyy = r
Be tHYO)=0 as
and
£20) 5 12712) =0, (4-88)
dz’
where the separation constants k, and , will, in general, be different from k,; but,
because of Eq. (4-83), the following condition must be satisfied:
Re + =0. (4-89)
Our problem has now been reduced to finding the appropriate solutions—X(x), Y(y),
and Z(2)—from the second-order ordinary diferential equations Eqs. (4-86), (4-87),
and (4-88), respectively. The possible solutions of Eq. (4-86) are known from our
atudy of ordinary differential equations with constant coefiicionte, They are listed in
Table 4-1. That the listed solutions satisfy Eq. (4-86) is easily verified by direct
substitution.
Of the listed solutions in Table 4~1, the first one, Aox + Bo for ky =
line with a slope Ag and an intercept Re at x = 0. When Ay = 0, XC) = Bo.
means that V, the solution of Laplace's equation, is independent of the dimension x.
We are, of course, familiar with the sine and cosine functions, both of which are
periodic with a period 2x. If plotted versus x, sin kx and cos kx have a period
2n/k. Frequently, a careful examination of a given problem enables us to decide
Whether a sine of a cosine function is the proper choice. For example, ifthe solution
is to vanish at x =0, sin kx must be chosen; on the other hand, if the solution is
expected to be symmetrical with respect to x = 0, then cos kx is the right choice. Inm7
FIGURE 4-14
Hyperbolic and exponential functions,
the general case, both terms are required. Sometimes it may be desirable to write
Ay sin kx + By cos kx as A, sin (kx +) or A, cos (kx + Yt
For k, = jk the solution converts to hyperbolic functions:
sin jkx = —j sinh kx
and
cos jl = cosh ke.
Hyperbolic functions are combinations of exponential functions with real exponents,
and are nonperiodic. They are plotted in Fig. 4~14 for easy reference. The important
characteristics of sinh kx are that it is an odd function of x and that its value ap-
proaches +00 as x goes to +t w. The function cosh Ax is an even function of x, equals
unity at x = 0, and approaches + co as x goes to +00 or —co.
‘The specified boundary conditions will determine the choice of the proper form
of the solution and of the constants 4 and B or C and D. The solutions of Eqs. (4-87)
and (4-88) for ¥(y) and Z(e) are entirely similar.
EXAMPLE 4-6 Two grounded, semi-infinite, parallel-plane electrodes ate separated
by a distance b. A third electrode perpendicular to and insulated from both is main-
tained at a constant potential Vo (sec Fig 4~15). Determine the potential distribution
in the region enclosed by the electrodes.
Aen (Ai BP,
Asin Ve. By =
cosy) sin x + (A, sin) 08 bx, Ay = A, 608 ys By = A, sin
Y. = tan" (By/Ay) A, cos ke + 9) = A, sin) sin hx + (Ace 9 cos kx A,
A, 808 Ya: A= (AF + BM, = tan! (—A,/B,)178
4 Solution of Electrostatic Problems
FIGURE 4-15
Cross-sectional figure for
Example 4-6. The plane
electrodes are in
direction.
Fig. 4-15, we write down the boundary
conditions for the potential function V(x, y, 2) as follows.
With V independent of z:
V(x 9,2) = Vos, 9) (4-900)
In the x-direction:
VO, y) = Yo (4-908)
¥(00, y) = (4-906)
In the y-direction:
¥¢x,0)=0 (4-908)
V(x, 6) = 0. (4-90e)
Condition (4-90a) implies k, = 0, and from Table 4-1,
2G) = Bo. 90)
The wonstant Ao vanishes because Z in independent of s. From Eq, (1 89) we have
-K=k, (4-92)
where k is a real number. This choice of k implies that kis imaginary and that k,
is real, The use of k, = jk, together with the condition of Eq. (4-900), requires us to
choose the exponentially decreasing form for X(x), which is
X() — Dye 4-99
In the y-direction, k,
Y()) from Table 4-1 is
k. Condition (4-904) indicates that the proper choice for
Y(y) = A; sin ky. (4-94)
Combining the solutions given by Eqs. (4-91), (4-93), and (4-94) in Eq. (4-82), we
obtain an appropriate solution of the following form:
Vales y) = (BoD 2A,)e™ sin ky
= Ce ™ sin ky, (4-99)
where the arbitrary constant C, has been written for the product BoD 24,
‘Now, of the five boundary conditions listed in Eqs. (4-90) through (4-90e) we
have used conditions (4-90a), (4-90), and (4-904). To meet condition (4-90), we4-5 Boundary-Value Problems in Cartesian Courdinates 179
require
Vi{x, b) = C,e-™ sin kb = 0, (4-96)
which can be satisfied, for all values of x, only if
sinkb =0
kb = nn
or
4 , 49
Therefore, Eq. (4-95) becomes
Vise y) — Co cin = » (4 98)
Question: Why are O and negative integral values of n not included in Eq. (4-977.
We can readily verify by direct substitution that V(x, y)in Eq. (4-98) satisfies the
Laplace's equation (4-81). However, V(x ») alone cannot satisy the remaining
boundary condition (4-90b) at x = 0 forall values of y from 0 to b. Since Laplace's
‘equation is a.linear partial differential equation, a sum (superposition) of V(x, y) of
the form in Eq, (4-98) with diferent values of mis also a solution. At x = 0, we write
v0.9=¥ won= > c, sin
=h, Vb
In this region the potential must decrease to zero as r ~+ . Terms containing
the factor #* cannot exist, and the appropriate form of solution is, from Eq.
(4-125),
M0, = Y Var 9)
4-140)
V(b, @) =)" Bb sin mp
_[ % ford b, we note the following
boundary conditions:
V(b, 6) =0" (4-1550)
V{K, 0) = —Bo2 = —BoK cos U, tor Rb, (4-1556)
Equation (41556) is a statement that the original E, is not disturbed at points
very far away from the sphere. By using Eq, (4-154) we write the general solution
"For this problem itis convenient to assume that V = 0in the equatorial plane (9 = x/2, which leads to
V{b,0)=0, since the surface of the conducing sphere is equipotental, (See Problem P4-28 for
YO, )= Ve)»
indary-Value Problems in Spherical Coordinates 191
as
V(R,O)= J [AR + BR“ }P,(c0s), RZb (4-156)
However, in view of Eq, (4-155b), all 4, except A, must vanish, and A, = Fo.
‘We have, from Eq, (4-156) and Table 4-2,
VIR, &) =
ERP,(c0s 6) + )° BAR" »P,{c0s 8)
eR BR? = BRON O+ SBR MPfonO, REL
~ (157)
‘Actually, the first term on the right side of Eq, (4-157) corresponds to the potential
of a charged sphere. Since the sphere is uncharged. Bo = 0. and Ea. (4-157)
becomes
VR) = (
Now applying boundary condition (4 1$5a) at R= b, we require
O= @ - fo) 008 0+ x B,b-"*P,c0s 0),
‘By
8 alone Sarg, RB eH
from which we obtain
By = Eb
and
B= 0, n>2.
We have. finally from Ea. (4~158
VIR, 0) = ~Fa[1—(3) ]reose Reb. (4-155)
The electric field intensity E(R,) for R > b can be easily determined from
=WWER, OF
HR.) = agg + ME 6-160)
where
aft +2(3) ‘oor R>b 4-160)
and
= -61 S ) ‘sa R>b. (4-160)
‘The surface charge density on the sphere can be found by noting that
P40) = €oE nly, = 3€0E0 008 8, (ste
which is proportional to cos @, being zero at @ = x/2. Some equipotential and
field lines are sketched in Fig. 4-21. -12
4 Solution of Electrostatic Problems
It is interesting to note from Eq, ($-159) that the potential is the sum of two
terms: ~ZoR cos 6 due to the applied uniform electric field; and (Egb? cos 6)/R? due
to an electric dipole of a dipole moment:
p= anecb®Ey (4-162)
at the center of the sphere. The contribution of the equivalent dipole can be verified
by referring to Eq. (3-53). The expressions in Eqs. (4~160a) and (4-160b) for the
resultant electric field intensity, being derived from the potential, obviously also rep-
resent the combination of the applied uniform field and that of the equivalent dipole,
piven in Eg. (3-54),
In this chapter we have discussed the analytical solution of electrostatic problems
by the method of images and by direct solution of Laplace’s equation. The method
of images is useful when charges exist rear conducting bodies of a simple and com-
patible geometry: a point charge near a conducting sphere or an infinite conducting
plane; and a line charge near a parallel conducting cylinder or a parallel conducting
plane. The solution of Laplace's equation by the method of separation of variables
requires that the boundaries coincide with coordinate surfaces, These requirements
restrict the usefulness of both methods. In practical problems we are often faced with
‘more complicated boundaries, which are not amenable to neat analytical solutions.
In such cases we must resort to approximate graphical or numerical methods. These
‘methods are beyond the scope of this took.t
Review Questions
RA-1 Write Poisson's eau
4) for a simple medium,
}) for a linear and isotropic but inhomogeneous medium.
Repeat in Cartesian coordinates both parts of Question R-4-1
‘Write Laplace's equation fora simple medium
4) in vector notation, —_b) in Cartesian coordinates,
RAM4 IF VU = 0, why docs it not follow that U is identically zero?
RA-~5 A fixed voltage is connected across s parallel-plate capacitor.
a) Does the elesric eid intensity inthe space berween the plates depend on the
permittivity ofthe medium?
') Does the electric fux density depend on the permittivity of the medium?
Explain
RA 6 Acoume that ned shargwe | Q and are daporited on the plate of an igolated
parallel-plate capacitor.
4) Does the electric field intensity in the space between the plates depend on the
permittivity ofthe medium?
) Does the eletrc fx density depend on the permittivity of the medium?
Explain
jon in vector notation
Ra.
RA.
See, for instance, B. D. Popovit, Introductory Engineering Electromagnetics, Chapter S, Addison-Wesley
Publishing Co, Reading, Mass, 1971