Int. J. Engng Ed. Vol. 15, No. 6, pp. 456±468, 1999 0949-149X/91 $3.00+0.

00
Printed in Great Britain. # 1999 TEMPUS Publications.

The Use of MathCad in Teaching Ideal

Fluid Flow with Complex Variables*
MICHAEL REX MAIXNER
Maine Maritime Academy, Castine, Maine ME, USA. E-mail: [email protected]
The use of MathCad for the visualization of two-dimensional, irrotational, steady, incompressible
flow patterns is discussed, along with the appropriate mathematical background. MathCad's
contour plotting capabilities and its ability to handle complex numbers permit rapid depiction of
basic flows and linear combinations of these flows with relative ease, allowing students to
concentrate on understanding the material and broader concepts. Sample output, recommended
lecture topics, and suggested topics for further study are presented, and the address of a web site
containing a prepared lecture is provided (http://www.mathsoft.com/appsindex.html).

NOMENCLATURE
 source (or sink) strength
a cylinder radius  doublet strength
f complex function used in definition of  = Re(F), potential function
Milne-Thompson circle theorem (see text) = Im(F), stream function
F ˆ  ‡ ^i , complex potential
dF
F0 ˆ ˆ u ÿ ^iv, complex velocity
dz
0
F ˆ up‡^iv,
 velocity INTRODUCTION
^i ˆ ÿ1 (Note the carat (^) atop the letter
i to differentiate it from the letter i when FLOW VISUALIZATION has been an invaluable
employed as a subscript or index.) tool in the study of ideal and real fluid flows,
i index associated with x-direction providing a synergism that allows the student to
j index associated with y-direction gain an understanding not only of the phenomen-
n number of subdivisions in x- and y- ological aspects of the flow, but also of the under-
directions lying mathematics. Detailed analysis of these flow
Q array of source and sink strengths in von visualizations permits not only the streamline and
KaÂrmaÂnpproblem (see text) equipotential patterns to be obtained, but also the
r ˆ jzj ˆ x2 ‡ y2 , modulus of position location of critical points in the flow, such as
in the complex plane stagnation points. Numerous methods of flow
Uo magnitude of free stream velocity visualization exist, and include electric analogs,
u velocity in the x-direction experimental techniques (hydrogen bubble
v velocity in the y-direction method, smoke tunnels, aluminum flakes sprinkled
V ˆ jF 0 j ˆ jF  0 j, magnitude of velocity on the surface of dye-colored water, dye injection,
x real component of position in the etc. [1, 2]). Given the requisite time and patience,
complex plane qualitative sketches of potential flow patterns
X array of source and sink locations in von provide additional insight into the underlying
KaÂrmaÂn problem (see text) physics. A natural outgrowth of sketches is the
y imaginary component of position in the use of computer graphics to assist in these visual-
complex plane izations, but early attempts at the use of computers
^
z ˆ x ‡ ^iy ˆ rei , position in the complex for this purpose were generally limited to main-
plane frame machines with, eventually, time-sharing
 orientation angle of free-stream relative workstations with graphics terminals [3±6]. More
to positive x-axis recent developments in flow visualization with
 orientation angle of doublet relative to personal computers are provided in the references
positive x-axis [7±9]. MathCad, a relatively inexpensive and read-
ÿ vortex strength ily available calculation software product, com-
 angle of position in the complex plane, bines an excellent contour plotting capability and
measured counter-clockwise from posi- the ability to perform calculations with complex
tive x-axis functions. This makes it an extremely powerful
tool in the visualization of ideal fluid flow patterns
* Accepted 6 June 1999. by individual users on personal computers.

456
 The Use of MathCad in Teaching Ideal Fluid Flow with Complex Variables 457

Students with a background in basic fluid BACKGROUND

mechanics and complex variables are quickly able
to obtain plots of streamlines and equipotential Students who took the pilot version of Es490 in
lines. the fall of 1998 and spring of 1999 were required to
At Maine Maritime Academy, the course purchase the student edition of MathCad, version
`Numerical and Computer Methods for Engineer- 7. The professional edition of MathCad was also
ing Design' (Es490) is taken in the spring term of available for both classroom instruction and after-
the third year of a five-year marine engineering hours use on the local-area network. The topic of
systems curriculum. Prior to that, students have potential flows is presented in the latter part of the
had basic courses in fluid mechanics and struc- term, so students will already be familiar with
tured computer programming, and will also have MathCad when this subject is reached.
been introduced to complex variables. Es490 also A detailed lesson plan file in MathCad work-
follows an introductory design course, and is sheet format is provided to all students to follow
taken in conjunction with an intermediate design along during the lectures and to use as a study
course. The course provides students with a variety guide. A copy of the lesson plan may be obtained
of numerical methods and techniques (numerical through the MathCad web site or directly from the
integration, differentiation, finite difference author by e-mail. Due to the number of illus-
methods, interpolation, curve fitting, solution of trations embedded in the lesson plan and the grid
differential equations and systems of differential size used for illustrations, it is recommended that
equations, etc.) that may be used throughout the program be run on a computer with at least a
the remainder of their curriculum, including the Pentium processor, otherwise the refresh times for
two-semester capstone design sequence taken in successive screens becomes excessively long. Alter-
the fifth year. Excel spreadsheets, QBASIC, natively, the grid coarseness may be increased by
and MathCad are utilized to complete a variety reducing the number of grid points (see next
of problems (structural, electrical, fluid mech- section) with a concomitant reduction in graphic
anics, optimization, aerodynamics, heat transfer, quality.
etc.). The background for this topic may be found in
As part of this course, students receive various mathematics [10,11] and fluid mechanics
instruction in the representation of ideal fluid [12] texts; many of the details are omitted here for
flows as a complex function whose real and the sake of brevity, but may be found in the lesson
imaginary parts are, respectively, the potential plan placed on the MathCad web site. Instruction
and stream functions. Following this basic begins with a review of the basics of complex
instruction and several examples, students are number theory, whereby an imaginary number, z,
required to solve problems involving linear may be represented in either rectangular (x- and y-
combinations of these potential flows, to draw components) or polar (r- and -components):
elementary conclusions from these representa-
^
tions, and, in general, to recognize the power z ˆ x ‡ ^iy ˆ rei ˆ r…cos  ‡ ^i sin † …1†
of this method, especially when used in conjunc-
tion with boundary layer methods at fluid- where  = arctan (y/x).
structure interfaces. The prototype of this For the two-dimensional, irrotational, steady
instructional method was employed in the flow of an incompressible fluid, the u- and v-
early 1980s by the author when on the faculty components of velocity (in the x- and y-directions,
of the Department of Mechanical Engineering respectively) are related by the Cauchy-Riemann
at the Naval Postgraduate School in Monterey, equations:
California. At that time, an IBM 360 main-
frame was required to run the program, with @ @
vˆÿ ˆ …2†
students utilizing time-sharing stations and separ- @x @y
ate Tektronix graphics screens to view the flow @ @
patterns. With the advent and proliferation of uˆ ˆ :
@y @x
personal computers and software since then, the
same capability is now available on desktop and where  and are, respectively, the equipotential
laptop computers. The intent of this instructional and stream functions for the flow; they are ortho-
package (originally at the Naval Postgraduate gonal, and give rise to the complex potential
School and in its current incarnation at Maine function:
Maritime Academy) was to relieve students of
the laborious plotting associated with flow visual- F …z† ˆ …x; y† ‡ ^i …x; y† …3†
ization. The student is thus able to investigate
many more combinations of basic potential flows Most students readily grasp the significance of a
and, consequently, obtain a better understanding streamline (a line of constant ) and recognize
of the subject matter. It is this method and its that there can be no velocity normal to a stream-
specific implementation in Es490 at Maine Mari- line. The concept of an equipotential (a line of
time Academy with which the remainder of this constant ) is usually not as easily mastered, and
article is concerned. must be reinforced through the correlation of the
458 M. Maixner

Cauchy-Riemann equations with plots of equi- Once the grid is established and and/or  are
potential lines in ideal flow patterns. evaluated at each grid point, MathCad's contour
The student is then shown that the complex plotting capability is used to plot or . MathCad
velocity may therefore be obtained from: performs a linear interpolation between gridpoints
to establish contours of the quantity being plotted
dF @F @ ^ @ (either  or ); the ij or ij are stored as a matrix
F 0 …z† ˆ ˆ ˆ ‡i ˆ u ÿ ^iv …4†
dz @x @x @x of equally spaced values, without reference to the
coordinates at which  or are evaluated. The
MathCad's built-in derivative function may be
default plotting limits for grid positions are
employed in this last equation. Note that in
ÿ1  x  1 and ÿ1  y  1, and must be changed
order to obtain the actual velocity, we must take:
by the user to reflect the values previously chosen
(i.e., xmin  x  xmax and ymin  y  ymax ).
F 0 …z† ˆ u ‡ ^iv …5†
Version 7 of MathCad limits the user to one
Finally, the magnitude of the velocity may be contour plot per graph. If, instead, two contour
obtained from: plots were allowed, then lines of constant  and
lines of constant could be viewed on the same
V …z† ˆ jF 0 …z†j ˆ jF 0 …z†j ˆ ‰F 0
F 0 j1=2 ˆ …u2 ‡ v2 †1=2 plot, thereby illustrating the orthogonality of these
two functions. Instead, the user must plot  and
…6† on separate graphs; if placed in close proximity to
one another, though, the orthogonality is readily
DISCRETIZATION OF FLOW FIELD discernable. MathCad Version 8 has enhanced
three-dimensional and contour plotting capabil-
Prior to displaying any of the more interesting ities, which allow the plotting of two sets of
potential flows, a grid must be established, with the contours on the same graph.
grid intersections being the points at which  and
will be calculated. A grid is employed with x and
y each bounded by 1.5, with the domain sub- BASIC FLOWS
divided into a number of rectangles in the x- and y-
directions. All grid points are then offset slightly With the preliminaries out of the way, the
by the same amount xoffset = yoffset = 0.0001; student is now introduced to the basic potential
should the user decide to place a singularity at, flows.
say, z ˆ 0 ‡ ^i0, (a common choice), then calcula-
tion will not be performed `at' the singularity. This Free stream
shift is essentially transparent to the user from this A free stream of magnitude Uo and inclined at
point on. An equal number of grid points is an angle to the positive x-axis of  will have the
utilized in each direction, with indices in the x- potential function:
and y-directions, respectively, of i ˆ 0 ! n and ^
j ˆ 0 ! n; with n ˆ 25, this gives, in all, (n + 1)2 = Ffreestream …z† ˆ Uo zeÿi …8†
676 points in the plotting area. MathCad's default
index counter is set to 0 so that all vector and matrix Recalling that  is the real part of the complex
subscripts begin with 0; the grid point corres- function, and that is the imaginary part, we
ponding to i = 0, j = 0 is situated at the lower left have i,j = Re(Ffreestream(zi,j)) and i,j =
corner of the plotting area. Im (Ffreestream(zi,j)). In the lesson plan, items such
The complex offset zoffset and complex grid- as  and Uo are included in the text as `math
points zi,j are defined as: regions' so that they may be varied as desired. In
fact, the student is encouraged to do so, and to
zoffset ˆ xoffset ‡ ^iyoffset …7† examine changes in the output ± this is the real
benefit of a program such as MathCad. As a
i
…xmax ÿ xmin † consequence of the Cauchy-Riemann equations,
xi ˆ xmin ‡
n the streamlines and equipotential lines are ortho-
j
…ymax ÿ ymin † gonal for all potential flows and for all linear
yj ˆ ymin ‡ combinations of these flows. MathCad allows
n various modifications to the contour plots, includ-
zi;j ˆ zoffset ‡ …xi ‡ ^iyj † ing colors or shades of gray between successive
contours, automatic contour plotting, choice of
(The choice of zoffset and the grid size (i.e., n) may numbers of contours, etc. Figure 1 depicts the
result in large values of F(z), depending on how equipotential lines and streamlines for a free
close a grid intersection lies to the location of a stream of magnitude Uo = 1 and  ˆ =3.
singularity. Since MathCad may arbitrarily select
the contour values, adjacent contours may then Source/Sink
appear very dense or very sparse. The user may be The complex potential for a source is given as:
required to make judicious choices of n and the
location of singularities to produce suitable 
Fsource …z† ˆ ln…z ÿ zsource † …9†
contour density.) 2
 The Use of MathCad in Teaching Ideal Fluid Flow with Complex Variables 459

Fig. 1. Equipotentials and streamlines for free stream of strength U0 ˆ 1 and  ˆ =3. Note the orthogonality of the two plots.

A sink is merely a source of negative strength. In tion relative to the positive x-axis of angle  (fluid
Fig. 2 the equipotential lines and streamlines are emanating from the doublet on the side originally
plotted for a source of strength of  = 2 and occupied by the source, and fluid entering the
located at zsource = 0, using the relations i,j = doublet on the side originally occupied by the
Im(Fsource(zi,j)) and i,j = Re(Fsource(zi,j)). Notice sink); the complex potential is given as:
that the streamlines all emanate from the origin
^
of the source, and that the equipotential lines and ÿeÿi
streamlines are everywhere perpendicular. Note Fdoublet …z† ˆ …10†
2…z ÿ zdoublet †
also the concentrations of the streamlines along
the negative x-axis ± this is the location (within Equipotential lines and streamlines for a doublet at
MathCad) of the so-called Riemann cut; the value the origin and with strength  = 2 and orientation
of jumps by a value of 2 as the Riemann cut is  =  are depicted in Fig. 3.
traversed.
Vortex
Doublet For a vortex of strength ÿ (a positive value of ÿ
A doublet is formed when a source and sink of results in a counter-clockwise velocity), the
equal magnitude are brought together, maintain- complex potential is:
ing the product of strength and separation at a
constant value. In the limit, as the separation ÿ^iÿ
between the source and sink tends towards zero, Fvortex …z† ˆ ln…z ÿ zvortex † …11†
2
a doublet of strength  will result with an orienta-

Fig. 2. Equipotentials and streamlines for source of strength  ˆ 2.

460 M. Maixner

Fig. 3. Equipotentials and streamlines for a doublet of strength  ˆ 2 and orientation  ˆ .

Figure 4 depicts the equipotentials and streamlines toward the origin. Were a source used instead of
for a vortex of strength ÿ = 2 situated at the a sink, the flow would spiral outward from the
origin. At this point, students are shown how the origin, much as is obtained in a centrifugal pump.
equipotentials for a source are analogous to the
streamlines for a vortex, and vice-versa; this is due Aircraft trailing vortex system
to the fact that their complex potentials are similar, When two vortices of equal magnitude, opposite
differing only by a factor of ÿ^i. sign, and located at z1 and z2 are combined, the
following complex potential results:

COMBINATIONS OF BASIC FLOWS ÿ^iÿ

FTV …z† ˆ ‰ln…z ÿ z1 † ÿ ln…z ÿ z2 †Š …13†
2
`Bathtub' vortex  
By combining a sink and a vortex, the following ÿ^iÿ z ÿ z1
complex potential results: ˆ ln
2 z ÿ z2
 ^iÿ Figure 6 depicts the streamlines obtained from two
Fbathtub …z† ˆ ÿ ln…z ÿ zsink † ÿ ln…z ÿ zvortex †
2 2 such vortices of magnitude 2 and situated z = 1.
…12† In fact, what this situation represents is the flow
pattern exhibited by a lifting wing, with a trailing
Choosing  = ÿ = 2 and zsink = zvortex = 0, a plot vortex emanating from each wingtip. The vortices
of the streamlines for this flow (Fig. 5) shows that are `images' of one another, and result in a plane of
the streamlines spiral, with the flow directed symmetry at mid-span (x = 0), where = 0.

Fig. 4. Equipotentials and streamlines for vortex of strength ÿ ˆ 2.

 The Use of MathCad in Teaching Ideal Fluid Flow with Complex Variables 461

Fig. 5. Streamlines for ``bathtub'' vortex.

The Magnus effect the round after exiting the muzzle. Magnus discov-
In the early 1850's, the Prussian army noted that ered that when the projectiles were subjected to a
rounds fired from their artillery pieces were falling crosswind in flight, the velocity on either the top or
long or short of the target, for no apparent reason. bottom of the spinning projectiles (depending on
Gustav Magnus [13] was given the task of investi- the wind's direction) was augmented by the wind.
gating this phenomenon. The bores of the artillery This resulted in reduced pressure on that side of
pieces were rifled to impart a spin to the rounds as the projectile and a net force on the projectile in
they traversed the barrel to introduce stability to that direction. This tended to either keep the

Fig. 6. Streamlines resulting from aircraft trailing vortex system.

462 M. Maixner

projectile aloft for a longer time (low pressure cent will be realized when compared to the fuel
region on top) or a shorter time (low pressure consumption of a more conventional vessel of the
region on the bottom). It is this same phenomenon same size [16].
which causes golf balls, baseballs, and other spin-
ning spherical objects to deflect in flight [14]. (See
the article by Swanson [15] for a more recent PROBLEMS FOR FURTHER STUDY
discussion of this effect.) The power of complex
variables may be used to depict this region of The basic flows and combinations thereof are
higher velocity for the two-dimensional, rather contained in the lesson plan posted on the Math-
than the three-dimensional case (i.e., the Magnus Cad web site; the following constitute additional
problem of a spinning cylinder, rather than a problems which may be employed as homeworks,
spinning sphere). This particular problem may be projects, or to merely spark discussion.
easily handled with MathCad.
Consider the combination of a freestream of Wing in ground effect
magnitude Uo = 1 oriented  = 0 (flow in the As an extension of the lifting wing problem,
positive x-direction) and a doublet of strength consider an aircraft in close proximity to the
2Uo oriented  =  (in the negative x-direction) ground (i.e. during take-off or landing) [17, 18].
and situated at the origin. This results in a cylinder In addition to the plane of symmetry along the
of unit radius whose surface is the locus of points centerline of the aircraft discussed previously,
where = 0. Since there can be no flow through there must now be another along the ground
the surface of the cylinder (i.e. no normal velocity since there can be no flow through that plane
on the cylinder), the tangential velocity is the total either. The standard model of this is to add to
velocity along this surface. We will also include a each wingtip (located at z1 and z2) a vortex which
vortex of strength ÿ at the origin ± this will later be must have an image of opposite sign situated as far
varied to investigate the effect of circulation on below the ground (located at, respectively, z4 and
the flow pattern. The complex potential that z3) as the wingtip vortex is above the ground. The
results is: appropriate complex potential is;
 
^
^
eÿi ^iÿ ÿ^i ÿ …z ÿ z1 †…z ÿ z3 †
FMagnus …z† ˆ Uo eÿi z ÿ ÿ ln z …14† FGE ˆ ln …15†
2z 2 2 …z ÿ z2 †…z ÿ z4 †
With ÿ = 0, the resulting streamline plot is shown The streamlines which result from a vortex
in Fig. 7(a); stagnation points exist on the cylinder strength of ÿ = 2, a wingspan of 2, and an altitude
at the intersection of the cylinder surface with the of one-quarter wingspan are depicted in Fig. 8(a).
x-axis. As the vortex circulation is increased to Since there can be no flow through the ground, the
ÿ ˆ ÿ2 (i.e. circulation in a clockwise direction), horizontal velocity there is the total velocity.
the stagnation points move to locations on the If positions along the ground are represented in
lower two quadrants of the cylinder, as seen in Fig. complex number form and inserted in:
7(b). Further increasing the circulation to ÿ ˆ ÿ4  
results in the stagnation points meeting at the d 
bottom of the cylinder, Fig. 7(c), and a further

Vground …z† ˆ  FGE …z† …16†
dz
increase results in one stagnation point moving
into the flow, and the other moving inside the the velocities induced on the ground are obtained
cylinder, Fig. 7(d)! as in Fig. 8(b). It can be seen that the maximum
The region of more densely packed streamlines induced velocities on the ground occur immedi-
on the upper surface of the cylinder corresponds to ately beneath the real vortices. An additional topic
a region of higher velocity since the free stream that may be discussed is the path taken by the
velocity adds to the rotational velocity imparted by wingtip vortices after they are shed; the point must
the spinning cylinder. From Bernoulli's theorem, be made that each vortex (real or image) cannot
this is also a region of lower pressure, so that a net induce any velocity on itself. Therefore, to calcu-
force in the positive y-direction results. At this late the velocity of a real vortex, the resulting
point, a discussion of viscous effects and separ- complex potential must exclude the term corres-
ation is appropriate, since the real flow would ponding to that vortex prior to calculating its
separate from the cylinder somewhere in the first velocity. The distance traveled by the vortex is
quadrant (i.e. on the downstream side of the upper then obtained via an Eulerian scheme over a
surface). It is also instructive to plot the velocity on suitably small time interval; the velocity is multi-
the surface of the cylinder at several points for the plied by the time interval, and this result is added
ideal (i.e. unseparated) flow. A device employing to the previous position of that vortex to update its
this principle (a 26-meter high `turbosail') will be position. The other real vortex and the image
installed as a means of supplementing more vortices move a similar distance, but symmetrically
conventional means of propulsion on Calypso II, relative to the aircraft centerline and the ground.
the research vessel for the Cousteau Society ± it is The resulting path, Fig. 8(c), is hyperbolic, with the
anticipated that fuel savings on the order of 30 per vortices moving toward the ground and outward.
 The Use of MathCad in Teaching Ideal Fluid Flow with Complex Variables 463

Fig. 7. Streamlines for flow about a cylinder with: (a) No circulation, with two stagnation points situated on x-axis; (b) Circulation of
intermediate strength ÿ ˆ ÿ2, with two stagnation points moving down and towards each other; (c) Critical circulation of ÿ ˆ ÿ4,
resulting in one stagnation point at the bottom of the cylinder; (d) Circulation of strength ÿ ˆ ÿ4:03, with one stagnation point
moving towards the origin (inside cylinder) and the other in the flow moving away from the cylinder.

At this point, a discussion is appropriate of the discussed above. For a detailed discussion of the
effect that the trailing vortices shed by a large method, see References [19±22].
aircraft could have on smaller aircraft, and why Von KaÂrmaÂn's method begins by placing a series
there can be a significant wait on the ground as of sources towards the bow along the centerline
aircraft queue for take-off. (i.e. the x-axis) of the body, and a series of sinks on
the centerline towards the stern. The body is
Von KaÂrmaÂn problem presumed to be operating head-on into a free
In the 1920s, Theodore von KaÂrmaÂn was stream of magnitude Uo. The sum of all sources
contracted by the Zeppelin company to investigate should equal the sum of all sinks, so that there will
the pressure distributions on airship hulls. His not be any flow through the body. What is
ingenious method of combining various singula- required is the proper combination of source and
rities to obtain the required airship body profiles, sink strengths and locations so that the body's
although created for the axisymmetric case, may outline is matched by the = 0 streamline.
be easily demonstrated with MathCad using com- A drawback of MathCad (before version 8) is
binations of the two-dimensional singularities that no other plot may be superposed on a contour
464 M. Maixner

Fig. 8. (a) Streamlines resulting from aircraft in ground effect; (b) Velocities induced on ground; (c) Path of vortex shed from starboard
wingtip.

plot. (Version 8 of Mathcad supports multiple Figure 9 depicts the streamlines obtained from:
plots of different types, as well as multiple contour 2 3 2 3
plots.) Consequently, an ellipse or other outline 0:8 ÿ1:0
which represents the airship hull may not be placed 6 1:0 7 6 ÿ0:9 7
6 7 6 7
on a plot, so that the = 0 streamline may be 6 7 6 7
6 0:9 7 6 ÿ0:8 7
compared to it as singularities are changed. Qˆ6 6 7 6 7
7 and X ˆ 6 0:8 7 …18†
Instead, the student may be told to make the = 6 ÿ0:9 7 6 7
0 streamline touch certain points on the x- and y- 6 7 6 7
4 ÿ1:0 5 4 0:9 5
axes. Specifically, a combination of sources and
sinks situated in a free stream is required which ÿ0:8 1:0
will place the = 0 streamline at y = 0.75 and x = As the student varies any of the singularity para-
1.25. An array Q containing the strengths of meters, the MathCad plot immediately updates
three sources and three sinks is used, along with itself and provides instantaneous feedback
an array X which contains the x-coordinates of the regarding the effect of each change. Outlines of
singularities. These will be used in the complex actual airships are, in fact, not nearly as blunt
potential: as the outline depicted in Fig. 9; suitable choices
1 X of source and sink locations and strengths in the
FVK …z† ˆ U0 z ‡ ln…z ÿ Xisource † …17† Q- and X-arrays could provide more realistic
2 isource shapes.
 The Use of MathCad in Teaching Ideal Fluid Flow with Complex Variables 465

Fig. 9. Illustration of von KaÂrmaÂn's method: two-dimensional free stream about a series of sources and sinks situated along the x-axis.
The streamline ˆ 0 represents the outline of the body.

Circle theorem Studying the right hand side of this last equation
With the aircraft trailing vortex example, the reveals that the first term is the real vortex; the
aircraft centerline was seen to be a plane through second term is an image vortex of equal (but
which no flow passed; similarly, an aircraft in opposite) strength situated at 1=zreal ; the third
ground effect also had no flow through the plane term is a vortex of the same magnitude and sense
that represented the ground. This technique is as the real vortex, but located at the origin; the
referred to as the method of images, and is not last term is a constant which ensures that the
restricted to plane boundaries. In fact, Milne- locus of = 0 coincides with |z| = a = 1. Had
Thompson's circle theorem [23] states that if the the last term been left off, there would still be a
complex potential f(z) represents a flow without circular streamline on |z| = a = 1, but its
singularities for |z| < a, then: value would be non-zero. It is a simple matter
 2 to plot the resulting streamlines using either of
 a the last two equations in MathCad; additionally,
F …z† ˆ f …z† ‡ f …19†
z if the second equation is used, the change in
streamline values which occurs when the last
represents the same flow at infinity with a circle of term is included or left out can be readily and
radius `a' at the origin. (When used on a function, quickly observed.
the overbar notation f indicates that all complex Figure 10 illustrates either of the last two equa-
constants in the original function f are now their tions with ÿ = 2 and zreal = 1.2 + 1.2^i; note that
complex conjugates.) In the case of a vortex situated the image vortex is located at zimage ˆ 1=zreal ˆ
at zreal in the vicinity of a circular cylinder of unit :417 ‡ :417^i on a line between the origin and the
radius, the circle theorem gives the complex potential: real vortex. Students should be encouraged to
  invoke the circle theorem for multiple vortices
ÿ^iÿ ^iÿ 1 to simulate the classic KaÂrmaÂn vortex street, or
F …z† ˆ ln…z ÿ zreal † ‡ ln ÿ zreal …20†
2 2 z to verify the validity of the circle theorem when
applied to sources, sinks, and/or doublets.
Following a bit of algebra, this equation may be
recast as: Kelvin oval
  Numerous other possibilities for further study
ÿ^iÿ ^iÿ 1 may be found in the references [24±26] or in the
F …z† ˆ ln…z ÿ zreal † ‡ ln z ÿ
2 2 zreal current literature. As a final illustrative example,
though, the Kelvin oval is shown in Fig. 11. This
^iÿ ^iÿ
ÿ ln z ‡ ln…ÿzreal † …21† results from a vortex pair of equal magnitude and
2 2 opposite sign whose axis is perpendicular to an
466 M. Maixner

Fig. 10. Illustration of the circle theorem for a real vortex of strength ÿ ˆ 2 situated at z ˆ 1:2 ‡ 1:2^i in the presence of a cylinder of
unit radius centered on the origin.

opposing free stream, resulting in a complex poten- 2. It can be seen that this constitutes a recircula-
tial of: tion cell within the streamline = 0. In fact, if the
  vortices are brought together in such a manner
^iÿ z ÿ zupper that the product of their strength and spacing
F …z† ˆ Uo z ÿ ln …22†
2 z ÿ zlower approaches a finite value, a doublet results, just
as was obtained from a similar source/sink combi-
where zupper ˆ ^i=2, zlower = zupper, U0 = 1, and ÿ = nation. The only difference is that the vortex pair is

Fig. 11. Streamlines of Kelvin oval representing from free stream opposing the flow from two oppositely-signed vortices of magnitude
2 located at z ˆ …^i=2†.
 The Use of MathCad in Teaching Ideal Fluid Flow with Complex Variables 467

oriented perpendicular to the free stream, while the excellent choice for teaching ideal fluid flow with
source/sink combination lies on an axis parallel to complex variables. The lesson plan provided on the
the free stream. MathCad web site may be easily modified to suit
individual teaching preferences or topics. With
minimal effort, the lesson plan may be changed
CONCLUSION to provide instruction in electrostatics. The stream
function for the fluid flow problem is analogous
MathCad is an inexpensive and readily available to the potential function in the electrostatic analog;
calculation software package which, even in the likewise, the potential function  (and the equipo-
student version, contains a powerful contour plot- tentials) of the fluid flow problem find their
ting routine and the ability to perform calculations parallels in the stream function (and lines of
with complex numbers. These features make it an force) of electrostatics.

REFERENCES

1. G. J. Hokensen and R. D. W. Bowersox, Flow visualization, in Handbook of Fluid Dynamics and

Fluid Machinery, Volume II: Experimental and Computational Fluid Dynamics, ed. J. A. Schetz and
A. E. Fuhs, Wiley, New York, (1996) pp. 1019 ±1039.
2. B. Carr and V. E. Young, Videotapes and movies on fluid dynamics and fluid machines, in
Handbook of Fluid Dynamics and Fluid Machinery, Volume II: Experimental and Computa-
tional Fluid Dynamics, ed. J. A. Schetz and A. E. Fuhs, Wiley, New York, (1996) pp. 1171±
1189.
3. J. C. Bruch, Jr. and R. C. Wood, The teaching of hydrodynamics using computer generated
displays, Bull. Mech. Engng. Educ., 9, (1970) pp. 105±115.
4. D. R. Broome, Interactive graphics display of inviscid fluid flows developed using conformal
transformations, IJMEE, 6 (4), (1978) pp.191±195.
5. J. C. Bruch, Jr., The use of interactive computer graphics in the conformal mapping area,
Computers & Graphics, 1, (1975) pp. 361±374.
6. R. C. Wood and J. C. Bruch, Jr., Teaching complex variables with an interactive computer system,
IEEE Transactions on Education, V-15 (1), (Feb 1972) pp. 73±80.
7. J. Mittleman and A. K. Mitra, Exploring two-dimensional potential flow on a personal computer,
Computers in Education Journal, IV (1), (Jan±Mar 1994) pp. 70±76.
8. T. A. Fenaish, Numerical experiments on potential flow configurations, Computers in Education
Journal, II (4), (Oct±Dec 1992) pp. 17±20.
9. R. C. Ertekin and B. Padmanabhan, Graphical aid in potential flow problems: computer program
POTFLO, Computers in Education Journal, IV (4), (Oct±Dec 1994) pp. 24±31.
10. F. B. Hildebrand, Advanced Calculus for Applications, Prentice-Hall, Englewood Cliffs NJ (1962)
Ch. 10.
11. E. Kreyszig, Advanced Engineering Mathematics, 8th ed., Wiley, New York (1999) Ch. 12.
12. V. L. Streeter, E. B. Wylie and L. W. Bedford, Fluid Mechanics, 9th ed., WCB/McGraw-Hill,
Boston (1998) Ch. 8.
13. G. Magnus, U È ber die Abweichung von Geschossen [On the deviation of shots (projectiles)], Abh.
Der Berlin Akad, 1851; U È ber die Abweichung der Geschosse und Auffallende Erscheinungen bei
Rotierenden KoÈrpern, Ann. Phys., 88, 1853, p. 1.
14. J. W. Strutt, Lord Rayleigh, On the irregular flight of a tennis ball, Messenger of Math., 7 (1877);
Scientific Papers, 1, pp. 344±46.
15. W. M. Swanson, The Magnus effect: a summary of investigations to date, Journal of Basic
Engineering, Trans. Am. Soc. Mech. Eng., 83D, (September 1961) pp. 461±470.
16. The Cousteau Society, Calypso IIÐA Revolutionary, Environmentally Friendly Ship, web site http://
www.pilot.infi.net/ ~merriam/cousteau/page/calyp2.htm (March 10, 1999).
17. J. K. Harvery and F. J. Perry, Flowfield produced by trailing vortices in the vicinity of the ground,
AIAA Journal, 9, (8), (1971) pp. 1659±60.
18. Z. C. Zheng and R. L. Ash, Study of aircraft wake vortex behavior near the ground, AIAA Journal,
34 (3), (March 1996) pp. 580±589.
19. T. von KaÂrmaÂn, Berechnung der Druckverteilung an LuftschiffkoÈrpoÂrn, Abhandlungen aus dem
Aerodynamischen Institut an der Technischen Hochschule Aachen, 6, (1927) pp. 3±17. Translated
as NACA TM 574: Calculation of Pressure Distribution on Airship Hulls, by M. Miner (July
1930).
20. J. M. Robertson, Hydrodynamics in Theory and Application, Prentice-Hall, Englewood Cliffs NJ
(1965).
21. C. Y. Chow, An Introduction to Computational Fluid Mechanics, John Wiley & Sons, New York
(1979) Ch. 2.
22. J. L. Hess, Potential Flow, in Handbook of Fluid Dynamics and Fluid Machinery, Volume I:
Fundamentals of Fluid Dynamics, ed. J. A. Schetz and A. E. Fuhs, Wiley, New York, (1996)
pp. 55±56.
23. L. M. Milne-Thompson, Theoretical Hydrodynamics, 5th ed., MacMillan, London (1968).
24. H. R. Vallentine, Applied Hydrodynamics, Butterworths, London (1969).
25. J. Katz and A. Plotkin, Low-Speed Aerodynamics: From Wing Theory to Panel Methods, McGraw-
Hill, New York (1991).
26. H. Glauert, The Elements of Aerofoil and Airscrew Theory, 2nd ed., Cambridge University Press,
Cambridge (1947).
468 M. Maixner

Michael Rex Maixner graduated with distinction from the United States Naval Academy
(1972), and served as a commissioned officer in the United States Navy for 25 years; his first
12 years were spent as a line (shipboard) officer, while his remaining service was spent
strictly in engineering assignments. He received his Ocean Engineer and SMME degrees
from MIT (1977), and his Ph.D. in mechanical engineering from the Naval Postgraduate
School (1994). His military service included 2 years on the faculty at the Naval Post-
graduate School. Upon completing his military service in 1997, he accepted a position as
Associate Professor of Engineering at Maine Maritime Academy in Castine, Maine.