Complex variables applied to two-dimensional potential flows - - Prepared by Gilberto E.

Urroz, September 2001

The theory of complex variables and complex functions can be used to analyze two-dimensional potential flows by defining a complex function known as
the complex potential. This function is defined as

Ω = φ ( x, y ) − i ψ ( x , y ) ,
or
Ω = φ ( r ,θ ) − i ψ ( r ,θ ) ,

where φ( r, θ ) or φ( r, θ ) represents the velocity potential and ψ( r, θ ) or ψ( r, θ ) represents the stream function of the flow.

By using the definitions of the velocity components as derivatives of the velocity potential and stream functions, we can prove that the function

dΩ
w= ,
dz

known as the complex velocity, contains the flow velocity components. In fact, we can write:

dΩ
w= = −u + iv = ( −vr + ivθ ) e −iθ ,
dz

( i θ) ( i θ)
or u=-Re(w), v=Im(w), vr=-Re(w e ), vθ=Im(w e ).

The process of analysis consists in identifying a complex potential function Ω = F( z ), from which we get a velocity potential and a stream function by using
φ = −ℜ( w ), ψ = ℑ( w ). Next, we proceed to calculate the complex velocity w = dΩ/dz and identify the velocity components through u=ℜ( w ), v= ℑ( w ), vr=
 q 
2

−ℜ( w e
( i θ)
), vθ= ℑ( w e
( i θ) 
) . Finally, pressure distributions are calculated from Bernoulli's equation as P = ρ  K − g h − .
 2 

Example 1 - Flow around a cylinder.

Consider the complex potential, F(z), given by:
> restart:F:=z->U*z+m/z;
m
F := z → U z +
z
The velocity potential and stream functions can be calculated as:
> phi:=evalc(Re(subs(z=x+I*y,F(z))));
mx
φ := U x +
x + y2
2

> psi:=evalc(Im(subs(z=x+I*y,F(z))));
my
ψ := U y −
x +y
2 2

The complex velocity is given by:

> w:=diff(F(z),z);
m
w := U −
z2
The velocity components are:
> u:=evalc(-Re(subs(z=x+I*y,w)));

m ( x2 − y2 )
u := −U + 2
(x − y ) + 4 x y
2 2 2 2

> v:=evalc(Im(subs(z=x+I*y,w)));
mxy
v := 2 2
(x − y ) + 4 x y
2 2 2 2

The denominator in the fractions included in u and v can be simplified by using:

> factor(expand((x^2-y^2)^2+4*x^2*y^2));
2
( x2 + y2 )
Thus, the velocity components can be re-defined as:
> u:=U-m*(x^2-y^2)/(x^2+y^2)^2;
 m ( x2 − y2 )
u := U − 2
( x 2 + y2 )
> v := 2*m*x*y/(x^2+y^2)^2;
mxy
v := 2 2
( x2 + y2 )
To produce plots of the flow and velocity field we provide specific values to U and m. For example, U = 1.5, m = 1:
> with(plots):
Warning, the name changecoords has been redefined

> equipot:=contourplot(subs({U=1.5,m=1},phi),x=-2..2,y=-2..2,contours=20,color=blue,scaling=constrained):
> stream:=contourplot(subs({U=1.5,m=1},psi),x=-2..2,y=-2..2,contours=20,color=red,scaling=constrained):
> display({equipot,stream});

We recognize the flow net for the flow around a cylinder without circulation. To determine the radius of the circle we look at the streamline ψ = 0 by
using:
> simplify(psi=0);

y (U x + U y − m )
2 2

=0
x +y
2 2
 m m
This is equivalent to the equation x2 + y2 = , i.e., the equation of a circle of radius a = .
U U
The pressure distribution in the flow can be obtained by using Bernoulli's equation:
> P:=rho*(K-g*h-(u^2+v^2)/2);
 2 
 1  m ( x2 − y2 )  2 m2 x2 y2 
P := ρ  K − g h −  U − 2 
− 4

 2  
  (x + y )  (x + y ) 
2 2 2 2

A stagnation point is found at infinity where we can make P = P0, u = 0, v = 0, thus P0 = ρ ( K − g h ), and we can re-write P as:
> P:=P0-rho*(u^2+v^2)/2;
 2
2 2 2
 m ( x − y )  4 m x y 
2 2
1
P := P0 − ρ   U − 2 
+ 
4
2   
 (x + y )  (x + y ) 
2 2 2 2

To determine the pressure distribution on the cylinder it is more convenient to work with polar coordinates, thus:
> Pp:=simplify(subs({x=r*cos(theta),y=r*sin(theta)},P));

1 4 cos ( θ )2 ρ U r2 m − ρ U 2 r4 − ρ m2 + 2 P0 r4 − 2 ρ U r2 m
Pp := 2
2
r4 ( cos ( θ )2 + sin ( θ )2 )
Here we replace the trigonometry identity cos ( θ ) + sin ( θ ) = 1 to get:
2 2

> Pp:=subs(cos(theta)^2+sin(theta)^2=1,Pp);

1 4 cos ( θ )2 ρ U r2 m − ρ U 2 r4 − ρ m2 + 2 P0 r4 − 2 ρ U r2 m
Pp :=
2 r4
m
The pressure distribution on the cylinder is obtained by replacing r = a = into Pp:
U
> Pc:=simplify(subs(r=sqrt(m/U),Pp));

Pc := 2 cos ( θ )2 ρ U2 − 2 ρ U2 + P0

Example 2 - Flow pattern for F(z) = z

This example shows the flow patter generated from the complex potential F( z ) = z .
> restart:F:=z->sqrt(z);phi:=evalc(Re(subs(z=x+I*y,F(z))));psi:=evalc(-Im(subs(z=x+I*y,F(z))));
F := sqrt
 1
φ := 2 x2 + y2 + 2 x
2
1
ψ := − csgn ( y − I x ) 2 x2 + y2 − 2 x
2
> with(plots):
Warning, the name changecoords has been redefined

> plot1:=contourplot(phi,x=-1..1,y=-1..1, contours=10, color = blue):

> plot2:=contourplot(psi,x=-1..1,y=-1..1, contours = 10, color = red):
> display({plot1,plot2},axes = boxed, scaling=constrained);

To obtain velocity potential and stream functions in polar coordinates, we use:

> restart:F:=z->sqrt(z);phi:=simplify(evalc(Re(subs(z=r*exp(I*theta),F(z)))));psi:=simplify(evalc(-Im(subs(z=r*exp(I*theta),F(z)))));
F := sqrt
1
φ := 2 r ( csgn ( r ) + cos ( θ ) )
2
1
ψ := csgn ( r ( −sin ( θ ) + I cos ( θ ) ) ) 2 r ( csgn ( r ) − cos ( θ ) )
2
> with(plots):
Warning, the name changecoords has been redefined
 > plot1:=contourplot(phi,r=0..2,theta=-Pi..Pi, contours=10, color = blue,scaling=constrained):
> plot2:=contourplot(psi,r=0..2,theta=-Pi..Pi, contours = 10, color = red,scaling=constrained):
> display({changecoords(plot1,polar),changecoords(plot2,polar)});

Analysis of basic two-dimensional potential flows using complex functions

In this section we analyze the flow generated by simple complex functions. We start with rectilinear flows.

Rectilinear flows
These are flows whose complex potential has the form F(z) = Ω = A zn, where A and n are constant. A can be a real or complex number, while n is integer.
The following are the general expressions for the velocity potential and stream functions:
> restart:assume(n,integer);F:=z->A*z^n;

F := z → A z
n

> phi:=simplify(evalc(Re(subs(z=r*exp(I*theta),F(z)))));
( 1 / 2 n~ )
φ := A ( r2 ) cos ( n~ arctan( r sin ( θ ), r cos ( θ ) ) )
> psi:=simplify(evalc(Im(subs(z=r*exp(I*theta),F(z)))));
( 1 / 2 n~ )
ψ := A ( r2 ) sin ( n~ arctan( r sin ( θ ), r cos ( θ ) ) )
 n
 
2
Here we recognize the fact that ( r ) 2
= rn, and arctan( r sin ( θ ), r cos ( θ ) ) = θ, thus, we re-write our functions as:
> phi:=-A*r^n*cos(theta);psi:=A*r^n*sin(theta);

φ := −A rn~ cos ( θ )
ψ := A r sin ( θ )
n~

The complex velocity is:

> w:=diff(F(z),z);
n~
A z n~
w :=
z
From which,
> u:=simplify(evalc(subs({z=r*exp(I*theta)},-Re(w))));
( 1 / 2 n~ )
( cos ( n~ arctan( r sin ( θ ), r cos ( θ ) ) ) cos ( θ ) + sin ( n~ arctan( r sin ( θ ), r cos ( θ ) ) ) sin ( θ ) )
2
n~ A ( r )
u := −
r
> v:=simplify(evalc(subs({z=r*exp(I*theta)},Im(w))));
( 1 / 2 n~ )
( sin ( n~ arctan( r sin ( θ ), r cos ( θ ) ) ) cos ( θ ) − cos ( n~ arctan( r sin ( θ ), r cos ( θ ) ) ) sin ( θ ) )
2
n~ A ( r )
v :=
r
n
 
2
= r , and arctan( r sin ( θ ), r cos ( θ ) ) = θ, we can simplify these results to:
2 n
Again, using ( r )
> u:=-A*r^(n-1)*(cos(n*theta)*cos(theta)+sin(n*theta)*sin(theta));
( n~ − 1 )
u := −A r ( cos ( n~ θ ) cos ( θ ) + sin ( n~ θ ) sin ( θ ) )
> v:=A*r^(n-1)*(-sin(n*theta)*cos(theta)+cos(n*theta)*sin(theta));
( n~ − 1 )
v := A r ( −sin ( n~ θ ) cos ( θ ) + cos ( n~ θ ) sin ( θ ) )
Which can in turn be simplified to:
> u:=-A*r^(n-1)*cos((n-1)*theta);
( n~ − 1 )
u := −A r cos ( ( n~ − 1 ) θ )
> v:=A*r^(n-1)*sin((n-1)*theta);
( n~ − 1 )
v := A r sin ( ( n~ − 1 ) θ )
Now we consider different types of values of A and n and check what type of flows are produced.
* Case 1: A is real, n = 1
Consider the case in which A is a real number and n = 1.
> restart:assume(A,real);F:=z->A*z;
F := z → A z
> phi:=evalc(subs(z=r*exp(I*theta),Re(F(z))));
φ := A~ r cos ( θ )
> psi:=evalc(subs(z=r*exp(I*theta),-Im(F(z))));
ψ := −A~ r sin ( θ )
We recognize that the velocity potential and stream functions can be also written as φ = −A x and ψ = A y, which correspond to a uniform flow.

*Case 2: A is real, n = 2
Consider the case in which A is a real number and n = 2.
> restart:assume(A,real):F:=z->A*z^2;

F := z → A z
2

> phi:=evalc(subs(z=r*exp(I*theta),Re(F(z))));

φ := A~ r ( cos ( θ ) − sin ( θ ) )
2 2 2

> psi:=evalc(subs(z=r*exp(I*theta),-Im(F(z))));

ψ := −2 A~ r cos ( θ ) sin ( θ )
2

The velocity potential and stream functions can also be written, for this case, as φ = −r2 cos ( 2 θ ), and ψ = r2 sin ( 2 θ ).
A portrait of the flow can be shown by using the following statements. Assume that A = 2:
> with(plots):
Warning, the name changecoords has been redefined

> p1:=contourplot(subs(A=2,phi),r=0..2,theta=-Pi..Pi,color=blue,scaling=constrained):
> p2:=contourplot(subs(A=2,psi),r=0..2,theta=-Pi..Pi,color=red,scaling=constrained):
> display({changecoords(p1,polar),changecoords(p2,polar)});
This portrait can be found also by using Cartesian coordinates as follows:
> restart:assume(A,real):F:=z->A*z^2;

F := z → A z2
> phi:=evalc(subs(z=x+I*y,Re(F(z))));

φ := A~ ( x2 − y2 )
> psi:=evalc(subs(z=x+I*y,-Im(F(z))));
ψ := −2 A~ x y
Assume that A = 2:
> with(plots):
Warning, the name changecoords has been redefined

> p1:=contourplot(subs(A=2,phi),x=-2..2,y=-2..2,color=blue,contours=15,scaling=constrained):
> p2:=contourplot(subs(A=2,psi),x=-2..2,y=-2..2,color=red,contours=15,scaling=constrained):
> display({p1,p2});
The red lines represent streamlines suggesting a flow concentrated in a quadrant of the plane. In more detail, this will be:
> p1:=contourplot(subs(A=2,phi),x=0..2,y=0..2,color=blue,contours=15,scaling=constrained):
> p2:=contourplot(subs(A=2,psi),x=0..2,y=0..2,color=red,contours=15,scaling=constrained):
> display({p1,p2});

To find the direction of the flow we can plot the velocity vectors:
> w:=diff(F(z),z);u:=evalc(subs(z=x+I*y,-Re(w)));v:=evalc(subs(z=x+I*y,Im(w)));
 w := 2 A~ z
u := −2 A~ x
v := 2 A~ y
> fieldplot(subs(A=2,[u,v]),x=0..2,y=0..2,arrows=thick,scaling=constrained,grid=[10,10]);

The field plot of velocity vectors indicate that the flow starts towards the right-hand side of the x-plane and bends upward towards the upper part of the y
axis. This represents a flow constrained by a corner, and we will refer to it as a corner flow. Other corner flows can be obtained by using a real value of A
and different values of n in the complex potential Ω = A z .
n

 π
 
α
* Case 3: corner flow, Ω = A z
π
, where α is a real number. It can be proven that the proposed complex potential represents
In this case we use A as a real number, and replace n with
α
π 2
flow around a corner with an angle . For example, consider the case α = .
α 3
> restart:assume(A,real):F:=z->A*z^(2/3);
( 2 / 3)
F := z → A z
> phi:=simplify(evalc(subs(z=r*exp(I*theta),Re(F(z)))));
( 1 / 3) 2 
φ := A~ ( r2 ) cos  arctan( r sin ( θ ), r cos ( θ ) ) 
3 
> psi:=simplify(evalc(subs(z=r*exp(I*theta),-Im(F(z)))));
 ( 1 / 3) 2 
ψ := −A~ ( r2 ) sin  arctan( r sin ( θ ), r cos ( θ ) ) 
3 
We can simplify these expressions to read:
> phi:=A*r^(2/3)*cos((2/3)*theta);
( 2 / 3) 2 
φ := A~ r cos  θ 
3 
> psi:=-A*r^(2/3)*sin((2/3)*theta);
( 2 / 3) 2 
ψ := −A~ r sin  θ 
3 
Assume that A = 2:
> with(plots):
Warning, the name changecoords has been redefined

> p1:=contourplot(subs(A=2,phi),r=0..2,theta=0..2*Pi,color=blue,scaling=constrained):
> p2:=contourplot(subs(A=2,psi),r=0..2,theta=0..2*Pi,color=red,scaling=constrained):
> display({changecoords(p1,polar),changecoords(p2,polar)});

3π
The red lines represent streamlines suggesting a flow around a corner of angle . In more detail, this will be:
2
3
π
2
 > p1:=contourplot(subs(A=2,phi),r=0..2,theta=0..3*Pi/2,color=blue,scaling=constrained):
> p2:=contourplot(subs(A=2,psi),r=0..2,theta=0..3*Pi/2,color=red,scaling=constrained):
> display({changecoords(p1,polar),changecoords(p2,polar)});

To find the direction of the flow we can plot the velocity vectors:
> w:=diff(F(z),z);u:=evalc(subs(z=r*exp(I*theta),-Re(w)));v:=evalc(subs(z=r*exp(I*theta),Im(w)));
2 A~
w :=
3 z( 1 / 3 )

1 
A~ cos  arctan( r sin ( θ ), r cos ( θ ) ) 
2 3 
u := − ( 1 / 6)
3
( r 2 cos ( θ ) 2 + r2 sin ( θ )2 )
1 
A~ sin  arctan( r sin ( θ ), r cos ( θ ) ) 
2 3 
v := − ( 1 / 6)
3
( r cos ( θ ) + r sin ( θ ) )
2 2 2 2

Simplifying the expressions we write:

> u:=-(2/3)*cos(theta/3)/r^(1/3);v:=-(2/3)*sin(theta/3)/r^(1/3);
1 
cos  θ 
2 3 
u := −
3 r 1 / 3)
(
 1 
sin  θ 
2 3 
v := −
3 r 1 / 3)
(

 π 
 
 π− β 
* Case 4. Wedge flow, Ω = A z
π
This is the flow about a wedge with an angle α = π − β. For example, consider the case β = .
3
> restart:assume(A,real):F:=z->A*z^(Pi/(Pi-Pi/3));
( 3 / 2)
F := z → A z

> phi:=simplify(evalc(subs(z=r*exp(I*theta),Re(F(z)))));
1
φ := − A~ 2 r ( csgn ( r ) + cos ( θ ) ) r ( −2 cos ( θ ) + csgn ( r ) )
2
> psi:=simplify(evalc(subs(z=r*exp(I*theta),-Im(F(z)))));
1
ψ := A~ csgn ( r ( −sin ( θ ) + I cos ( θ ) ) ) 2 r ( csgn ( r ) − cos ( θ ) ) r ( csgn ( r ) + 2 cos ( θ ) )
2
Assume that A = 2:
> with(plots):
Warning, the name changecoords has been redefined

> p1:=contourplot(subs(A=2,phi),r=0..2,theta=0..2*Pi,color=blue,scaling=constrained):
> p2:=contourplot(subs(A=2,psi),r=0..2,theta=0..2*Pi,color=red,scaling=constrained):
> display({changecoords(p1,polar),changecoords(p2,polar)});
 π
The red lines represent streamlines suggesting a flow around a wedge of angle . In more detail, this will be:
3
3
π
2
> p1:=contourplot(subs(A=2,phi),r=0..2,theta=0..Pi-Pi/3,color=blue,scaling=constrained):
> p2:=contourplot(subs(A=2,psi),r=0..2,theta=0..Pi-Pi/3,color=red,scaling=constrained):
> p3:=implicitplot(theta=2*Pi/3,r=0..2,theta=0..Pi-Pi/3,color=black,scaling=constrained):#Draws the left side of wedge
> display({changecoords(p1,polar),changecoords(p2,polar),changecoords(p3,polar)});
* Case 5. A is complex. Uniform flow.
( −i α ) ( −i α )
Consider the case in which A = K e , thus Ω = K e z.

> restart:assume(K,real);assume(alpha,real);
> F:=z->K*exp(-I*alpha)*z;
( −I α )
F := z → K e z
> phi:=evalc(subs(z=r*exp(I*theta),Re(F(z))));
φ := K~ ( r cos ( θ ) cos ( α∼ ) + r sin ( θ ) sin ( α∼ ) )
> psi:=evalc(subs(z=r*exp(I*theta),-Im(F(z))));
ψ := −K~ ( r sin ( θ ) cos ( α∼ ) − r cos ( θ ) sin ( α∼ ) )
π
Assume that K = 2, and α =
6
1
α∼ = π
6
> with(plots):
Warning, the name changecoords has been redefined

> p1:=contourplot(subs({K=2,alpha=Pi/6},phi),r=0..2,theta=0..2*Pi,color=blue,scaling=constrained):
> p2:=contourplot(subs({K=2,alpha=Pi/6},psi),r=0..2,theta=0..2*Pi,color=red,scaling=constrained):
> display({changecoords(p1,polar),changecoords(p2,polar)});
 ( −i α )
Which indicates that Ω = K e z, represents a uniform flow at an angle α with respect to the positive x-axis. So, if α = 0, we recover the complex
potential for a uniform flow in the x-direction: Ω = K z, where the velocity is U = -K.

* Case 6. Uniform flow in x, Ω = −U z

> restart:assume(U,real);
> F:=z->-U*z;
F := z → −U z
> phi:=evalc(subs(z=r*exp(I*theta),Re(F(z))));
φ := −U~ r cos ( θ )
> psi:=evalc(subs(z=r*exp(I*theta),-Im(F(z))));
ψ := U~ r sin ( θ )
> with(plots):
Warning, the name changecoords has been redefined

> p1:=contourplot(subs(U=1,phi),r=0..2,theta=0..2*Pi,color=blue,scaling=constrained):
> p2:=contourplot(subs(U=1,psi),r=0..2,theta=0..2*Pi,color=red,scaling=constrained):
> display({changecoords(p1,polar),changecoords(p2,polar)});
Radial and swirling flows
These flows are associated with complex potentials of the form z = A ln z.

* Case 1 - A is real
Let's analyze the case in which A is a real constant. The complex potential is:
> restart:assume(A,real);F:=z->A*log(z);
F := z → A log( z )
The corresponding velocity potential is obtained as:
> phi:=simplify(evalc(subs(z=r*exp(I*theta),Re(F(z)))));
φ := A~ ln ( csgn ( r ) r )
Since the function csgn(r) is either equal to 1.0 or to -1.0, we can write:
> phi:=A*log(r);
φ := A~ ln ( r )
The stream function is given by:
> psi:=evalc(subs(z=r*exp(I*theta),-Im(F(z))));
ψ := −A~ arctan( r sin ( θ ), r cos ( θ ) )
which simplifies to
> psi:=-A*theta;
ψ := −A~ θ
A picture of this flow is shown next for A = 1:
> with(plots):
> p1:=contourplot(subs(A=1,phi),r=0..2,theta=0..2*Pi,color=blue,scaling=constrained):
> p2:=contourplot(subs(A=1,psi),r=0..2,theta=0..2*Pi,color=red,scaling=constrained):
> display({changecoords(p1,polar),changecoords(p2,polar)});

We recognize this flow net as that of a source or sink located at the origin. The singularity flow is a source if A = −µs< 0 or a sink if A = −µs > 0. The
complex potential is, thus, Ω = −µs ln ( z ). If the source is located at a point z = z0, then the complex potential is written as Ω = −µs ln ( z − z0 ).

*Case 2 - A is imaginary
Consider the case in which the coefficient A is imaginary, say A = iB. The complex potential is:
> restart:assume(B,real);F:=z->I*B*log(z);
F := z → I B log( z )
The corresponding velocity potential is obtained as:
> phi:=simplify(evalc(subs(z=r*exp(I*theta),Re(F(z)))));
 φ := −B~ arctan( r sin ( θ ), r cos ( θ ) )
which simplifies to
> phi:=-B*theta;
φ := −B~ θ
The stream function is given by:
> psi:=evalc(subs(z=r*exp(I*theta),-Im(F(z))));

ψ := −B~ ln ( r2 cos ( θ )2 + r2 sin ( θ )2 )

which simplifies to:
> psi:=-B*ln(r);
ψ := −B~ ln ( r )
A picture of this flow is shown next for B = 1:
> with(plots):
Warning, the name changecoords has been redefined

> p1:=contourplot(subs(B=1,phi),r=0..2,theta=0..2*Pi,color=blue,scaling=constrained):
> p2:=contourplot(subs(B=1,psi),r=0..2,theta=0..2*Pi,color=red,scaling=constrained):
> display({changecoords(p1,polar),changecoords(p2,polar)});

We recognize this flow net as that of a vortex, of strength B, located at the origin. The vortex is counterclockwise if B = µv >0 or a sink if B = µv <0.
 A
Doublet flow - Ω =
z
Consider the case in which A is a real constant and the velocity potential is given by:
> restart:assume(A,real);F:=z->A/z;
A
F := z →
z
The corresponding velocity potential is obtained as:
> phi:=simplify(evalc(subs(z=r*exp(I*theta),Re(F(z)))));

A~ cos ( θ )
φ :=
r
The stream function is given by:
> psi:=evalc(subs(z=r*exp(I*theta),-Im(F(z))));
A~ sin ( θ )
ψ :=
r ( cos ( θ )2 + sin ( θ )2 )
which simplifies to:
> psi:=A*sin(theta)/r;
A~ sin ( θ )
ψ :=
r
A cos ( θ )
A picture of this flow is shown by recognizing that lines of constant φ are given by the polar curve r = , while those lines of constant ψ (i.e., the
φ
A sin ( θ ) φ
streamlines) are given by r = . The following commands produce the flow net. Here we use K = .
φ A
> with(plots):
> K:=[-5,-4,-3,-2,-1,1,2,3,4,5];
K := [ -5, -4, -3, -2, -1, 1, 2, 3, 4, 5 ]
> p1:=plot({seq(cos(theta)/K[j],j=1..nops(K))},theta = 0..2*Pi,coords=polar,scaling=constrained,color=blue):
> p2:=plot({seq(sin(theta)/K[j],j=1..nops(K))},theta = 0..2*Pi,coords=polar,scaling=constrained,color=red):
> display({p1,p2});
We recognize the flow net for a doublet of strength A = µd. The double strength µd is positive if the doublet -- aligned with the x axis -- is such that flow
leaves the doublet on the positive side of x and it returns to the doublet on the negative side of x. This is indicated in the following sketch with the

symbol (-+) located at the origin as illustrated in the following drawing.

If the outflow occurs on the negative side of x and the inflow in the positive side of x, the strength of the doubled µd is negative.
From the figure we can observe that both the equipotential lines and the streamlines are circles. This result can be obtained from the expressions for the
equipotential lines and stream functions, namely:
> EqPhi:=r=mu[d]*cos(theta)/_phi;EqPsi:=r=mu[d]*sin(theta)/_psi;
µd cos ( θ )
EqPhi := r =
_φ
µd sin ( θ )
EqPsi := r =
_ψ
Here we use _φ and _ψ, rather than φ and ψ, to emphasize constant values of these functions. Let's now manipulate these equations:
> EqPhi:=r*EqPhi;EqPsi:=r*EqPsi;
r µd cos ( θ )
EqPhi := r =
2

_φ
r µd sin ( θ )
EqPsi := r =
2

_ψ
and make the following replacements:
> EqPhi:=subs({r^2=x^2+y^2,cos(theta)=x/r},EqPhi);EqPsi:=subs({r^2=x^2+y^2,sin(theta)=y/r},EqPsi);
µd x
EqPhi := x + y =
2 2

_φ
µd y
EqPsi := x + y =
2 2

_ψ
These equations are re-written as:
> EqPhi:=lhs(EqPhi)-rhs(EqPhi)=0;EqPsi:=lhs(EqPsi)-rhs(EqPsi)=0;
µd x
EqPhi := x + y − =0
2 2

_φ
µd y
EqPsi := x + y − =0
2 2

_ψ
In these expressions we used functions lhs (left-hand side) and rhs (right-hand side) to manipulate the equations.
Next, we will use function completesquare from the student package to re-write our equations as those of a circle:
> with(student):
> EqPhi:=completesquare(x^2-mu[d]*x/_phi,x)+y^2 = 0;
 2 2
 1 µd  1 µd
EqPhi :=  x −  − +y =0
2

 2 _φ  4 _φ2
> EqPsi:=x^2+completesquare(y^2-mu[d]*y/_psi,y)= 0;
2 2
 1 µd  1 µd
EqPsi :=  y −  − + x2 = 0
 2 _ψ  4 _ψ2
These equations can be re-written as:
> EqPhi := (x-1/2*mu[d]/_phi)^2+y^2 = 1/4*mu[d]^2/(_phi^2);
2 2
 1 µd  1 µd
EqPhi :=  x −  +y =
2

 2 _φ  4 _φ2
> EqPsi := (y-1/2*mu[d]/_psi)^2+x^2 = -1/4*mu[d]^2/(_psi^2);
2 2
 1 µd  1 µd
EqPsi :=  y −  +x =−
2

 2 _ψ  4 _ψ2
µd µd
We recognize equation EqPhi as that of a circle of radius a = , centered at C(
, 0), i.e., circles centered on the x-axis. Also, we recognize equation
2φ 2φ
µd µd
EqPsi as that of a circle of radius a = , centered at C(0, ), i.e, circles centered on the y-axis.
2ψ 2ψ

Summary of complex potentials for simple flows

Here we summarize the diverse complex potentials presented above:

 π
 
α
* Corner flow of angle α, Ω = A z , A is real.

 π 
 
 π− β 
* Wedge flow of angle β, Ω = A z , A is real.

( −i α )
* Uniform flow of velocity U at angle α, Ω = −U e z.
* Source (0 < µs) or sink (µs < 0) at the origin, Ω = −µs ln ( z ).

* Source (0 < µs) or sink (µs < 0) at point z = z0, Ω = −µs ( ln ( z ) − z0 ).

* Vortex of strenght µv , at the origin µv , Ω = i µv ln ( z ), with 0 < µv for counterclockwise vortex, and µv < 0 for clockwise vortex.

* Vortex of strenght µv , at z = z0, Ω = i µv ln ( z − z0 ), with 0 < µv for counterclockwise vortex, and µv < 0 for clockwise vortex.

µd
* Doublet of strength µd at the origin oriented along the x-axis, Ω = , with 0 < µd when source is on the positive side of the x axis (- +).
z

µd
* Doublet of strength µd at z = z0 the origin oriented along the x-axis, Ω = , with 0 < µd when source is on the positive side of the x axis (- +).
z − z0

Combination of simple flows

The combination of a uniform flow in the positive x direction, i.e., Ω U = −U z, with singularity flows can produce the flow around a number of simple solid
bodies. Some of these flows are presented next.

Rankine half-body
This flow results from combining a uniform flow (Ω U = −U z) with a source located at the origin (Ω S = −µs log( z )). The resulting complex potential is:
> restart:F:=z->-U*z-mu[s]*log(z);
F := z → −U z − µs log( z )
with velocity potential and stream functions given by:
 > phi:=evalc(subs(z=r*exp(I*theta),Re(F(z))));
1
φ := −U r cos ( θ ) − µs ln ( r2 cos ( θ )2 + r2 sin ( θ )2 )
2
> psi:=evalc(subs(z=r*exp(I*theta),-Im(F(z))));
ψ := U r sin ( θ ) + µs arctan( r sin ( θ ), r cos ( θ ) )
simplified to:
> phi:=-U*r*cos(theta)-mu[s]*ln(r);
φ := −U r cos ( θ ) − µs ln ( r )
> psi:=U*r*sin(theta)+mu[s]*theta;
ψ := U r sin ( θ ) + µs θ
The velocity components in polar coordinates can be obtained from:
> w:=diff(F(z),z);
µs
w := −U −
z
These components are:
> v_r:=evalc(subs(z=r*exp(I*theta),-Re(w)));
µs cos ( θ )
v_r := U +
r ( cos ( θ ) + sin ( θ ) )
2 2

> v_theta:=evalc(subs(z=r*exp(I*theta),Im(w)));
µs sin ( θ )
v_theta :=
r ( cos ( θ )2 + sin ( θ )2 )
which simplify to:
> v_r:=U+mu[s]*cos(theta)/r;
µs cos ( θ )
v_r := U +
r
> v_theta:=mu[s]*sin(theta)/r;
µs sin ( θ )
v_theta :=
r
A stagnation point is found along the line θ = π. Along this line
> subs(theta=Pi,v_theta);evalf(%);
µs sin ( π )
r
0.
The stagnation point corresponds to v_r = 0, i.e.,
> Eq_v_r:=subs(theta=Pi,v_r);
µs cos ( π)
Eq_v_r := U +
r
Solving for r=a
> a:=solve(Eq_v_r,r);
µs
a :=
U
 µs 
Thus, the stagnation point occurs at ( r, θ ) =  , π . The streamline passing through this point evaluates to:
U 
> psi0 := subs({r = mu[s]/U, theta = Pi},psi);psi0:=simplify(psi0);
ψ0 := µs sin ( π ) + µs π
ψ0 := µs π
The equation of this streamline is then given by:
> Eq_psi0:=psi0=psi;
Eq_psi0 := µs π = U r sin ( θ ) + µs θ
The function describing this streamline, in polar coordinates, is given by:
> solve(Eq_psi0,r);
µs ( −π + θ )
−
U sin ( θ )
A plot of this function is achieved by using the following statement. We use values of U=1 and µs = 1.
> with(plots):
> p1:=plot(subs({mu[s]=1,U=1},-mu[s]*(-Pi+theta)/(U*sin(theta))),theta=0.1*Pi..0.99*Pi,coords=polar,color=black,thickness=2):
> p2:=plot(subs({mu[s]=1,U=1},-mu[s]*(-Pi+theta)/(U*sin(theta))),theta=1.01*Pi..1.9*Pi,coords=polar,color=black,thickness=2):
 > display({p1,p2});

The equations of other streamlines will be given by:

> EqPsi:=psi=K;
EqPsi := U r sin ( θ ) + µs θ = K
Corresponding to the following function in polar coordinates:
> solve(EqPsi,r);
µs θ − K
−
U sin ( θ )
Plots of other streamlines are given next:
> KK:=[seq(j*Pi/10,j=1..9),seq(j*Pi/10,j=11..19)];
 1 1 3 2 1 3 7 4 9 11 6 13 7 3 8 17 9 19 
KK :=  π, π, π, π, π, π, π, π, π, π, π, π, π, π, π, π, π, π
 10 5 10 5 2 5 10 5 10 10 5 10 5 2 5 10 5 10 
> pA:=plot({seq(subs({mu[s]=1,U=1,K=KK[j]},-(mu[s]*theta-K)/(U*sin(theta))),j=1..nops(KK))},theta = 0.1*Pi..0.9*Pi,coords=polar, color = red):
> pB:=plot({seq(subs({mu[s]=1,U=1,K=KK[j]},-(mu[s]*theta-K)/(U*sin(theta))),j=1..nops(KK))},theta = 1.1*Pi..1.9*Pi,coords=polar, color = red):
> display({p1,p2,pA,pB});
Rankine oval
The Rankine oval results from combining a uniform flow with a source at x=-a and a sink at x=a. Thus, the complex potential is given by:
> restart:Omega:=-U*z-mu[s]*log(z+a)+mu[s]*log(z-a);
Ω := −U z − µs ln ( z + a ) + µs ln ( z − a )
The velocity potential is:
> phi:=evalc(subs(z=x+I*y,Re(Omega)));
1 1
φ := −U x − µs ln ( ( x + a )2 + y2 ) + µs ln ( ( x − a )2 + y2 )
2 2
while the stream function is given by:
> psi:=evalc(subs(z=x+I*y,-Im(Omega)));
ψ := U y + µs arctan( y, x + a ) − µs arctan( y, x − a )
A sketch of the flow net, for U=1, a=1, and µs = 1, follows:
> with(plots):
Warning, the name changecoords has been redefined

> p1:=contourplot(subs({U=1,mu[s]=1,a=1},phi),x=-4..4,y=-4..4,color=blue,contours=10):
> p2:=contourplot(subs({U=1,mu[s]=1,a=1},psi),x=-4..4,y=-4..4,color=red,contours=10):
> display({p1,p2});
To determine the equation for the body streamline, we use the equation of the streamlines:
> psi=K;
U y + µs arctan( y, x + a ) − µs arctan( y, x − a ) = K
 y   y 
Suppose that we call θ1 = arctan , and θ2 = arctan
 
, then, the equation of the streamlines is:

 µs   µs 
x+  x− 
 U  U 
> EqPsi:=y*U+mu[s]*(theta[1]-theta[2])=K;
EqPsi := y U + µs ( θ1 − θ2 ) = K
From which we can write:
> EqPsi:=theta[1]-theta[2]=(K-y*U)/mu[s];
K−yU
EqPsi := θ1 − θ2 =
µs
Taking the tangent of this expression,
> EqPsi:=tan(theta[1]-theta[2])=tan((K-y*U)/mu[s]);
K−yU
EqPsi := tan( θ1 − θ2 ) = tan 

 µs 
with the expansion
> expand(tan(theta[1]-theta[2]));
 −tan( θ2 ) + tan( θ1 )
1 + tan( θ2 ) tan( θ1 )
y y
And since tan( θ1 ) = , and tan( θ2 ) = ,
µs µs
x+ x−
U U
> EqPsi:=simplify(subs({tan(theta[1]) = y/(x+mu[s]/U),tan(theta[2]) = y/(x-mu[s]/U)},(-tan(theta[2])+tan(theta[1]))/(1+tan(theta[2])*tan(theta[1]))))=
tan((K-y*U)/mu[s]);
y U µs K−yU
EqPsi := −2 = tan 

 µs 
2
U x − µs + y U
2 2 2 2

This equation can be re-written as

> EqPsi:=(U^2*x^2-mu[s]^2+y^2*U^2)*tan((K-y*U)/mu[s])=-2*y*U*mu[s];
K−yU
EqPsi := ( U2 x2 − µs + y2 U2 ) tan  = −2 y U µs
2

 µs 
or,
> EqPsi := (U^2*x^2-mu[s]^2+y^2*U^2)*tan((K-y*U)/mu[s]) +2*y*U*mu[s]=0;
K−yU
EqPsi := ( U2 x2 − µs + y2 U2 ) tan  + 2 y U µs = 0
2

 µs 
A plot of the equation for K = 0, gives:
> EqPsi0:=subs(K=0,EqPsi);
 yU
EqPsi0 := ( U2 x2 − µs + y2 U2 ) tan −  + 2 y U µs = 0
2

 µs 
> p0:=implicitplot(subs({U=1,mu[s]=1},(U^2*x^2-mu[s]^2+y^2*U^2)*tan(-y*U/mu[s])+2*y*U*mu[s] =
0),x=-1.8..1.8,y=-1.5..1.5,color=black,thickness=4):
> display({p1,p2,p0});
Kelvin oval
Γ
A Kelvin oval is obtained by combining two vortices, a clockwise vortex at y=a and a counterclockwise vortex at y = -a, both of strenght µv = , with a
2π
uniform flow of magnitude Γ = 4 π a U. The relationship between U and Γ results from the fact that, in order to keep the vortex from moving from their
required positions, the velocity U must equal that velocity produced by each of the vortices in the location of the other vortex. Since the transverse
µs Γ
velocity for a vortex is given by vθ = , and the vortices are separated by a distance r = 2a, the velocity of each vortex is vvortex = . Making this result
r 4 πa
equal to U, results in Γ = 4 π a U, or µv = 2 a U.
The combined complex potential is given by
> restart:Omega:=-U*z+I*2*a*U*log(z+I*a)-I*2*a*U*log(z-I*a);
Ω := −U z + 2 I a U ln ( z + I a ) − 2 I a U ln ( z − I a )
with velocity potential:
> phi:=evalc(subs(z=x+I*y,Re(Omega)));
φ := −U x − 2 a U arctan( y + a , x ) + 2 a U arctan( y − a , x )
and streamfunction:
> psi:=evalc(subs(z=x+I*y,-Im(Omega)));

ψ := U y − a U ln ( x2 + ( y + a )2 ) + a U ln ( x2 + ( y − a )2 )
The streamlines can be obtained by using:
> contourplot(subs({a=1,U=1},psi),x=-3..3,y=-3..3,contours=20,scaling=constrained,color = red);
The velocity components are calculated from:
> w:=diff(Omega,z);
2 I a U -2 I a U
w := −U + +
z+Ia z−Ia
These components are:
> u:=evalc(subs(z=x+I*y,-Re(w)));
2 a U (y + a) 2 a U (y − a)
u := U − +
x + (y + a) x + (y − a)
2 2 2 2

> v:=evalc(subs(z=x+I*y,Im(w)));
a Ux 2a Ux
v := 2 −
x + (y + a)
2
x + ( y − a )2
2 2

We expect the stagnation point to occur somewhere along the x axis, i.e., at y=0:
> u0:=subs(y=0,u);v0:=subs(y=0,v);

4 a2 U
u0 := U −
x2 + a 2
v0 := 0
The stagnation points occur where u0 = 0, i.e.,
> solve(u0=0,x);

3 a, − 3 a