School of Mechanical Aerospace and Civil Engineering

Introduction

3rd Year Fluid Mechanics

Inviscid Flows

One class of flows for which a simplified form of the Navier Stokes
equation can be considered is where the viscosity can be neglected.

These are known as inviscid flows.

Such flows might arise where the Reynolds number (Re = UL/ ) is high.

Since the Reynolds number represents a ratio of convective to diffusive

influences, the limit of very high Re implies negligible diffusion.

In high-speed external flows around streamlined bodies, for example, the

flow well away from the body might be treated as inviscid (although the
approximation is not valid in the boundary layer or wake regions).

T. J. Craft
George Begg Building, C41

Contents:
Navier-Stokes equations
Inviscid flows
Boundary layers
Transition, Reynolds averaging
Mixing-length models of turbulence
Turbulent kinetic energy equation
One- and Two-equation models
Flow management

Reading:
F.M. White, Fluid Mechanics
J. Mathieu, J. Scott, An Introduction to Turbulent Flow
P.A. Libby, Introduction to Turbulence
P. Bernard, J. Wallace, Turbulent Flow: Analysis Measurement & Prediction
S.B. Pope, Turbulent Flows
D. Wilcox, Turbulence Modelling for CFD
Notes: http://cfd.mace.manchester.ac.uk/tmcfd
- People - T. Craft - Online Teaching Material

Boundary layer
Wake

Inviscid Flows

In some cases solutions can be built

up by matching the inviscid solution
(in the outer flow region) to boundary
layer approximation solutions (see
later lectures) close to walls.

2010/11

The Euler Equations

Boundary layer
Wake

If viscous effects are neglected, the governing equations are known as

the Euler equations.

In 2-D, incompressible, flow these can be written as

At the other end of the scale from inviscid flows are creeping flows
(which will not be considered in this course).

Continuity:

These occur where the Reynolds number is very small (Re << 1), and
convection terms can thus be neglected.

 2
1 P
U U 2 UV
U 2U
=
(1)
+
+
+
+
t
x
y
x
x2
y2
{z
}
|

Momentum:

U V
+
=0
x
y

(2)

U U 2 UV
1 P
+
+
=
t
x
y
x

(3)

1 P
V UV V 2
+
+
=
t
x
y
y

(4)

or (making use of continuity)

1 P
U
U
U
+U
+V
=
t
x
y
x

(5)

V
V
V
1 P
+U
+V
=
t
x
y
y
Inviscid Flows

2 / 22

2010/11

3 / 22

Inviscid Flows

(6)
2010/11

4 / 22

The Stream Function

In 2-D we can define a stream function, , such that the velocity

components are given by

U=
y

Z B
A

U dy

Z B

Contours of therefore give the streamlines

of a flow.

From equation (7),

=0
y x
x y

The gradient of is thus normal to the velocity vector, so does not

change in the direction following the fluid velocity.
5 / 22

In steady inviscid flow, subtracting / x of equation (4) (V momentum)

from / y of equation (3) (U momentum) leads to a partial differential
equation for :


(10)
(2 ) +
(2 ) = 0
y x
x y

In principle this can be solved (with appropriate boundary conditions),

although numerical solution methods are usually required.

2010/11

(9)

7 / 22

Q=1 2

=2

Note that the above definition and properties of are not restricted to
inviscid flows.

Inviscid Flows

2010/11

6 / 22

Vorticity

In a 2-D flow the vorticity, z , is given by

z =

U V

y
x

(11)

This is associated with rotation of a fluid element.

In inviscid flow there are no viscous shear stresses to deform fluid

elements. It might, therefore, seem a reasonable approximation to
assume that the flow will be irrotational, ie. z = 0.

This is an approximation that

can often be made in free
stream regions, away from
boundary layers.

Having obtained , the velocity components can then be calculated from

equation (7).

Inviscid Flows

V dx

The difference in stream function values

between two streamlines therefore gives the
total volume flow rate between the lines.

where 2 is the Laplacian operator 2 / x 2 + 2 / y 2 .

Z B

=1
(8)

2010/11

U dy

gives the total volume flow rate across a line between points A and B.

V dx

=Const

The integral
Z B

Inviscid Flows

(7)

From the expressions in equation (7), the stream function can be

evaluated by integrating along a path in the fluid:

U . =

V =
x

=Const

Hence is constant along a streamline.

Note that this definition ensures continuity is satisfied.

Inviscid Flows

Boundary layer
Wake

2010/11

8 / 22

One can show that in an incompressible, inviscid, flow the vorticity is

constant along a streamline.
Taking / y of the U momentum equation (5) gives






U
U
U U
U
V U
1 2P
+V
=
(12)
+U
+
+
t y
x y
y x
y y
y y
xy

Recall that the total derivative D/Dt represents the total rate of change as
a result of being convected with the flow.

Hence, if such a flow starts off irrotational (eg. at some far upstream
location where the flow is uniform, so z = 0) it must remain irrotational.

Since U/ x + V / y = 0 by continuity, we obtain







U
U
U
1 2P
+U
+V
=
t y
x y
y y
xy

z =
(13)

Similarly, taking / x of the V momentum equation results in







1 2P
V
V
V
+U
+V
=
t x
x x
y x
xy

or

2 2
+
x2 y2

(16)

The condition z = 0 thus implies that the stream function satisfies the
Laplace equation, 2 = 0.

(15)
2010/11

9 / 22

Velocity Potential Function

For analyzing irrotational, inviscid, flow the velocity potential function, is
often used.

Inviscid Flows

V=




and, in 3-D, W =
z

Note that this ensures z = U/ y V / x = 0.

Continuity then dictates that satisfies the Laplace equation:

Note the correct wall boundary condition for an inviscid flow is not simply
all velocity components being zero.
The wall-normal velocity component should be zero at a wall, but since
there is no viscosity the wall-tangential velocity need not be its value
arises as part of the problem solution.

From the (steady) U momentum equation, noting that V / x = U/ y in

irrotational flow, we get

2010/11

11 / 22

i
1 P
U
U
U
V
h
0.5(U 2 + V 2 )
=U
+V
=U
+V
=
x
x
y
x
x
x

(19)

Similarly, from the V momentum equation one gets

(18)

With appropriate boundary conditions, the above equation can be solved

to obtain the flowfield for a particular problem.

10 / 22

(17)

2 2
+
=0
x2 y2

Inviscid Flows

2010/11

This is defined so that

U=

(14)

D z
=0
Dt

Inviscid Flows

U V

=
y
x
y

Subtracting equation (14) from equation (13) one obtains

z
z
z
+U
+V
=0
t
x
y

From the definition of ,

i
h
1 P
0.5(U 2 + V 2 )
=
y
y

(20)

Integrating the above equations we obtain

P + 0.5 (U 2 + V 2 ) = Constant

Inviscid Flows

(21)
2010/11

12 / 22

Having obtained (and hence the velocities) from solving equation (18),
the pressure field can then be determined from equation (21).

Consider a channel flow developing between two parallel plates.

Appropriate boundary conditions are:

The use of the velocity potential for irrotational flows can easily be
extended to 3-D flows, whereas the stream function cannot.

y=0

x=0

The function = Uo x satisfies the Laplace equation, and the above

boundary conditions.
1.2

0.5

1.8
1.6
1.5

1.3

0.7

1.9

1.1

1.4

0.8

0.4

0.1

0.6

1.7

0.6

0.9

0.8

0.3

Differentiating gives the

velocity field as U = Uo and
V = 0.

0.2

0.4

Potential contour lines and

streamlines are as shown.

Without viscosity there is no boundary layer development.

0.2
0

1.8

0.9

1.2

0.6

An attraction of the above approach is that the Laplace equation is

relatively easy to solve. There are some known analytical solutions, and
well-tested and inexpensive methods for numerical solutions.

Uo

0.5

y=1

V = 0, so / y = 0, at y = 0 and 1.
U = Uo , so / x = Uo , at x = 0.
U/ x = 0, so 2 / x 2 = 0, as x .

0.2

Example 1: Plane Channel Flow

This is a special case of Bernoullis equation. The right hand side of

equation (21) is now constant throughout the flow, not simply along a
streamline.

0.5

1.5

This can be shown to be true generally:

In this case it is convenient to work in polar coordinates (r , ).

The Laplace equation becomes



1
1 2

r
+ 2
=0
r r
r
r 2

There are a family of solutions to this of the form

For simplicity, take A = 1.

0.8

n = 1 gives uniform flow.

0.6

= VU + UV

(22)
y=0

x=0

= Ar n cos(n )

for constants A and n

1

-0
-0 .3
.2

0.1 0.2

0.3
0.1

-0.2

-0.1

0.1

0.2

-0.5

0.8

0.2

0.4

0
-1

.1

.2

-0.4
-0.3

0.7

Inviscid Flows

-0

-0.5
.1

0.4
0.2

-0

-0

-0

0.5

15 / 22

.4
-0
.3

-0.8

0.6

2010/11

n = 2 gives a flow
impinging onto a flat plate,
with potential lines and
streamlines as shown.

(23)
-0.7

-0.7

-0.6

3
0.

-0.6

0.4

The gradient of is thus perpendicular to the gradient of , and it follows

that contours of must be perpendicular to contours of (streamlines).

0.6

=0

Inviscid Flows

14 / 22

Example 2: Plane Stagnation Flow

. =
+
x x y y

2010/11

0.7

Note that the streamlines and potential function contour lines are
perpendicular to each other.

Inviscid Flows

0.5

0.9

13 / 22

0.2

2010/11

0.5

Inviscid Flows

X
2010/11

16 / 22

V =

1
r

and

Vr =

(24)

The radial velocity, Vr , in this flow is thus

= 2r cos(2 )
r

Example 3: Flow Past a Cylinder

In cylindrical polar coordinates the velocity components are given by

Consider a uniform stream of velocity Uo , flowing past a circular cylinder

of radius R as shown.

Working in cylindrical coordinates, potential

flow in such a case can be represented by

(25)

At = 90o , this gives / r = 2r

So the velocity along the stagnation line decreases linearly towards
the wall.

= Uo (r + R 2 /r ) cos( )

V =

The radial and circumferential velocity components are given by

Vr =

Along the impingement wall ( = 0o ) we get / r = 2r

So the velocity along the wall increases linearly with distance from
the stagnation point.

Uo

(26)

= Uo (1 R 2 /r 2 ) cos( )
r
1
= Uo (1 + R 2 /r 2 ) sin( )
r

In the far field (large r ), this gives

Flow deflections through angles other than 90o can be obtained by taking
other values of n.

Vr = Uo cos( )

V = Uo sin( )

and

corresponding to uniform flow in the x direction.

2010/11

On the cylinder r = R, we get

Vr = 0

17 / 22

Inviscid Flows

2010/11

P + 0.5 V2 = Const = 0.5 Uo2

The potential function contours and streamlines are as shown.

18 / 22

The cylinder surface pressure can be obtained from equation (21):

V = 2Uo sin( )

and

(27)

Thus

0.5

P = 0.5 Uo2 2 Uo2 sin2 ( )

= 0.5 Uo2 (1 4 sin2 ( ))

0
-0.5

Cp

Inviscid Flows

-1

Y/R

-1.5

0
-1
-2
-4

-2

There is no flow separation, and the streamline patterns are perfectly

symmetric fore and aft of the cylinder.

Inviscid Flows

2010/11

19 / 22

-2
-2.5
-3

30

60

90

120

180

150

Theta

This example illustrates a limitation of the inviscid, irrotational,

approximations.

In reality they do not hold near the cylinder surface, where boundary
layers form, leading to flow separation under adverse pressure gradient
conditions.

X/R

The pressure is perfectly symmetric

fore and aft, and the cylinder
experiences no drag force!

Inviscid Flows

2010/11

20 / 22

Superposition of Solutions

Since the Laplace equation is linear,

solutions to it can be added together, or
superimposed.

Summary

For example, the potential function

= k , for constant k , gives a vortex
centered at the origin.

In 2-D flows the stream function can be defined to aid in flow analysis.

Inviscid flows are governed by the Euler equations.

In an inviscid, irrotational, flow the velocity potential function can be

defined.

In potential flow problems the streamlines and potential function contours

are orthogonal to each other.

Inviscid flow approximations can typically be applied to high speed flow

away from solid surfaces. Such solutions can, in some instances, be
matched to approximate boundary layer solutions for the near-wall region.

-2
-2

This could be added to the

function used in example 3,
giving flow around a
cylinder with circulation:

= Uo (r +R 2 /r ) cos( )+k

Y/R

-2
-4

-2

X/R

Inviscid Flows

2010/11

21 / 22

Inviscid Flows

2010/11

22 / 22