This action might not be possible to undo. Are you sure you want to continue?

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

Dan S. Henningson Martin Berggren August 24, 2005

Contents

1 Derivation of the Navier-Stokes equations 1.1 Notation . . . . . . . . . . . . . . . . . . . . 1.2 Kinematics . . . . . . . . . . . . . . . . . . 1.3 Reynolds transport theorem . . . . . . . . . 1.4 Momentum equation . . . . . . . . . . . . . 1.5 Energy equation . . . . . . . . . . . . . . . 1.6 Navier-Stokes equations . . . . . . . . . . . 1.7 Incompressible Navier-Stokes equations . . 1.8 Role of the pressure in incompressible ﬂow . 2 Flow physics 2.1 Exact solutions . . . . . . . . 2.2 Vorticity and streamfunction 2.3 Potential ﬂow . . . . . . . . . 2.4 Boundary layers . . . . . . . 2.5 Turbulent ﬂow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 7 8 14 15 19 20 22 24 29 29 32 40 48 54 59 59 62 65 66 69 71 71 72 73 74 76 76 80 82 83 85 85 86 87 88 88

3 Finite volume methods for incompressible ﬂow 3.1 Finite Volume method on arbitrary grids . . . . . 3.2 Finite-volume discretizations of 2D NS . . . . . . 3.3 Summary of the equations . . . . . . . . . . . . . 3.4 Time dependent ﬂows . . . . . . . . . . . . . . . 3.5 General iteration methods for steady ﬂows . . . .

4 Finite element methods for incompressible ﬂow 4.1 FEM for an advection–diﬀusion problem . . . . . . . . . . 4.1.1 Finite element approximation . . . . . . . . . . . . 4.1.2 The algebraic problem. Assembly. . . . . . . . . . 4.1.3 An example . . . . . . . . . . . . . . . . . . . . . . 4.1.4 Matrix properties and solvability . . . . . . . . . . 4.1.5 Stability and accuracy . . . . . . . . . . . . . . . . 4.1.6 Alternative Elements, 3D . . . . . . . . . . . . . . 4.2 FEM for Navier–Stokes . . . . . . . . . . . . . . . . . . . 4.2.1 A variational form of the Navier–Stokes equations 4.2.2 Finite-element approximations . . . . . . . . . . . 4.2.3 The algebraic problem in 2D . . . . . . . . . . . . 4.2.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . 4.2.5 The LBB condition . . . . . . . . . . . . . . . . . . 4.2.6 Mass conservation . . . . . . . . . . . . . . . . . . 4.2.7 Choice of ﬁnite elements. Accuracy . . . . . . . . . A Background material A.1 Iterative solutions to linear systems . A.2 Cartesian tensor notation . . . . . . A.2.1 Orthogonal transformation . A.2.2 Cartesian Tensors . . . . . . A.2.3 Permutation tensor . . . . . . A.2.4 Inner products, crossproducts A.2.5 Second rank tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . and determinants . . . . . . . . . . . . 3 . . . . . . .

95 . 95 . 98 . 98 . 99 . 100 . 100 . 101

A.2.6 Tensor ﬁelds . . . . . . . . . . . . A.2.7 Gauss & Stokes integral theorems A.2.8 Archimedes principle . . . . . . . . A.3 Curvilinear coordinates . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

102 103 104 106 109 109 113 117 120 124 129 132 136 140 146 148 152 155 157 173

B Recitations 5C1214 B.1 Tensors and invariants . . . . . . . . . . . . . . . . . . . . . B.2 Euler and Lagrange coordinates . . . . . . . . . . . . . . . . B.3 Reynolds transport theorem and stress tensor . . . . . . . . B.4 Rankine vortex and dimensionless form . . . . . . . . . . . B.5 Exact solutions to Navier-Stokes . . . . . . . . . . . . . . . B.6 More exact solutions to Navier-Stokes . . . . . . . . . . . . B.7 Axisymmetric ﬂow and irrotational vortices . . . . . . . . . B.8 Vorticity equation, Bernoulli equation and streamfunction . B.9 Flow around a submarine and other potential ﬂow problems B.10 More potential ﬂow . . . . . . . . . . . . . . . . . . . . . . . B.11 Boundary layers . . . . . . . . . . . . . . . . . . . . . . . . B.12 More boundary layers . . . . . . . . . . . . . . . . . . . . . B.13 Introduction to turbulence . . . . . . . . . . . . . . . . . . . B.14 Old Exams . . . . . . . . . . . . . . . . . . . . . . . . . . . C Study questions 5C1212

Preface

These lecture notes has evolved from a CFD course (5C1212) and a Fluid Mechanics course (5C1214) at the department of Mechanics and the department of Numerical Analysis and Computer Science (NADA) A at KTH. Erik St˚ alberg and Ori Levin has typed most of the L TEXformulas and has created the electronic versions of most ﬁgures.

Stockholm, August 2004 Dan Henningson Martin Berggren

In the latest version of the lecture notes study questions for the CFD course 5C1212 and recitation material for the Fluid Mechanics course 5C1214 has been added.

Stockholm, August 2005 Dan Henningson

Chapter 1

**Derivation of the Navier-Stokes equations
**

1.1 Notation

∂u + (u · ∂t ·u 1 p+ν ρ

**The Navier-Stokes equations in vector notation has the following form )u = − = 0
**

2

u

where the velocity components are deﬁned u = (u, v, w) = (u1 , u2 , u3 ) the nabla operator is deﬁned as = the Laplace operator is written as

2

∂ ∂ ∂ , , ∂x ∂y ∂z

=

∂ ∂ ∂ , , ∂x1 ∂x2 ∂x3

=

∂2 ∂2 ∂2 + 2+ 2 2 ∂x ∂y ∂z

and the following deﬁnitions are used ν ρ p − kinematic viscosity − density − pressure

see ﬁgure 1.1 for a deﬁnition of the coordinate system and the velocity components. The Cartesian tensor form of the equations can be written ∂ui + uj ∂ui ∂t ∂xj ∂ui ∂xi = − = 0 1 ∂p ∂ 2 ui +ν ρ ∂xi ∂xj ∂xj

where the summation convention is used. This implies that a repeated index is summed over, from 1 to 3, as follows ui ui = u 1 u1 + u 2 u2 + u 3 u3 Thus the ﬁrst component of the vector equation can be written out as 7

y, x2 v, u2 u, u1

PSfrag replacements x, x1 w, u3

z, x3 Figure 1.1: Deﬁnition of coordinate system and velocity components PSfrag replacements x, x1 2 x y, x2 ˆ P t=0 z, x3 u, u1 xi 0 v, u2 w, u3

ri

xi

0

t ,ˆ

ˆ u i xi 0 , t x1

Figure 1.2: Particle path.

i=1

∂u1 ∂u1 ∂u1 1 ∂p ∂u1 + u1 + u2 + u3 =− +ν ∂t ∂x1 ∂x2 ∂x3 ρ ∂x1 ∂u ∂u ∂u ∂u 1 ∂p +u +v +w =− +ν ∂t ∂x ∂y ∂z ρ ∂x

∂ 2 u1 ∂ 2 u1 ∂ 2 u1 + + ∂x1 2 ∂x2 2 ∂x3 2

or

**∂2u ∂2u ∂2u + 2 + 2 ∂x2 ∂y ∂z
**

2u

1.2

Kinematics

**Lagrangian and Euler coordinates
**

Kinematics is the description of motion without regard to forces. We begin by considering the motion of a ﬂuid particle in Lagrangian coordinates, the coordinates familiar from classical mechanics. Lagrange coordinates: every particle is marked and followed in ﬂow. The independent variables are xi 0 ˆ t − initial position of ﬂuid particle − time ˆ r i = r i xi 0 , t and the velocity of the particle is the rate of change of the particle position, i.e.

where the particle path of P, see ﬁgure 1.2, is

ui =

∂ri ˆ ∂t

Note here that when xi 0 changes we consider new particles. Instead of marking every ﬂuid particle it is most of the time more convenient to use Euler coordinates. Euler coordinates: consider ﬁxed point in space, ﬂuid ﬂows past point. The independent variables are xi t − space coordinates − time

Thus the ﬂuid velocity ui = ui (xi , t) is now considered as a function of the coordinate xi and time t. ˆ The relation between Lagrangian and Euler coordinates, i.e. xi 0 , t and (xi , t), is easily found by noting that the particle position is expressed in ﬁxed space coordinates xi , i.e. ˆ at the time xi = r i xi 0 , t

Material derivative

t

ˆ = t

Although it is usually most convenient to use Euler coordinates, we still need to consider the rate of change of quantities following a ﬂuid particle. This leads to the following deﬁnition. Material derivative: rate of change in time following ﬂuid particle expressed in Euler coordinates. Consider the quantity F following ﬂuid particle, where ˆ ˆ F = FL xi 0 , t = FE (xi , t) = FE ri xi , t , t The rate of change of F following a ﬂuid particle can then be written ∂FE ∂F ∂FE ∂ri ∂FE ∂t ∂FE = · + · = + ui ˆ ˆ ˆ ∂xi ∂ t ∂t ∂ t ∂t ∂xi ∂t D Based on this expression we deﬁne the material derivative as Dt ∂ ∂ ∂ ∂ D ≡ = + ui = + (u · ) ˆ Dt ∂t ∂xi ∂t ∂t In the material or substantial derivative the ﬁrst term measures the local rate of change and the second measures the change due to the motion with velocity ui . As an example we consider the acceleration of a ﬂuid particle in a steady converging river, see ﬁgure 1.3. The acceleration is deﬁned aj = Dui ∂ui ∂ui = + uj Dt ∂t ∂xj

which can be simpliﬁed in 1D for stationary case to Du ∂u ∂u = +u Dt ∂t ∂x Note that the acceleration = 0 even if velocity at ﬁxed x does not change. This has been experienced by everyone in a raft in a converging river. The raft which is following the ﬂuid is accelerating although the ﬂow ﬁeld is steady. a=

Description of deformation

Evolution of a line element ˆ Consider the two nearby particles in ﬁgure 1.4 during time dt. The position of P2 can by Taylor expansion be expressed as P2 ˆ ˆ : ri dt + dri dt = ˆ = ri (0) + ∂rti (0) dt + dri (0) + ∂ˆ

∂ ˆ ( dri ) ∂t 0

ˆ dt

ˆ ˆ = xi 0 + dxi 0 + ui (0) dt + dui (0) dt

PSfrag replacements x, x1 y, x2 z, x3 u, u1 v, u2 w, u3 x1 x2 xi 0 ˆ r i xi 0 , t 0 ˆ u i xi , t ˆ P t=0 x

u1

u2 > u 1

Figure 1.3: Acceleration of ﬂuid particles in converging river.

PSfrag replacements x, x1 y, x2 z, x3 u, u1 v, u2 w, u3 x1 x2 xi 0 ˆ r i xi 0 , t 0 ˆ u i xi , t ˆ P t=0 x u1 u2 > u 1

x2 dxi0

d

ui d

ˆ t

P2

ˆ u i dt P1 xi

0

ˆ t ri d

x3 Figure 1.4: Relative motion of two nearly particles.

dr i

x1

dtˆ

where we have used

∂ ( dri ) = ˆ ∂t = and where ∂ui ∂xj 0 dxj 0 dui We can transform the expression line element in Euler coordinates

∂ ∂ri ∂ dxj 0 = ˆ ∂xj 0 ∂ t ∂xj 0 ∂ui dxj 0 = dui ∂xj 0

∂ri ˆ ∂t

dxj 0

− change of ui with initial pos. − diﬀerence in initial pos. − diﬀerence in velocity

∂ ( dri ) = dui in Lagrange coordinates to an equation for a material ˆ ∂t

D ( dri ) = dui Dt = {expand in Euler coordinates} ∂ui drj = ∂xj

where ∂ui ∂xj drj − change in velocity with spatial position − diﬀerence in spatial pos. of particles

Relative motion associated with invariant parts We consider the relative motion dui = ∂ui ∂ui drj by dividing in its invariant parts, i.e. ∂xj ∂xj ∂ui = ξij + eij + eij ∂xj

eij

where eij is the deformation rate tensor and

ξij eij eij

= = =

∂uj 1 ∂ui − anti-symmetric part 2 ∂uj ∂xi 1 ∂ui ∂uj 2 ∂ur + − δij traceless part 2 ∂xj ∂xi 3 ∂ur 1 ∂ur δij isotropic part 3 ∂ur

∂ui , eij , describes the deformation and is considered in detail below, whereas the ∂xj anti-symmetric part can be written in terms of the vorticity ωk and is associated with solid body rotation, i.e. no deformation. The anti-symmetric part can be written The symmetric part of

dxi0 ˆ dui dt ˆ r i dt ˆ dri dt P1 P2

r2

x2 R2

π 2

+ dϕ12

**R1 R x3 r3 Figure 1.5: Deformation of a cube.
**

3

x1 r1

1 [ξij ] = − 2 1 −2

0

∂u1 ∂x2 ∂u1 ∂x3

1 2

∂u1 ∂x2

− −

∂u2 ∂x1 ∂u3 ∂x1

− 0

∂u2 ∂x1

1 2 1 2

∂u1 ∂x3 ∂u2 ∂x3

− − 0

∂u3 ∂x1 ∂u3 ∂x2

1 −2

∂u2 ∂x3

−

∂u3 ∂x2

= {use ω = 0 1 = 2 ω3 1 2 ω2 Deformation of a small cube

× u , ωk =

1 2 ω2 − 1 ω1 2

kji

∂ ui } ∂xj

1 − 2 ω3 0 1 2 ω1

0

**Consider the deformation of the small cube in ﬁgure 1.5, where we deﬁne Rl k
**

l rk

= Rδkl = Rl + k

component k of side l ∂uk l R dt ∂xj j dul k relative motion of side l

= R δkl +

∂uk dt ∂xl

deformed cube

First, we consider the deformation on side 1, which can be expressed as dR1 = r1 − R = The inner product can be expanded as ∂u1 dt ∂x1

2 1 1 r k rk − R

1 1 rk rk

=

1+

+

∂u2 dt ∂x1

2

+

∂u3 dt ∂x1

2

R2

= {drop quadratic terms} = R2 1 + 2 ∂u1 dt ∂x1

We have dropped the quadratic terms since we are assuming that dt is small. dR1 becomes ∂u1 ∂u1 dt − R = R 1 + dt + ... − R ∂x1 ∂x1

dR1

= R = R

1+2

∂u1 dt = Re11 dt ∂x1

**which implies that dR1 = Re11 dt
**

∂ui Thus the deformation rate of side 1 depends on e11 , both traceless and isotropic part of ∂xj . Second, we consider the deformation of the angle between side 1 and side 2. This can be expressed as

cos

π + dφ12 2

= = = =

r1 · r2 1 2 1 1 2 2 = r k rk · rm rm · rn rn |r1 | |r2 | δk1 + ∂u1 ∂x2 ∂u1 ∂x2

−1/2 −1/2

∂uk ∂uk ∂u1 dt δk2 + dt · 1 + 2 dt ∂x1 ∂x2 ∂x1 ∂u2 ∂u1 ∂u2 dt + dt 1− dt 1− dt ∂x1 ∂x1 ∂x2 ∂u2 + dt = 2e12 dt ∂x1

1+2

∂u2 dt ∂x2

−1/2

where we have dropped quadratic terms. We use the trigonometric identity cos π π π + dϕ12 = cos · cos dϕ12 − sin · sin dϕ12 ≈ − dϕ12 2 2 2

which allow us to obtain the ﬁnial expression dϕ12 = −2e12 dt Thus the deformation rate of angle between side 1 and side 2 depends only on traceless part of Third, we consider the deformation of the volume of the cube. This can be expressed as r1 r2 r3 − R 3 1+ = R3

∂u1 ∂x1 dt ∂u2 ∂x1 dt ∂u3 ∂x1 dt ∂ui ∂xj .

dV

=

∂u1 ∂x2 dt 2 1 + ∂u2 dt ∂x ∂u3 ∂x2 dt

1

∂u1 ∂x3 dt ∂u2 ∂x3 dt 3 + ∂u3 dt ∂x

− R3

= R3 1 +

∂u1 ∂u2 dt 1+ dt ∂x1 ∂x2 ∂u1 ∂u2 ∂u3 = R3 + + dt ∂x1 ∂x2 ∂x3 ∂uk = R3 dt ∂xk

1+

∂u3 dt − R3 ∂x3

**where we have again omitted quadratic terms. Thus we have dV = R3 err dt
**

∂ui and the deformation rate of volume of cube (or expansion rate) depends on isotropic part of ∂xj . In summary, the motion of a ﬂuid particle with velocity ui can be divided into the following invariant parts

π 2

R3 R2 R1 r3 r2 r1 + dϕ12 V (t)

S

Figure 1.6: Volume moving with the ﬂuid.

**i) ii) iii) iv) ξij = eij =
**

1 2

**ui ∂ui ∂uj − ∂xj ∂xi
**

1 = −2 kij ωk

1 2

**∂ui ∂uj 2 ∂ur + − δij ∂xj ∂xi 3 ∂ur eij =
**

1 3

∂ur δij ∂ur

1.3

Reynolds transport theorem

**Volume integral following with the ﬂuid
**

Consider the time derivative of a material volume integral, i.e. a volume integral where the volume is moving with the ﬂuid. We obtain the following expressions 1 lim ∆t→0 ∆t

∆t→0 ∆t

D Dt

V (t)

Tij dV

=

=

lim

=

** 1 + Tij (t + ∆t) dV − Tij (t) dV ∆t V (t) V (t) 1 ∂Tij lim Tij (t + ∆t) dV + dV ∆t→0 ∆t ∂t
**

V (t+∆t)−V (t) V (t)

The volume in the ﬁrst integral on the last line is represented in ﬁgure 1.6, where a volume element describing the change in volume between V at time t and t + ∆t can be written as u · n∆t = uk nk ∆t ⇒ dV = uk nk ∆t dS

¡¡¡¡¢ ¢ ¢ ¢¡¢¡¢¡¢¡ ¡¢¡¢¡¢¡ ¡¢¡ ¡ ¢ ¢¡ ¡ ¡ ¡¢ ¡ ¡ ¡ ¡¢ ¡ ¡¢ ¡¡ ¡¢¡¢¡¢¡¡¢¡¢¡¢¡¡¢¡¡¢ ¡ ¡ ¡ ¡¢¡ ¡ ¡ ¡¢¡ ¡¢¡¡ ¢¡¡¡¡ ¡¡¡¡ ¡¡ ¡ ¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡ ¢ ¢¡¢ ¡¢ ¡¢ ¡ ¡¢ ¡¢ ¡¢ ¡ ¡¢ ¡ ¡ ¢¡ ¢¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡¡ ¡¡¡¡¢¡¡¡¡¢¡¡¢¡ ¡¢¡¢¡¢¡ ¡¢¡¢¡¢¡ ¡¢¡ ¡¢ ¢¡¡¡¡¢¡¡¢¡ ¢¡¢ ¡¢ ¡¢ ¡¡¢ ¡¢ ¡¢ ¡¡¢ ¡¡ ¢¡ ¢¡ ¡ ¡ ¡¢¡ ¡ ¡ ¡¢¡ ¡¢¡¡ ¡¡¡¡ ¢¡¢¡¢¡¢¡ ¡¢¡¢¡¢¡ ¡¢¡ ¡¢ ¡ ¡ ¡ ¡¢¡ ¡ ¡ ¡¢¡ ¡¢¡¡ ¡¢¡¢¡¢¡ ¡¢¡¢¡¢¡ ¡¢¡ ¡ ¢¢¡ ¢¡¡¡¡¢¡¡¢¡ ¢¡ ¢ ¡ ¢ ¡ ¢ ¡¡ ¢ ¡ ¢ ¡ ¢ ¡¡ ¢ ¡¡¡ ¡¡¡¡¢¡¡¡¡¢¡¡¢¡ ¢¢¡¡¡¡¢¡¡¡¡¢¡¡¢¡ ¡¢¡¢¡¢¡ ¡¢¡¢¡¢¡ ¡¢¡ ¡ ¢¢¡ ¡ ¢ ¡ ¢ ¡ ¢ ¡¡ ¢ ¡ ¢ ¡ ¢ ¡¡ ¢ ¡¡¡ ¢¡¢¡¢¡¢¡ ¢ ¢ ¢ ¢ ¢ ¢ ¢ ¢ ¢¡ ¢ ¡ ¢ ¡ ¢ ¡¢¡ ¢ ¡ ¢ ¡ ¢ ¡¢¡ ¢ ¡¢¡¡ ¢¡¡¡¡ ¢¡¡ ¢¡ ¢¡ ¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡ ¢ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡¡ ¢ ¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡ ¢ ¡¢ ¡¢ ¡¢ ¡ ¡¢ ¡¢ ¡¢ ¡ ¡¢ ¡ ¡ ¢¡ ¢¡¡¡¡ ¢¡¡¡¡ ¢¡¡ ¢¡ ¢¡ ¡ ¢ ¡ ¢ ¡ ¢ ¡¡ ¢ ¡ ¢ ¡ ¢ ¡¡ ¢ ¡¡¡

V (t+∆t)

V (t + ∆t) − V (t) S (t + ∆t)

(t)

u

n

V (t + ∆t)

solid body translation solid body rotation volume constant deformation volume expansion rate

Tij (t + ∆t) dV −

V (t)

1

V (t+∆t)

Tij (t + ∆t) dV −

V (t)

Tij (t) dV

Tij (t + ∆t) dV

This implies that the volume integral can be converted to a surface integral. This surface integral can in turn be changed back to a volume integral by the use of Gauss (or Greens) theorem. We have D Dt

V (t)

Tij dV

=

∆t→0

lim

Tij (t + ∆t) uk nk dS +

S(t) V (t)

∂Tij dV ∂t

=

S(t)

Tij uk nk dS +

V (t)

∂Tij dV = {Gauss/Green’s theorem} ∂t

=

V (t)

∂Tij ∂ + (uk Tij ) dV ∂t ∂xk

which is the Reynolds transport theorem.

Conservation of mass

By the substitution Tij → ρ and the use of the Reynolds transport theorem above we can derive the equation for the conservation of mass. We have D Dt

V (t)

ρ dV =

V (t)

∂ρ ∂ + (uk ρ) = 0 ∂t ∂xk

Since the volume is arbitrary, the following must hold for the integrand 0= ∂ρ ∂ ∂ρ ∂uk Dρ ∂uk ∂ρ + (uk ρ) = + uk +ρ = +ρ ∂t ∂xk ∂t ∂xk ∂xk Dt ∂xk

1 2 3 4

where we have used the deﬁnition of the material derivative in order to simplify the expression. The terms in the expression can be given the following interpretations:

1 2 3 4

: : : :

accumulation of mass in ﬁxed element net ﬂow rate of mass out of element rate of density change of material element volume expansion rate of material element

By considering the transport of a quantity given per unit mass, i.e. Tij = ρtij , we can simplify Reynolds transport theorem. The integrand in the theorem can then be written 0 (cont. eq.) ¨0 ! ¡ ∂ ∂ ∂ρ ∂tij ∂ ∂tij Dtij ¨B (ρtij ) + (uk ρ tij ) = tij ¡ + ρ + tij ¨¨ k ρ) + ρuk (u =ρ ∂t ∂xk ∂t ∂t ∂x ∂xk Dt ¡ ¨k which implies that Reynolds transport theorem becomes D Dt

V (t)

ρ tij dV =

V (t)

ρ

Dtij dV Dt

where V (t) again is a material volume.

1.4

Momentum equation

Conservation of momentum

The momentum equation is based on the principle of conservation of momentum, i.e. that the time rate of change of momentum in a material region = sum of the forces on that region. The quantities involved are: Fi Ri ρui body forces per unit mass surface forces per unit area momentum per unit volume

V (t + ∆t) − V (t) ˆ r i dt nIII(k) S (t + ∆t) ˆ dri dt u dl P1 n P2 V (t + ∆t) x1 R(nI ) x2 x3 R3 nII nI R2 R1 dS r3 2 R(nII ) r r1 π + dϕ12 2 Figure 1.7: (t) V Momentum balance for ﬂuid element S (t) V (t + ∆t) − V (t) n S (t + ∆t) u n R V (t + ∆t) R Figure 1.8: Surface force and unit normal. We can put the momentum conservation in integral form as follows D Dt

V (t)

∆s (k)

ρui dV =

V (t)

ρFi dV +

S(t)

Ri dS

**Using Reynolds transport theorem this can be written ρ
**

V (t)

Dui dV = Dt

V (t)

ρFi dV +

S(t)

Ri dS

which is Newtons second law written for a volume of ﬂuid: mass · acceleration = sum of forces. To proceed Ri must be investigated so that the surface integral can be transformed to a volume integral. In order to do that we have to deﬁne the stress tensor.

**The stress tensor
**

Remove a ﬂuid element and replace outside ﬂuid by surface forces as in ﬁgure 1.7. Here R(n) is the surface force per unit area on surface dS with normal n, see ﬁgure 1.8. Momentum conservation for the small ﬂuid particle leads to ρ Dui dS dl = ρFi dS dl + Ri nI dS + Ri nII dS + j j Dt Ri nIII(k) ∆s(k) dl j

k

Letting dl → 0 gives 0 = Ri nI dS + Ri nII dS j j Now nj = nI = −nII which leads to j j Ri (nj ) = −Ri (−nj ) implying that a surface force on one side of a surface is balanced by an equal an opposite surface force at the other side of that surface. Note that it is a general principle that the terms proportional to the volume of a small ﬂuid particle approaches zero faster than the terms proportional to the surface area of the particle. Thus the surface forces acting on a small ﬂuid particle has to balance, irrespective of volume forces or acceleration terms.

u P1 R(e1 ) = ( T11 , T21 , T31 ) n V (t + ∆t) P2 x1 T21 x2 x1 T11 x3 T31 PSfrag replacementsR3 2 x, x1 R y, x2 R1 z, x3 r3 Figure 1.9: Deﬁnition of surface force components on a surface with a normal in the 1-direction. u, u1 r2 x2 v, u2 r1 π R(e2 ) = ( T12 , T22 , T32 ) + dϕ w, u3 12 2 V x1 (t) T22 S x2 (t) V (t + ∆t) −xi0 (t) V ˆ ri Sxi0 + ∆t) (t , t ˆ T12 u i xi 0 , t u ˆ P t=0 n V (t + ∆t) x T32 u1 u2 > u 1 x1 x2 x3 xi 0 Figure 1.10: Deﬁnition u dt of surface force components on a surface with a normal in the 2-direction. i ˆ dxi0 ˆ dui dt We now divide the surfaceˆforces into components along the coordinate directions, as in ﬁgures 1.9 and r i dt 1.10, with corresponding deﬁnitions for the force components on a surface with a normal in the 3-direction. dri dˆ Thus, Tij is the i-component t the surface force on a surface dS with a normal in the j-direction. of P1 Consider a ﬂuid particle with surfaces along the coordinate directions cut by a slanted surface, as in P2 ﬁgure 1.11. The areas of the surface elements are related by x1 x2 dSj = ej · n dS = nj dS x3 where dS is the area of the slanted surface. Momentum balance require the surface forces to balance R3 in the element, we have R2 R1 3 r0 = R1 dS − T11 n1 dS − T12 n2 dS − T13 n3 dS r2 which implies that the total surface force Ri can be written in terms of the components of the stress r1 tensor Tij as π + dϕ12 2 V (t) R = T n + T n + T n = T n i i1 1 i2 2 i3 3 ij j S (t) V (t + ∆t) − V (t) e2 S (t + ∆t) R u n n V (t + ∆t)

R1 dS1 dS2 dS3 e1

e3 Figure 1.11: Surface force balance on a ﬂuid particle with a slanted surface.

P ˆ Here the diagonal components are normal stresses oﬀ-diagonal components are t = 0 stresses. shear Further consideration of the moment balance around a ﬂuid particle show that Tij is ax symmetric tensor, u1 i.e. u2 > u 1 x1 Tij = Tji x2 x3 Momentum equation xi 0 ˆ i dt Using the deﬁnition of the surface force in terms of the stress tensor, the momentumuequation can now be dxi0 written ˆ dui dt ˆ r i dt Dui ∂Tij ρ dV = ρFi dV + Tij nj dS = ρFi + dV ˆ Dt ∂xj dri dt V (t) V (t) S(t) V (t) P1 P2 The volume is again arbitrary implying that the integrand must itself equal zero. We have x1 x2 Dui ∂Tij = ρFi + ρ x3 Dt ∂xj R3 R2 Pressure and viscous stress tensor R1 r3 For ﬂuid at rest only normal stresses present otherwise ﬂuid element would deform. We thus divide the r stress tensor into an isotropic part, the hydrodynamic pressure p, and a part depending2 on the motion of r1 the ﬂuid. We have π + dϕ12 p 2 V (t) Tij = −pδij + τij S (t) V (t + ∆t) − V (t) p - hydrodynamic pressure, directed inward τij - viscous stress tensor, depends on ﬂuid motion S (t + ∆t) u n V (t + ∆t) p

p

Newtonian ﬂuid

It is natural to assume that the viscous stresses are functions of deformation rate e ij or strain. Recall that the invariant symmetric parts are eij = 1 2 ∂ui ∂uj 2 ∂ur + − δij ∂xj ∂xi 3 ∂xr

eij

+

**1 ∂ur δij 3 ∂xr
**

eij

where eij is the volume constant deformation rate and eij is the uniform rate of expansion. For an isotropic ﬂuid, the viscous stress tensor is a linear function of the invariant parts of e ij , i.e. τij = λeij + 2µeij where we have deﬁned the two viscosities as µ (T ): λ (T ): dynamic viscosity (here T is the temperature) second viscosity, often=0

For a Newtonian ﬂuid, we thus have the following relationship between the viscous stress and the strain (deformation rate) τij = 2µeij = µ which leads to the momentum equation ρ Dui ∂p ∂ =− + µ Dt ∂xi ∂xj ∂ui ∂uj 2 ∂ur + − δij ∂xj ∂xi 3 ∂xr + ρFi ∂ui ∂uj 2 ∂ur + − δij ∂xj ∂xi 3 ∂xr

1.5

Energy equation

The energy equation is a mathematical statement which is based on the physical law that the rate of change of energy in material particle = rate that energy is received by heat and work transfers by that particle. We have the following deﬁnitions

1 ρ e + 2 ui ui dV

**energy of particle, with e the internal energy work rate of Fi on particle
**

force · velocity

ρui Fi dV

uj Rj dS ni qi dS

work rate of Rj on particle heat loss from surface, with qi the heat ﬂux vector, directed outward

**Using Reynolds transport theorem we can put the energy conservation in integral form as D Dt
**

V (t)

1 ρ e + ui ui dV = 2

V (t)

ρFi ui dV +

S(t)

[ni Tij uj − ni qi ] dS =

V (t)

ρFi ui +

∂ (Tij uj − qi ) dV ∂xi

˙ Compare the expression in classical mechanics, where the momentum equation is mu = F and the associated kinetic energy equation is (u · u) work rate ( work

m d 2 dt

= = =

F·u force · velocity force · dist. )

From the integral energy equation we obtain the total energy equation by the observation that the volume is arbitrary and thus that the integrand itself has to be zero. We have 1 ∂ ∂ ∂qi D e + ui ui = ρFi ui − (pui ) + (τij uj ) − Dt 2 ∂xi ∂xi ∂xi The mechanical energy equation is found by taking the dot product between the momentum equation and u. We obtain ρ D 1 ∂p ∂τij ui ui = ρFi ui − ui + ui Dt 2 ∂xi ∂xj Thermal energy equation is then found by subtracting the mechanical energy equation from the total energy equation, i.e. ρ ∂ui ∂ui ∂qi De = −p + τij − Dt ∂xi ∂xj ∂xi The work of the surface forces divides into viscous and pressure work as follows ρ − ∂ (pui ) ∂xi = −p ∂ui ∂xi − ui ∂p ∂xi

2

1

∂ (τij uj ) ∂xi

=

τij

∂ui ∂xj

+ uj

∂τij ∂xi

**where following interpretations can be given to the thermal and the mechanical terms
**

1: 2:

thermal terms ( force · deformation ): heat generated by compression and viscous dissipation mechanical terms ( velocity · force gradients ): gradients accelerate ﬂuid and increase kinetic energy

The heat ﬂux need to be related to the temperature gradients with Fouriers law qi = −κ ∂T ∂xi

where κ = κ (T ) is the thermal conductivity. This allows us to write the thermal energy equation as ρ De ∂ui ∂ = −p +Φ+ Dt ∂xi ∂xi κ ∂T ∂xi

where the positive deﬁnite dissipation function Φ is deﬁned as ∂ui 1 = 2µ eij eij − ∂xj 3 1 ∂uk δij 3 ∂xk

2

Φ = τij

∂uk ∂xk

2

= 2µ eij −

>0

Alternative form of the thermal energy equation can be derived using the deﬁnition of the enthalpy h=e+ We have Dh De 1 Dp p = + − 2 Dt Dt ρ Dt ρ Dρ Dt

− ρ ∂xi

i ∂u

p ρ

which gives the ﬁnal result ρ Dh Dp ∂ = +Φ+ Dt Dt ∂xi κ ∂T ∂xi

To close the system of equations we need a i) ii) thermodynamic equation equation of state e = e (T, p) p = ρRT simplest case: e = cv T or h = cp T

where cv and cp are the speciﬁc heats at constant volume and temperature, respectively. At this time we also deﬁne the ratio of the speciﬁc heats as γ= cp cv

1.6

Navier-Stokes equations

The derivation is now completed and we are left with the Navier-Stokes equations. They are the equation describing the conservation of mass, the equation describing the conservation of momentum and the equation describing the conservation of energy. We have Dρ ∂uk +ρ =0 Dt ∂xk ρ ρ where Dui ∂p ∂τij =− + + ρFi Dt ∂xi ∂xj κ ∂T ∂xi

De ∂ui ∂ = −p +Φ+ Dt ∂xi ∂xi

Φ

= τij

and the thermodynamic relation and the equation of state for a gas e = e (T, p) p = ρRT We have 7 equations and 7 unknowns and therefore the necessary requirements to obtain a solution of the system of equations. unknown ρ 1 ui 3 p 1 e 1 T 1 7 equations continuity momentum energy thermodyn. gas law 1 3 1 1 1 7

τij

∂ui ∂xj ∂uj 2 ∂ur ∂ui + − δij = µ ∂xj ∂xi 3 ∂xr

**Equations in conservative form
**

A slightly diﬀerent version of the equations can be found by using the identity ρ to obtain the system ∂ρ ∂ ∂t + ∂x (uk ρ) = 0 k ∂ ∂ ∂p ∂τij (ρui ) + (ρuk ui ) = − + + ρFi ∂xk ∂xi ∂xj ∂t ∂ 1 ∂ 1 ∂ ∂ ∂ = ρFi ui − (pui ) + (τij uj ) + ∂t e + 2 ui ui + ∂xk ρuk e + 2 ui ui ∂xi ∂xi ∂xi ∂U ∂G(i) + =J ∂t ∂xi where U = (ρ, ρu1 , ρu2 , ρu3 , ρ (e + ui ui /2)) is the vector of unknowns, J = (0, ρF1 , ρF2 , ρF3 , ρui Fi ) is the vector of the right hand sides and G(i) =

T T

Dtij ∂ ∂ = (ρtij ) + (uk ρtij ) Dt ∂t ∂xk

κ

∂T ∂xi

These are all of the form

or

∂U ∂G ∂G ∂G + + + =J ∂t ∂x ∂y ∂z

ρui ρu1 ui + pδ1i − τ1i ρu2 ui + pδ2i − τ2i ρu3 ui + pδ3i − τ3i ∂T ρ (e + ui ui /2) ui + pui + κ ∂xi − uj τij

is the ﬂux of mass, momentum and energy, respectively. This form of the equation is usually termed the conservative form of the Navier-Stokes equations.

1.7

Incompressible Navier-Stokes equations

The conservation of mass and momentum can be written Dρ + ρ ∂ui = 0 Dt ∂xi Dui ∂p ∂τij ρ =− + + ρFi ∂xi ∂xj Dt τ = µ ∂ui + ∂uj − 2 ∂uk δ ij ij ∂xj ∂xi 3 ∂uk

for incompressible ﬂow ρ = constant, which from the conservation of mass equation implies

∂ui =0 ∂xi implying that a ﬂuid particle experiences no change in volume. Thus the conservation of mass and momentum reduce to ∂ui + uj ∂ui = − 1 ∂p + ν 2 ui + Fi ∂t ∂xj ρ ∂xi ∂ui ∂xi = 0 since the components of the viscous stress tensor can now be written ∂ ∂xj ∂ui ∂uj 2 ∂uk + − δij ∂xj ∂xi 3 ∂xk = ∂ 2 ui = ∂xj ∂xj

2

ui

We have also introduced the kinematic viscosity, ν, above, deﬁned as ν = µ/ρ The incompressible version of the conservation of mass and momentum equations are usually referred to as the incompressible Navier-Stokes equations, or just the Navier-Stokes equations in case it is evident that the incompressible limit is assumed. The reason that the energy equation is not included in the incompressible Navier-Stokes equations is that it decouples from the momentum and conservation of mass equations, as we will see below. In addition it can be worth noting that the conservation of mass equation is sometimes referred to as the continuity equation. The incompressible Navier-Stokes equations need boundary and initial conditions in order for a solution to be possible. Boundary conditions (BC) on solid surfaces are ui = 0. They are for obvious reasons usually referred to as the no slip conditions. As initial condition (IC) one needs to specify the velocity ﬁeld at the initial time ui (t = 0) = u0 . The pressure ﬁeld does not need to be speciﬁed as it can be obtained once the i velocity ﬁeld is speciﬁed, as will be discussed below.

**Integral form of the Navier-Stokes eq.
**

We have derived the diﬀerential form of the Navier-Stokes equations. Sometimes, for example when a ﬁnite-volume discretization is derived, it is convenient to use an integral form of the equations. An integral from of the continuity equation is found by integrating the divergence constraint over a ﬁxed volume VF and using the Gauss theorem. We have ∂ (ui ) dVF = ∂xi ui ni dSF = 0

S

V

An integral form of the momentum equation is found by taking the time derivative of a ﬁxed volume integral of the velocity, substituting the diﬀerential form of the Navier-Stokes equation, using the continuity equation and Gauss theorem. We have ∂ ∂t

VF

ui dV =

VF

∂ui dV = − ∂t

VF

**∂ 1 ∂p ∂ 2 ui (uj ui ) + −ν ∂xj ρ ∂xi ∂xj ∂xj dV = −
**

SF

dV =

−

VF

∂ 1 ∂ui uj ui + pδij − ν ∂xj ρ ∂xj

p ∂ui ui uj n j + n i − ν nj dS ρ ∂xj

Tw

T0 , U 0

L

Figure 1.12: Examples of length, velocity and temperature scales. Note that the integral from of the continuity equation is a compatibility condition for BC in incompressible ﬂow.

Dimensionless form

It is often convenient to work with a non-dimensional form of the Navier-Stokes equations. We use the following scales and non-dimensional variables x∗ = i xi L t∗ = tU0 L u∗ = i ui U0 p∗ = p p0 ∂ 1 ∂ = ∂xi L ∂x∗ i ∂ U0 ∂ = ∂t L ∂t∗ ⇒

where ∗ signiﬁes a non-dimensional variable. The length and velocity scales have to be chosen appropriately from the problem under investigation, so that they represent typical lengths and velocities present. See ﬁgure 1.12 for examples. If we introduce the non-dimensional variables in the Navier-Stokes equations we ﬁnd 2 ∗ U 2 ∂u∗ p0 ∂p∗ νU0 ∂ 2 u∗ U0 ∂ui i i · ∗ + 0 u∗ ∗ = − + 2 j L ∂t L ∂xj ρL ∂x∗ L ∂x∗ ∂x∗ i j j U0 ∂u∗ i · =0 L ∂x∗ i ∂ui + uj ∂ui = − p0 ∂p + ν 2 ∂t ∂xj ρU0 ∂xi U0 L ∂ui =0 ∂xi

2 U0 L U0 /L = ν νU0 /L2 2

2 Now drop * as a sign of non-dimensional variables and divide through with U0 /L, we ﬁnd

We have deﬁned the Reynolds number Re as Re =

“inertial forces” “viscous forces”

which can be interpreted as a measure of the inertial forces divided by the viscous forces. The Reynolds number is by far the single most important non-dimensional number in ﬂuid mechanics. We also deﬁne the pressure scale as

2 p0 = ρU0

which leaves us with the ﬁnal form of the non-dimensional incompressible Navier-Stokes equations as ∂ui + uj ∂ui = − ∂p + 1 ∂t ∂xj ∂xi Re ∂ui =0 ∂xi

2

**¥¤¤¤¤¤¤¤¤¤¤¥ £ ¥£ ¥£ ¥£ ¥¥£ ¥£ ¥£ ¥£ ¥£ ¥£ ¥¤¥¤¥¤¥¤¤¥¤¥¤¥¤¥¤¥¤£ £¤£¤£¤£¤¤£¤£¤£¤£¤£¤ ¤¤¤¤¤¤¤¤¤¤£¥ £
**

T0 U0

S (t + ∆t) u n V (t + ∆t) p

Tw L

**§¤¤¤¤¤¤¤¤¤¤§ ¦ §¦ §¦ §¦ §§¦ §¦ §¦ §¦ §¦ §¦ §¤§¤§¤§¤¤§¤§¤§¤§¤§¤¦ ¦¤¦¤¦¤¦¤¤¦¤¦¤¦¤¦¤¦¤ ¤¤¤¤¤¤¤¤¤¤¦§ ¦
**

ui ui

**Non-dimensional energy equation
**

We stated above that the energy equation decouple from the rest of the Navier-Stokes equations for incompressible ﬂow. This can be seen from a non-dimensionalization of the energy equation. We use the deﬁnition of the enthalpy (h = cp T ) to write the thermal energy equation as DT Dp = + Φ + κ 2T Dt Dt and use the following non-dimensional forms of the temperature and dissipation function ρcp T∗ = This leads to DT ∗ Dt∗ T − T0 Tw − T 0 Φ∗ = L2 2 Φ U0 µ

= =

µ κ · · ρU0 L µcp 1 1 · · Re P r

∗2

∗2

T∗ +

T∗ +

2 U0 cp (Tw − T0 ) M a2 cp (Tw0 −T

a2 0)

2 U0 Dp∗ µ + φ∗ cp (Tw − T0 ) Dt∗ ρU0 L

Dp∗ 1 + Φ∗ Dt∗ R

where a0 speed of sound in free-stream and M a = to obtain

U0 is Mach number. We now drop ∗ and let M a → 0 a0

DT 1 1 = · · 2T Dt Re P r µc where the Prandtl number P r = κp , which takes on a value of ≈ 0.7 for air. As the Mach number approaches zero, the ﬂuid behaves as an incompressible medium, which can be seen from the deﬁnition of the speed of sound. We have a0 = γ , ρα α= 1 ∂ρ ρ ∂p

T

where α is the isothermal compressibility coeﬃcient. If the density is constant, not depending on p, we have that α → 0 and a0 → ∞, which implies that M a → 0.

1.8

Role of the pressure in incompressible ﬂow

The role of the pressure in incompressible ﬂow is special due to the absence of a time derivative in the continuity equation. We will illustrate this in two diﬀerent ways.

Artiﬁcial compressibility

First we discuss the artiﬁcial compressibility version of the equations, used sometimes for solution of the steady equations. We deﬁne a simpliﬁed continuity equation and equation of state as ∂ρ ∂ui + =0 and p = βρ ∂t ∂xi This allows us to write the Navier-Stokes equations as ∂ui + ∂ (uj ui ) = − ∂p + 1 ∂t ∂xj ∂xi Re ∂ui ∂p +β = 0 ∂t ∂xi Let p βu βv u p + u2 uv u= e= f = v uv p + v2 w uw vw

2

ui

βw uw g= vw p + β2

be the vector of unknowns and their ﬂuxes. The Navier-Stokes equations can then be written ∂u ∂e ∂f ∂g 1 + + + = ∂t ∂x ∂y ∂z Re or

2

Du

2

∂u ∂u ∂u ∂u 1 +A +B +C = ∂t ∂x ∂y ∂z R 0 0 0 0 , B = ∂f = u 0 ∂u 0 u 0 0 β 0 0 v u 0 , D = 0 0 2v 0 0 w v 0

Du

the eigenvalues of A, B, C are the wave speeds of plane waves in the respective coordinate directions, they are and w, w, w ± w2 + β √ Note that the eﬀective acoustic or pressure wave speed ∼ β → ∞ for incompressible ﬂow. u, u, u ± v, v, v ± u2 + β , v2 + β

where the matrices above are deﬁned 0 β ∂e 1 2u = A= ∂u 0 v 0 w 0 ∂g 0 C= = ∂u 1 0

0 0 β 0 v u 1 0 2v 0 0 w 0 1 0 0 0 0 1 0 0 0 0 1

0 0 0 v

**Projection on a divergence free space
**

Second we discuss the projection of the velocity ﬁeld on a divergence free space. We begin by the following theorem. Theorem 1. Any wi in Ω can uniquely be decomposed into wi = u i + where ∂ui = 0, ∂xi ∂p ∂xi on ∂Ω.

u i · ni = 0

i.e. into a function ui that is divergence free and parallel to the boundary and the gradient of a function, here called p. Proof. We start by showing that ui and ui , ∂p ∂xi =

Ω

**∂p is orthogonal in the L2 inner product. ∂xi ∂p ∂ ui dV = (ui p) dV = pui ni dS = 0 ∂xi ∂xi
**

Ω ∂Ω

The uniqueness of the decomposition can be seen by assuming that we have two diﬀerent decompositions and showing that they have to be equivalent. Let wi = u i The inner product between ui gives 0=

Ω (1) (1)

+

∂p(1) ∂p(2) (2) = ui + ∂xi ∂xi

− ui

(2)

+

∂ p(1) − p(2) ∂xi

and ui

(1)

− ui

(2)

ui

(1)

− ui

(2)

2

+ ui

(1)

− ui

(2)

∂ p(1) − p(2) ∂xi

dV =

Ω

ui

(1)

− ui

(2)

2

dV

Thus we have

ui

(1)

= ui ,

(2)

p(1) = p(2) + C

r + dϕ12 V (t) S (t) V (t + ∆t) − V (t) S (t + ∆t) u n V (t + ∆t) p T0 , U 0 Tw L T0 U0

π 2

Gradient ﬁelds

wi

∂p ∂xi

ui

Divergence free vectorﬁelds // to boundary Figure 1.13: Projection of a function on a divergence free space. We can ﬁnd an equation for a p with above properties by noting that

2

p=

∂wi ∂xi

in Ω with

∂p = wi ni on ∂Ω ∂n

has a unique solution. Now, if ui = w i − we have 0= ∂ui ∂wi ∂2p = − ; ∂xi ∂xi ∂xi ∂xi

∂p ⇒ ∂xi 0 = u i ni = w i ni − ∂p ∂xi ∂p ∂n

and

**To sum up, to project wi on divergence free space let wi = ui + 1) solve for p : ∂wi = ∂xi ∂p ∂xi
**

2

p,

∂p =0 ∂n

2) let ui = wi −

This is schematically shown in ﬁgure 1.13. Note also that (wi − ui ) ⊥ui .

**Apply to Navier-Stokes equations
**

Let P be orthogonal projector which maps wi on divergence free part ui , i.e. Pwi = ui . We then have the following relations wi = Pwi + Pui = ui , P ∂p ∂xi ∂p =0 ∂xi

Applying the orthogonal projector to the Navier-Stokes equations, we have P ∂ui ∂p + ∂t ∂xi = P −uj ∂ui 1 + ∂xj Re

2

ui

Now, ui is divergence free and parallel to boundary, thus

P However, P

∂ui ∂ui = ∂t ∂t

2

ui =

2

ui

since 2 ui need not be parallel to the boundary. Thus we ﬁnd an evolution equation without the pressure, we have ∂ui ∂ui 1 = P −uj + ∂t ∂xj Re

wi 2

The pressure term in the Naiver-Stokes equations ensures that the right hand side w i is divergence free. We can ﬁnd a Poisson equation for the pressure by taking the divergence of w i , according to the result above. We ﬁnd

2

ui

p=

∂ ∂xi

thus the pressure satisﬁes elliptic Poisson equation, which links the velocity ﬁeld in the whole domain instantaneously. This can be interpreted such that the information in incompressible ﬂow spreads inﬁnitely fast, i.e. we have an inﬁnite wave speed for pressure waves, something seen in the analysis of the artiﬁcial compressibility equations.

** ∂ui 1 + −uj ∂xj Re
**

wi

2

∂uj ∂ui ui = − ∂xi ∂xj

**Evolution equation for the divergence
**

It is common in several numerical solution algorithms for the Naiver-Stokes equations to discard the divergence constraint and instead use the pressure Poisson equation derived above as the second equation in the incompressible Naiver-Stokes equations. If this is done we may encounter solutions that are not divergence free. Consider the equation for the evolution of the divergence, found by taking the divergence of the momentum equations and substituting the Laplacian of the pressure with the right hand side of the Poisson equation. We have ∂ ∂t ∂ui ∂xi = 1 Re

2

∂ui ∂xi

Thus the divergence is not automatically zero, but satisﬁes a heat equation. For the stationary case the solution obeys the maximum principle. This states that the maximum of a harmonic function, a function satisfying the Laplace equation, has its maximum on the boundary. Thus, the divergence is zero inside the domain, if and only if it is zero everywhere on the boundary. Since it is the pressure term in the Navier-Stokes equations ensures that the velocity is divergence free, this has implications for the boundary conditions for the pressure Poisson equation. A priori we have none speciﬁed, but they have to be chosen such that the divergence of the velocity ﬁeld is zero on the boundary. This may be a diﬃcult constraint to satisfy in a numerical solution algorithm.

Chapter 2

z, x3 u, u1 v, u2 w, u3 x1 x2 xi 0 ˆ r i xi 0 , t 0 ˆ u i xi , t ˆ P t=0

Flowuxphysics

1

u2 > u 1 x1 x2 x3 2.1 Exact solutions xi 0 ˆ u i dt Plane Pouseuille ﬂow - exact solution for channel ﬂow dxi0 ˆ dui dt The ﬂow inside the two-dimensional channel, see ﬁgure 2.1, is driven by a pressure diﬀerence p 0 −p1 between ˆ r i dt the inlet and the outlet of the channel. We will assume two-dimensional, steady ﬂow, i.e. ˆ dri dt P1 ∂ = 0, u3 = 0 P2 ∂t x1 and that the ﬂow is fully developed, meaning that eﬀects of inlet conditions have disappeared. We have x2

x3 u1 = u1 (x2 ) , u2 = u2 (x2 ) R3 2 R The boundary conditions are u1 = u2 = 0 at x2 = ±h. The continuity equation reads R1 r3 0 0 r2 U ∂u2 U ∂u1 ∂u3 r1 + =0 + π ∂x ∂x2 3 ∂x + dϕ12 2 1 V (t) where the ﬁrst and the last term disappears due to the assumption of two-dimensional ﬂow and that S (t) the ﬂow is fully developed. This implies that u2 = C = 0, where the constant is seen to be zero from the V (t + ∆t) − V (t) boundary conditions. S (t + ∆t) The steady momentum equations read u n ρu ∂u1 + ρu ∂u1 = − ∂p + µ 2 u V (t + ∆t) 1 2 1 ∂x1 ∂x2 ∂x1 p ρu ∂u2 + ρu ∂u2 = − ∂p + µ 2 u − ρg 1 2 2 T0 , U 0 ∂x1 ∂x2 ∂x2 Tw which, by applying the assumptions above, reduce to L T0 U0 x2 L Gradient ﬁelds wi ∂p h ∂xi ui ree vectorﬁelds // to boundary p0 > p 1 x1 p

**©¨©¨©¨©©¨©¨©¨©¨©©¨©©¨©©¨©¨©©¨©©¨©© ¨ ¨ ¨ ¨ ¨ ¨ ¨ ¨ ©©©©©©©©©©©©©©©©©©©©©¨ ©©©©©©©©©©©©©©©©©©©©© ©©©©©©©©©©©©©©©©©©©©©
**

29

1

Figure 2.1: Plane channel geometry.

ui Divergence free vectorﬁelds // to boundary L h x1 x2 p0 > p 1 p1 dS

u n

u·n ∆t

Figure 2.2: Volume of ﬂuid ﬂowing through surface dS in time ∆t

0=−

∂p d2 u 1 +µ ∂x1 dx2 2 ∂p − ρg 0=− ∂x2

The momentum equation in the x2 -direction (or normal direction) implies p = −ρgx2 + P (x1 ) ⇒ ∂p dP = ∂x1 dx1

showing that the pressure gradient in the x1 -direction (or streamwise direction) is only a function of x1 . The momentum equation in the streamwise direction implies 0=− Since P (x1 ) and u1 (x2 ) we have d2 u 1 1 dP = = const. (independent of x1 , x2 ) dx2 µ dx1 2 We can integrate u1 in the normal direction to obtain u1 = 1 dP 2 x + C 1 x2 + C 2 2µ dx1 2 dP d2 u 1 +µ dx1 dx2 2

The constants can be evaluated from the boundary conditions u1 (±h) = 0, giving the parabolic velocity proﬁle u1 = h2 dP x2 · 1− 2µ dx1 h

2

Evaluating the maximum velocity at the center of the channel we have umax = − h2 dP · >0 2µ dx1

if ﬂow is in the positive x1 -direction. The velocity becomes u1 x2 =1− umax h Flow rate From ﬁgure 2.2 we can evaluate the ﬂow rate as dQ = u · n dS. Integrating over the channel we ﬁnd

h 2

Q=

S

ui ni dS = {channel} =

−h

u1 dx2 = −

2h3 dP 4 · = humax 3µ dx1 3

L h x1 x2 p0 > p 1 p1 u n dS u · n∆t

y

u0 p0 x

Wall shear stress The wall shear stress can be evaluated from the velocity gradient at the wall. We have τ12 = τ21 = µ implying that τwall = τ Vorticity The vorticity is zero in the streamwise and normal directions, i.e. ω1 = ω2 = 0 and has the following expression in the x3 -direction (or spanwise direction) ω3 = ∂u2 ∂u1 du1 2umax · x2 − =− = ∂x1 ∂x2 dx2 h2

x2 =−h

**Stokes 1:st problem: instantaneously started plate
**

Consider the instantaneously starting plate of inﬁnite horizontal extent shown in ﬁgure 2.3. This is a time dependent problem where plate velocity is set to u = u0 at time t = 0. We assume that the velocity can be written u = u(y, t) which using continuity and the boundary conditions imply that normal velocity v = 0. The momentum equations reduce to ∂u ∂t 1 ∂p ∂2u +ν 2 ρ ∂x ∂y 1 ∂p 0 = − −g ρ ∂y = −

The normal momentum equation implies that p = −ρgy + p0 , where p0 is the atmospheric pressure at the plate surface. This gives us the following diﬀusion equation, initial and boundary conditions for u ∂u ∂2u = ν 2 ∂t ∂y u(y, 0) = 0 0 ≤ y ≤ ∞ u(0, t) = u0 t > 0 u(∞, t) = 0 t > 0

We look for a similarity solution, i.e. we introduce a new dependent variable which is a combination of y and t such that the partial diﬀerential equation reduces to an ordinary diﬀerential equation. Let

2µ · umax · x2 du1 =− dx2 h2 2µ umax h =

Figure 2.3: Stokes instantaneously plate.

u η = C · y · tb u0 where C and b are constants to be chosen appropriately. We transform the time and space derivatives according to f (η) = ∂ ∂t ∂ ∂y ∂2 ∂y 2 ∂η d d = Cbytb−1 ∂t dη dη ∂η d d = = Ctb ∂y dη dη d2 = C 2 t2b 2 dη =

If we substitute this into the equation for u and use the deﬁnition of η, we ﬁnd an equation for f df d2 f = νC 2 t2b 2 dη dη where no explicit dependence on y and t can remain if a similarity solution is to exist. Thus, the coeﬃcient of the two η-dependent terms have to be proportional to each other, we have bt−1 η bt−1 = κνC 2 t2b where κ is a proportionality constant. The exponents of these expressions, as well as the coeﬃcients in front of the t-terms have to be equal. This gives b=− 1 2 C=√ 1 −2κν

We choose κ = −2 and obtain the following f + 2ηf = 0

y η= √ 2 νt

d where = dη . The boundary and initial conditions also have to be compatible with the similarity assumption. We have

**u(y, 0) u(0, t) = f (∞) = 0 = f (0) = 1 u0 u0 This equation can be integrated twice to obtain
**

η

u(∞, t) = f (∞) = 0 u0

f = C1

0

eξ dξ + C2

2

**Using the boundary conditions and the deﬁnition of the error-function, we have
**

η 2 2 f =1− √ eξ dξ = 1 − erf(η) π 0 This solution is shown in ﬁgure 2.4, both as a function of the similarity variable η and the unscaled normal coordinate y. Note that the time dependent solution is diﬀusing upward. From the velocity we can calculate the spanwise vorticity as a function of the similarity variable η and the unscaled normal coordinate y. Here the maximum of the vorticity is also changing in time, from the inﬁnite value at t = 0 it is diﬀusing outward in time.

∂v ∂u u0 −η2 − =√ e ∂x ∂y πνt The vorticity is shown in ﬁgure 2.5, both as a function of the similarity variable η and the unscaled normal coordinate y. Note again that the time dependent solution is diﬀusing upward. ω3 = ω z =

2.2

Vorticity and streamfunction

Vorticity is an important concept in ﬂuid dynamics. It is related to the average angular momentum of a ﬂuid particle and the swirl present in the ﬂow. However, a ﬂow with circular streamlines may have zero vorticity and a ﬂow with straight streamlines may have a non-zero vorticity.

rﬁelds // to boundaryn Divergence free vectorﬁelds // to boundary n dS dS L L u · n∆t u · n∆t hx hx x1 y x1 y √2 xu x y η = √2 0 yp=> p 4νt η u0 t 4νt p0 > p 1 0 1 p0 p0 y p1 p1 η = √4νt u u √n n y = η 4νt dS dS t u u u · n∆t u · n∆t u0 u0 x x Figure 2.4: Velocity above the instantaneously started plate. a) U as a function of the similarity variable y y η. b) U (y, t) for various times. u0 u0 p0 p0 y y η = √4νt η = √4νt

2

8

1.8 1.6 1.4 1.2 1

7

6

5

4

0.8 0.6 0.4 0.2 0

3

2

1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

2

8

1.8 1.6 1.4 1.2 1 0.8 0.6

√ y = η 4νt t η y ωz =

ω(η)·u0 ˜ √ πνt

u u0

√ y = η 4νt ty η

u u0

7

6

5

4

3

t

2

0.4 0.2 0

t

1

0

0.1

0.2

0.3

ωz = = ω(η) ˜ √ Figure 2.5: Vorticity above the instantaneously started plate. a) ωz πνt/u0 as a function of the similarity variable η. b) ωz (y, t) for various times.

ωz

√ πνt u0

0.4

0.5

0.6

0.7

0.8

0.9

1

√ ωz πνt u0

= ω (η) ˜

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

ω(η)·u0 ˜ √ πνt

**Vorticity and circulation
**

The vorticity is deﬁned mathematically as the curl of the velocity ﬁeld as ω= In tensor notation this expression becomes ωi =

ijk

×u ∂uk ∂xj

**Another concept closely related to the vorticity is the circulation, which is deﬁned as Γ=
**

C

ui dxi =

C

ui ti ds

In ﬁgure 2.6 we show a closed curve C. Using Stokes theorem we can transform that integral to one over the area S as ∂uk Γ= dS = ωi ni dS ijk ni ∂xj S S This allows us to ﬁnd the following relationship between the circulation and vorticity dΓ = ω i ni dS, which can also be written dΓ = ω i ni dS Thus, one interpretation is that the vorticity is the circulation per unit area for a surface perpendicular to the vorticity vector. Example 1. The circulation of an ideal vortex. Let the azimuthal velocity be given as u θ = C/r. We can calculate the circulation of this ﬂow as

2π

Γ=

C

uθ r dθ = C

0

dθ = 2πC

Thus we have the following relationship for the ideal vortex uθ = Γ 2πr

√ u0 y = η 4νt t η y ˜ (η)·u ωz = ω√πνt0 t

dxi = ti ds

S

C

Figure 2.6: Integral along a closed curve C with enclosed area S. The ideal vortex has its name from the fact that its vorticity is zero everywhere, except for an inﬁnite value at the center of the vortex.

**Derivation of the Vorticity Equation
**

We will now derive an equation for the vorticity. We start with the dimensionless momentum equations. We have ∂ui ∂ui ∂p 1 2 + uj =− + ui ∂t ∂xj ∂xi Re We slightly modify this equation by introducing the following alternative form of the non-linear term uj ∂ui ∂ = ∂xj ∂xi 1 uj uj 2 +

ijk ωj uk ∂ pqi ∂xq )

**this can readily be derived using tensor manipulation. We take the curl ( equations and ﬁnd ∂ ∂t
**

pqi

of the momentum ∂ui ∂xq

∂ui ∂xq

+

pqi

∂2 ∂xq ∂xi

1 uj uj 2

+

pqi

∂ ( ∂xq

ijk ωj uk )

=−

pqi

∂2p 1 + ∂xq ∂xi Re

pqi

2

**The second and the fourth terms are zero, since i, q
**

pqi

pqi

is anti-symmetric in i, q and ∂xq ∂xi is symmetric in

∂2 =0 ∂xq ∂xi

The ﬁrst term and the last term can be written as derivatives of the vorticity, and we now have the following from of the vorticity equation ∂ 1 2 ∂ωp + pqi ( ijk ωj uk ) = ωp ∂t ∂xq Re The second term in this equation can be written as

ipq ijk

∂ ∂ ∂ ∂ (ωj uk ) = (δpj δqk − δpk δqj ) (ωj uk ) = (ωp uk ) − (ωj up ) = ∂xq ∂xq ∂xk ∂xj uk ∂ωp ∂uk ∂ωj ∂up + ωp − up − ωj ∂xk ∂xk ∂xj ∂xj

Due to continuity

∂ωj ∂xj

=

∂ 2 uq jpq ∂xj ∂xp

**= 0 and we are left with the relation
**

ipq ijk

∂ ∂ωp ∂up (ωj uk ) = uk − ωj ∂xq ∂xk ∂xj

**Thus, the vorticity equation becomes ∂ωp ∂ωp ∂up 1 + uk = ωj + ∂t ∂xk ∂xj Re or written in vector-notation Dω = (ω · Dt )u + 1 Re
**

2 2

ωp

ω

**Components in Cartesian coordinates
**

In cartesian coordinates the components of the vorticity vector in three-dimensions become ex ω=

∂ ∂x

ey

∂ ∂y

ez

∂ ∂z

=

u

v

w

∂w ∂v − ∂y ∂z

ex +

∂u ∂w − ∂z ∂x

ey +

∂v ∂u − ∂x ∂y

ez

For two-dimensional ﬂow it is easy to see that this reduces to a non-zero vorticity in the spanwise direction, but zero in the other two directions. We have ωx = 0, ωy = 0, ωz = ∂v ∂u − ∂x ∂y

This simpliﬁes the interpretation and use of the concept of vorticity greatly in two-dimensional ﬂow.

**Vorticity and viscocity
**

There exists a subtle relationship between ﬂows with vorticity and ﬂows on which viscous forces play a role. It is possible to show that × ω = 0 ⇐⇒ viscous forces = 0 This means that without viscous forces there cannot be a vorticity ﬁeld in the ﬂow which varies with the coordinate directions. This is expressed by the following relationship ∂τij =µ ∂xj

2

ui = −µ

ijk

∂ωk ∂xj

**This can be derived in the following manner. We have −µ
**

ijk

∂ωk ∂xj

**∂ ∂ un ∂xj ∂xm ∂ 2 un = −µ(δim δjn − δin δjm ) ∂xj ∂xm = −µ
**

ijk kmn

= µ = µ

∂ 2 ui ∂xj ∂xj

2

ui

**Terms in the two-dimensional vorticity equation
**

In two dimensions the equation for the only non-zero vorticity component can be written ∂ω3 ∂ω3 ∂ω3 + u1 + u2 =ν ∂t ∂x1 ∂x2

2

ω3

We will take two examples elucidating the meaning of the terms in the vorticity equation. First the diﬀusion of vorticity and second the advection or transport of vorticity. Example 2. Diﬀusion of vorticity: the Stokes 1st problem revisited. From the solution of Stokes 1st problem above, we ﬁnd that the vorticity has the following form ω3 = ωz = ω(y, t) and that it thus is a solution of the following diﬀusion equation ∂ω ∂2ω =ν 2 ∂t ∂y The time evolution of the vorticiy can be seen in ﬁgure 2.7, and is given by u0 −η2 y ω= √ e , where η = √ πνt 2 νt √ where the factor νt can be identiﬁed as the diﬀusion length scale, where we have used the following dimensional relations √ L2 [ν] = , [t] = T ⇒ [ νt] = L T

i

**ree vectorﬁelds // to boundary L h x1 x2 p0 > p 1 p1 u n dS u · n∆t x y u0 p0 y η = √4νt √ y = η 4νt t η y ω(η)·u0 ˜ √ ωz = πνt t
**

u u0

η

y = η 4νt t ηy y ˜ (η)·u ωz = ω√πνt0 t

y = √4νt u √ u0

u0 p0

8

7

6

5

4

3

t

2

1

0

0

0.1

0.2

0.3

0.4

ω

0.5

0.6

0.7

0.8

0.9

1

Figure 2.7: A sketch of the evolution of vorticity in Stokes 1st problem. v, y

ω=0 x, u Figure 2.8: Stagnation point ﬂow with a viscous layer close to the wall. In this ﬂow the streamlines are straight lines and therefore the non-linear term, or the transport/advection term, is zero. We now turn to an example where the transport of vorticity is important. Example 3. Transport/advection of vorticity: Stagnation point ﬂow. The stagnation point ﬂow is depicted in ﬁgure 2.8, and is a ﬂow where a uniform velocity approaches a plate where it is showed down and turned to a ﬂow parallel two the plates in both the positive and negative x-direction. At the origin we have a stagnation point. The inviscid stagnation point ﬂow is given as follows inviscid: u = cx v = −cy ⇒ ωz = ω = ∂v ∂u − =0 ∂x ∂y

Note that this ﬂow has zero vorticity and that it does not satisfy the no-slip condition (u = v = 0, y = 0). In order to satisfy this condition we introduce viscosity, or equally, vorticity. Thus close to the plate we have to fulﬁll the equation ∂ω ∂ω ∂2ω ∂2ω u +v =ν + ∂x ∂y ∂x2 ∂y 2 The two ﬁrst terms represent advection or transport of vorticity. We will try a simple modiﬁcation of the inviscid ﬂow close to the boundary, which is written as u = xf (y) v = −f (y) This satisﬁes the continuity equation, and will be zero on the boundary and approach the inviscid ﬂow far above the plate if we apply the following boundary conditions u = v = 0, y=0 u ∝ cx, v ∝ −cy, y → ∞ The vorticity can be written ω = −xf (y). We introduce this and the above deﬁnitions of the velocity components into the vorticity equation, and ﬁnd −f f + f f + νf =0

0

η=

y = η 4νt t η y F,˜F , F √ ωz = ω(η)·u0 πνt t η

√y 4νt u √ u0

p0

1.5

F

1

F

0.5

F

0 0 1

η

2

Figure 2.9: Solution to the non-dimensional equation for stagnation point ﬂow. F is proportional to the normal velocity, F to the streamwise velocity and F to the vorticity. The boundary conditions become f =f =0 y=0 f ∝ cy, f ∝ c, y → ∞ An integral to the equation exists, which can be written f

2

− f f − νf

= K = {evaluate y → ∞} = c2

This is as far as we can come analytically and to make further numerical treatment simpler we make the equation non-dimensional with ν and c. Note that they have the dimensions [ν] = and that we can use the length scale dependent variables as L2 , T [c] = 1 T √ νc to scale the independent and the

ν/c and the velocity scale y ν/c f (y) √ νc

η F (η)

= =

This results in the following non-dimensional equation F with the boundary conditions F = F = 0, η = 0 F ∝ η, η→∞ In ﬁgure 2.9 a solution of the equation is shown. It can be seen that diﬀusion of vorticity from the wall is balanced by transport downward by v and outward by u. The thin layer of vortical ﬂow close to the wall has the thickness ∼ ν/c.

2

− FF − F

=1

**Stretching and tilting of vortex lines
**

In three-dimensions the vorticity is not restricted to a single non-zero component and the three-dimensional vorticity equation governing the vorticity vector becomes ∂ωi ∂ωi ∂ui 1 + uj = ωj + ∂t ∂xj ∂xj Re

advection of vorticity 2

ωi

diﬀusion of vorticity

η=

y = η 4νt t η y ˜ (η)·u ωz = ω√πνt0 t

√ 4νt u √ u0

dr

C(t)

Figure 2.10: A closed material curve. The complex motion of the vorticity vector for a full three-dimensional ﬂow can be understood by an analogy to the equation governing the evolution of a material line, discussed in the ﬁrst chapter. The equation for a material line is D ∂ui dri = drj Dt ∂xj Note that if we disregard the diﬀusion term, i.e. the last term of the vorticity equation, the two equations are identical. Thus we can draw the following conclusions about the development of the vortex vector i) Stretching of vortex lines (line ω) produce ωi like stretching of dri produces length

ii) Tilting vortex lines produce ωi in one direction at expense of ωi in other direction

**Helmholz and Kelvin’s theorems
**

The analogy with the equation for the material line directly give us a proof of Helmholz theorem. We have Theorem 1. Helmholz theorem for inviscid ﬂow: Vortex lines are material lines in inviscid ﬂow. A second theorem valid in inviscid ﬂow is Kelvin’s theorem. Theorem 2. Kelvin’s theorem for inviscid ﬂow: Circulation around a material curve is constant in inviscid ﬂow. Proof. We show this by considering the circulation for a closed material curve, see ﬁgure 2.10, Γm =

C(t)

ui dri

We can evaluate the material time derivative of this expression with the use of Lagrangian coordinates as follows DΓm Dt = D ui dri Dt C(t) Lagrange coordinates ˆ ˆ ui = ui (x0 , t) = ui (x0 (m), t) i i 0 ˆ 0 ˆ) ri = ri (xi , t) = ri (xi (m), t d ∂ri dm u ˆ m i ∂m dt ∂ui ∂ri ∂ ∂ri dm + ui ˆ ∂m ˆ ∂ t ∂m m ∂t m

= = =

dm

**The last term of this expression will vanish since ui
**

m

∂ ∂ri ˆ ∂ t ∂m

dm =

m

ui

∂ui dm = ∂m

m

∂ ∂m

1 ui ui 2

dm = 0

y, √ u0 v y = η 4νt t η y ˜ √ ωz = ω(η)·u0 πνt t dx

u = (u, v)

dy

x, u Figure 2.11: A streamline which is tangent to the velocity vector. Thus we ﬁnd the following resulting expression DΓm Dt =

C(t)

Dui dri Dt −

C(t)

=

C(t)

∂p 1 + ∂xi Re

2

2

ui dri Re → ∞

=

1 Re

ui dri → 0,

Which shows that the circulation for a material curve changes only if the viscous force = 0

Streamfunction

We end this section with the deﬁnition and some properties of the streamfunction. To deﬁne the streamfunction we need the streamlines which are tangent to the velocity vector. In two dimensions, see ﬁgure 2.11, we have dx u dx dy = ⇐⇒ = ⇐⇒ u dy − v dx = 0 dy v u v and in three dimensions we can write dx dy dz = = u v w For two-dimensional ﬂow we can ﬁnd the streamlines using a streamfunction. We can introduce a streamfunction ψ = ψ(x, y) with the property that the streamfunction is constant along streamlines. We can make the following derivation dψ = ∂ψ ∂ψ dx + dy ∂x ∂y ∂ψ ∂ψ , v=− and check for consistency = assume u = ∂y ∂x = u dy − v dx = 0, on streamlines

We can also quickly check that the deﬁnition given above satisﬁes continuity. We have ∂u ∂v ∂2ψ ∂2ψ + = − =0 ∂x ∂y ∂x∂y ∂y∂x Another property of the streamfunction is that the volume ﬂux between two streamlines can be found from the diﬀerence by their respective streamfunctions. We use ﬁgure 2.12 and ﬁgure 2.13 in the following derivations of the volume ﬂux:

ωz =

**t n dS u · n∆t x y u0 p0 y η = √4νt y = η 4νt t η y ˜ (η)·u ωz = ω√πνt0 t dy √
**

u u0

√ πνtu

B Ψ = ΨB S n A Ψ = ΨA Figure 2.12: Integration paths between two streamlines.

ds

nx

ny

|n| = 1

− dx Figure 2.13: The normal vector to a line element.

B

QAB

=

A B

ui ni ds (nx u + ny v) ds

A

= = =

A

1 ds = , dy nx

B

−

ds 1 = dx ny

B A

(u dy − v dx) =

dψ = ψB − ψA

2.3

Potential ﬂow

Flows which can be found from the gradient of a scalar function, a so called potential, will be dealt with in this section. Working with the velocity potential, for example, will greatly simplify calculations, but also restrict the validity of the solutions. We start with a deﬁnition of the velocity potential, then discuss simpliﬁcations of the momentum equations when such potentials exist, and ﬁnally we deal with two-dimensional potential ﬂows, where analytic function theory can be used.

Velocity potential

Assume that the velocity can be found from a potential. In three dimensions we have ui = ∂φ ∂xi or u = φ

with the corresponding relations in two dimensions u = v =

∂φ ∂x ∂φ ∂y

There are several consequences of this deﬁnition of the velocity ﬁeld.

i) Potential ﬂow have no vorticity, which can be seen by taking the curl of the gradient of the velocity potential, i.e. ∂uk ∂2φ ωi = ijk = ijk =0 ∂xj ∂xj ∂xk ii) Potential ﬂow is not inﬂuenced by viscous forces, which is seen by the relationship between viscous forces and vorticity, derived earlier. We have ∂τij =µ ∂xj

2

ui = −µ

ijk

∂ωk ∂xj

iii) Potential ﬂow satisﬁes Laplace equation, which is a consequence of the divergence constraint applied to the gradient of the velocity potential, i.e. ∂ui ∂2φ = = ∂xi ∂xi ∂xi

2

φ=0

Bernoulli’s equation

For potential ﬂow the velocity ﬁeld can thus be found by solving Laplace equation for the velocity potential. The momentum equations can then be used to ﬁnd the pressure from the so called Bernoulli’s equation. We start by rewriting the momentum equation ∂ui ∂ui 1 ∂p + uj =− +ν ∂t ∂xj ρ ∂xi using the alternative form of the non-linear terms, i.e. uj and obtain the following equation ∂ui ∂ + ∂t ∂xi 1 p uk uk + + gx3 2 ρ =

ijk uj ωk 2

ui − gδi3

∂ui ∂ = ∂xj ∂xi

1 uk uk 2

−

ijk uj ωk

+ν

ijk

∂ωk ∂xj

This equation can be integrated in for two diﬀerent cases: potential ﬂow (without vorticity) and stationary inviscid ﬂow with vorticity.

∂φ i) Potential ﬂow. Introduce the deﬁnition of the velocity potential ui = ∂xi into the momentum equation above, we have ∂ ∂φ 1 p + uk uk + + gx3 = 0 ⇒ ∂xi ∂t 2 ρ

p ∂φ 1 2 + (u1 + u2 + u2 ) + + gx3 = f (t) 2 3 ∂t 2 ρ ii) Stationary, inviscid ﬂow with vorticity. Integrate the momentum equations along a streamline, see ﬁgure 2.14. We have ti ∂ ∂xi 1 p uk uk + + gx3 2 ρ ds =

ijk

ui uj ωk ds = 0 u

⇒

1 2 p (u + u2 + u2 ) + + gx3 = constant along streamlines 2 3 2 1 ρ Thus, Bernoulli’s equation can be used to calculate the pressure when the velocity is known, for two diﬀerent ﬂow situations. Example 4. Ideal vortex: Swirling stationary ﬂow without vorticity. The velocity potential u = polar coordinates can be written ur = ∂φ ∂r u θ = 1 ∂φ r ∂θ φ in

h u t x1 xη 2 pω(η)·uy 0 > p0 1 ωz = ˜√πνt 1 p u t n t= u u dS u · n∆t Figure 2.14: Integration of the momentum equation along a streamline. x y u0 r p0 y η = √4νt y = η 4νt t η y ˜ √ ωz = ω(η)·u0 πνt t Figure 2.15: Streamlines for the ideal vortex. The velocity potential satisﬁes Laplace equation, which in polar coordinates becomes

2

√

u u0

φ=

1 ∂ r ∂r

1 ∂φ r ∂r

+

1 ∂2φ =0 r ∂θ2

The ﬁrst term vanishes since

∂φ ∂r

= ur = 0. Integrate the resulting equation twice and we ﬁnd φ = Cθ + D

The constant D is arbitrary so we can set it to zero. Using the deﬁnition of the circulation we ﬁnd uθ = C , r C= Γ 2π

We can easily ﬁnd the streamfunction for this ﬂow. In polar coordinates the streamfunction is deﬁned ur = 1 ∂ψ r ∂θ Note that the streamfunction satisﬁes continuity, i.e. potential to ﬁnd the streamfunction, we have ∂ψ Γ =− ∂r 2πr ⇒ u θ = − ∂ψ ∂r · u = 0. We use the solution from the velocity ψ=− Γ ln r 2π

The streamfunction is constant along streamlines resulting in circular streamlines, see ﬁgure 2.15 We can now use Bernoulli’s equation to calculate the pressure distribution for the ideal vortex as 1 2 u + p = p∞ 2 θ ⇒ p = p∞ − Γ2 8π 2 r2

Note that the solution has a singularity at the origin.

**Two-dimensional potential ﬂow using analytic functions
**

For two-dimensional ﬂows we can use analytic function theory, i.e. complex valued functions, to calculate the velocity potential. In fact, as we will show in the following. Any analytic function represents a twodimensional potential ﬂow.

We start by showing that both the velocity potential and the streamfunciton satisfy Laplace equation. Velocity potential φ for two-dimensional ﬂow is deﬁned u = ∂φ ∂x Using the continuity equation gives ∂u ∂v + = ∂x ∂y u = v v = ∂φ ∂y

2

φ=0

The streamfunction ψ is deﬁned

∂ψ ∂y ∂ψ ∂x

2

= −

By noting that potential ﬂow has no vorticity we ﬁnd −ω = − ∂v ∂y + = ∂x ∂y ψ=0

Thus, both the velocity potential φ and the streamfunction ψ satisfy Laplace equation. We can utilize analytic function theory by introducing the complex function F (z) = φ(x, y) + iψ(x, y) where z = x + iy = reiθ = r(cos θ + i sin θ) which will be denoted the complex potential. We now recall some useful results from analytic function theory. F (z) exists and is unique ⇐⇒ F (z) is an analytic function

The fact that a derivative of a complex function should be unique rests on the validity of the Cauchy– Riemann equations. We can derive those equations by requiring that a derivative of a complex function should be the same independent of the direction we take the limit in the complex plane, see ﬁgure 2.16. We have 1 F (z) = lim [F (z + ∆z) − F (z)] ∆z→0 ∆z 1 = lim [φ(x + ∆x, y) − φ(x, y) + iψ(x + ∆x, y) − iψ(x, y)] ∆x→0 ∆x 1 = lim [φ(x, y + ∆y) − φ(x, y) + iψ(x, y + ∆y) − iψ(x, y)] i∆y→0 i∆y ∂φ = ∂ψ real part ∂x ∂y ⇒ ∂ψ 1 ∂φ ∂ψ ∂φ i = ⇒ =− imaginary part ∂x i ∂y ∂x ∂y Thus we have the Cauchy–Riemann equations as a necessary requirement for a complex function to be analytic, i.e. ∂φ ∂ψ ∂φ ∂ψ = and =− ∂x ∂y ∂y ∂x Using the Cauchy-Riemann equations we can show that the real φ and imaginary part ψ of the complex potential satisfy Laplace equations and thus that they are candidates for identiﬁcation with the velocity potential and streamfunction of a two-dimensional potential ﬂow. We have 2 ∂ ∂ψ ∂ ∂ψ ∂2φ ∂ φ = = =− 2 ∂x2 ∂x ∂y ∂y ∂x ∂y 2 ∂ ψ ∂ = 2 ∂y ∂y ∂φ ∂x = ∂ ∂x ∂φ ∂y =− ∂ 2ψ ∂x2

η=

y = η 4νt t η y ω (η)·u0 ˜ √ ωz = πνt t x

√y 4νt u √ u0

u0 p0

iy

i∆y

∆x

Figure 2.16: The complex plane and the deﬁnitions of analytic functions. In addition to satisfying the Laplace equations, we also have to show that the Cauchy-Riemann equations are consistent with the deﬁnition of the velocity potential and the streamfunciton. The Cauchy–Riemann equations are ∂φ ∂ψ u= = ∂x ∂y which recovers the deﬁnitions of the velocity potential and the streamfunction. Thus any analytic complex function can be identiﬁed with a potential ﬂow, where the real part is the velocity potential and the imaginary part is the streamfunction. We can now deﬁne the complex velocity as the complex conjugate of the velocity vector since W (z) = dF ∂φ ∂ψ = +i = u − iv dz ∂x ∂x v = ∂φ = − ∂ψ ∂y ∂x

It will be useful to have a similar relation for polar coordinates, see ﬁgure 2.17. In this case we have u = ur cos θ − uθ sin θ v = ur sin θ + uθ cos θ W (z) = u − iv = [ur (r, θ) − iuθ (r, θ)] e−iθ

cos θ−i sin θ

We will now take several examples of how potential velocity ﬁelds can be found from analytic functions. Example 5. Calculate the velocity ﬁeld from the complex potential F = U e−iα z. The complex velocity becomes W (z) = F = U e−iα = U (cos α − i sin α) which gives us the solution for the velocity ﬁeld in physical variables, see ﬁgure 2.18, as u = U cos α v = U sin α Example 6. Calculate the velocity ﬁeld from the complex potential F = Az n . In polar coordinates the complex potential becomes Arn einθ = Arn cos(nθ) + iAr n sin(nθ) Thus the velocity potential and the streamfunction can be identiﬁed as φ = Arn cos(nθ) ψ = Arn sin(nθ)

**u · n∆t P2 x xy 1 x2 u0 x3 p0 √y 3 R η = 4νt u R2 u1 √ R0 y y = η 4νt r3 t 2 v rη 1 ry π dϕ12 2 + (η)·u ˜ ωz = ω√πνt0 u ˜ V (t) S (t) t uθ V (t + ∆t) − V (t) S (t + ∆t) ur u u n V (t + ∆t) r p T0 , U 0 θ Tw x L T0 Figure 2.17: The deﬁnition of polar coordinates. U0 Gradient ﬁelds wi
**

∂p ∂xi

**ui Divergence free vectorﬁelds // to boundary L h x1 x2 p0 > p 1 p1 u n dS u · n∆t x y u0 p0 y η = √4νt y = η 4νt t η y ˜ (η)·u ωz = ω√πνt0 t Figure 2.18: Constant velocity at an angle α. √
**

u u0

α

xi y =u · n∆t η 4νt Ψ=0 ˆ u i dt x t 0 dxiy η Ψˆ= 0 dui uy d0 t ˜ (η)·u 0 ˆ p rω√ dt 0 i ωz = πνt y η = √4νt ˆ dri dt ut π u1 √ P0 n y = η 4νt P2 Ψ=0 t x1 η x2 Figure 2.19: Flow towards a corner. xy 3 ω (η)·u0 ˜ √ ωz = πνt 3 R R2 t Ψ=0 R1 r3 r2Figure 2.20: Uniform ﬂow over a ﬂat plate. r1 π + dϕ12 2 and we can calculate the complex velocity as V (t) S (t) (z) = F = nAz n−1 W V (t + ∆t) − V (t) = nArn−1 einθ e−iθ S (t + ∆t) = nArn−1 [cos(nθ) + i sin(nθ)]e−iθ u n This gives us the solution for the velocity in polar coordinates as V (t + ∆t) ur = nArn−1 cos(nθ) p uθ = −nArn−1 sin(nθ) T0 , U 0 Tw This is a ﬂow which approaches a corner, see ﬁgure 2.19. The streamfunction is zero on the walls and on L the stagnation point streamline. By solving for the zero streamlines we can calculate the angle between a T0 wall aligned with the x-axis and the stagnation point streamline. We have U0 Gradient ﬁelds ψ = Arn sin(nθ) = 0 ⇒ θ = πk wi n

∂xi Figure 2.20 shows the uniform ﬂow over a ﬂat plate which results when n = 1. ui Figure 2.21 shows the stagnation point ﬂow when n = 2. Here the analysis can be done in Cartesian Divergence coordinates free vectorﬁelds // to boundary L F = Az 2 h W (z) = F = 2Az x1 x2 = 2A(x + iy) p0 > p 1 u = 2Ax p1 ⇒ v = −2Ay u n Figure 2.22 shows the ﬂow towards a wedge when 1 ≤ n ≤ 2. The velocity along the wedge edge (x axis dS in ﬁgure 2.22) is u · n∆t ur = nArn−1 cos(nθ) ⇒ {θ = 0} ⇒ u = ur = nArn−1 = Cxn−1 x The wedge angle is y 2π n−1 u0 ϕ = 2π − = 2π n n p0 y η = √4νt ∂p

Figure 2.21: Stagnation point ﬂow when n = 2.

!! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !! !!

y = η 4νt t η y ω (η)·u0 ˜ √ ωz = πνt t

√

u u0

L y = η 4νt h t x1 η x2 py > p1 0 ϕ ˜ (η)·u p1 ωz = ω√πνt0 u t n dS u · 2.22: Figuren∆t The ﬂow over a wedge when 1 ≤ n ≤ 2. x y u0 p0 y η = √4νt √ y = η 4νt t η y ˜ (η)·u ωz = ω√πνt0 t Figure 2.23: Line source. We now list several other examples of potential ﬂow ﬁelds obtained from analytic functions. Example 7. Line source. F (z) = = = m ln z 2π m ln(reiθ ) 2π m [ln r + iθ] 2π √

u u0

u u0

The velocity potential and the streamfunction become φ = m ln r 2π The complex velocity can be written as ψ= mθ 2π

W (z) = F =

m m −iθ = e 2πz 2πr

which gives the solution in polar coordinates, see ﬁgure 2.23, as ur = m 2πr uθ = 0 We can now calculate the volume ﬂux from the source as

2π

Q=

0

m r dθ = m 2πr

Example 8. Line vortex. F (z) = −i = Γ ln z 2π

Γ [θ − i ln r] 2π

η=

y = η 4νt t η y ˜ √ ωz = ω(η)·u0 πνt t Figure 2.24: Ideal vortex. The velocity potential and the streamfunction become φ= Γ θ 2π ψ = − Γ ln r 2π W (z) = F = −i Γ −iθ e 2πr

√y 4νt u √ u0

x y u0 p0

The complex velocity can be written as

which gives the solution in polar coordinates, see ﬁgure 2.24, as ur = 0 uθ = Γ 2πr F (z) = m πz Example 9. The dipole has the complex potential

**Example 10. The potential ﬂow around a cylinder has the complex potential F (z) = U z + U
**

2 r0 z

**Example 11. The potential ﬂow around a cylinder with circulation has the complex potential F (z) = U z + U
**

2 r0 Γ z −i ln z 2π r0

Example 12. Airfoils. The inviscid ﬂow around some airfoils can be calculated with conformal mappings of solutions from cylinder with circulation.

2.4

Boundary layers

For ﬂows with high Reynolds number Re, where the viscous eﬀects are small, most of the ﬂow can be considered inviscid and a simpler set of equations, the Euler equations, can be solved. The Euler equations are obtained if Re → ∞ in the Navier-Stokes equations. However, the highest derivatives are present in the viscous terms, thus a smaller number of boundary conditions can be satisﬁed when solving the Euler equations. In particular, the no ﬂow condition can be satisﬁed at a solid wall but not the no slip condition. In order to satisfy the no slip condition we need to add the viscous term. Thus, there will be a layer close to the solid walls where the viscous terms are important even for very high Reynolds number ﬂow. This layer will be very thin and the ﬂow in that region is called boundary layer ﬂow.

ωz =

}δ t ω = η 4νt ≈0 y U→ t η L y ω(η)·u0 ˜ ωz = √πνt Figure 2.25: Estimatet of the viscous region for an inviscid vorticity free inﬂow in a plane channel.

ω (η)·u0 ˜ η √ πνt

=

0 √y 4νt u √ u0

L v y, δ ω≈0 U→

}

U

δ L x, u

Figure 2.26: Boundary layer proﬁle close to a solid wall. We begin this section by estimating the thickness of the viscous region in channel with inviscid inﬂow (zero vorticity), see ﬁgure 2.25. By analogy to the Stokes solution for the instantaneously starting plate, we can estimate the distance √ the vorticity has diﬀused from the wall toward the channel centerline as δ ∼ νts . Where ts is time of a L ﬂuid particle with velocity U has travelled length L, i.e. ts = . Thus we can estimate δ/L as U √ 1/2 δ νts 1 νL ν 1/2 1 → 0 as Re → ∞ ⇒ = = = =√ L L L U UL Re We will now divide the ﬂow into a central/outer inviscid part and a boundary layer part close to walls, where the no slip condition is fulﬁlled. Since we expect this region to be thin for high Reynolds numbers, we expect large velocity gradients near walls. We choose diﬀerent length scales inside boundary layer and in inviscid region as indicated in ﬁgure 2.26, as Inviscid : U velocity scale for u, v : L length scale for x, y BL : : : : U α L δ velocity scale for velocity scale for length scale for length scale for u v x y

The pressure scale is assumed to be ρU 2 for both the inner and the outer region. The unknown velocity scale for the normal velocity α is determined using continuity ∂u ∂v + =0 ∂x ∂y

O

U L

O

α δ

which implies that U α = L δ ⇒ α= δ U L

In this manner both terms in the continuity equation have the same size. Let us use these scalings in the steady momentum equations and disregard small terms. We obtain the following estimates for the size of the terms in the normal momentum equation

↑y:

∂v ∂x 1 δU U· L L u δ L

2

+

∂v ∂y δU 1 δU · · L δ L v δ L

2

= −

1 ∂p ρ ∂y

U2 δ

+

ν

ν L2

∂2v ∂x2 ·

δU L

+

∂2v ∂y 2 ν δU δ2 L ν 1 Re

1

1 Re

δ L

2

where the last line was obtained by dividing all the terms on the second line by U 2 /δ. When R → ∞ and δ/L is small, only pressure term O(1), which implies that ∂p =0 ∂y ⇒ p = p(x)

Thus, the pressure is constant through the boundary layer, given by the inviscid outer solution. Next we estimate the size of the terms in the streamwise momentum equation as →x: u ∂u ∂x + v ∂u ∂y = − 1 ∂p ρ ∂x U2 L 1 + ν ∂2u ∂x2 + ν ∂2u ∂y 2

U2 L 1

δU U · L δ 1

ν ·U L2 1 Re

ν U δ2 1 · Re L δ

2

The ﬁrst three terms are order one and the fourth term negligible when the Reynolds number become large. In order to keep the last term, the only choice that gives a consistent approximation, we have to assume that 1 δ ∼√ L Re which is the same estimate as we found for the extent of the boundary layer in the channel inﬂow case. Thus we are left with the parabolic equation in x since the second derivative term in that direction is neglected. We have u ∂u ∂u 1 ∂p ∂2u +v =− +ν 2 ∂x ∂y ρ ∂x ∂y

In summary, the steady boundary layer equations for two-dimensional ﬂow is 2 u ∂u + v ∂u = − 1 ∂p + ν ∂ u ∂x ∂y ρ ∂x ∂y 2 ∂u ∂v + =0 ∂x ∂y

with the initial (IC) and boundary conditions (BC) IC : u (x = x0 , y) = uin (y) BC : u (x, y = 0) v (x, y = 0) u (x, y → ∞) = = =

0 0 ue (x) outer inviscid ﬂow

where the initial condition is at x = x0 and the equations are marched in the downstream direction. There are several things to note about these equations. First, vin cannot be given as its own initial condition. By using continuity one can show that once the u0 is given the momentum equation can be written in the form − ∂v + f (y) v = g (y) ∂y

which can be integrated in the y-direction to give the initial normal velocity. Second, the system of equations is a third order system in y and thus only three boundary conditions can be enforced. We have the no ﬂow and the no slip conditions at the wall and the u = ue in the free stream. Thus no boundary condition for v in the free stream can be given.

Figure 2.27: Boundary layer ﬂow over a ﬂat plate with zero pressure gradient. Third, ue (x) = uinv (x, y = 0), i.e. the boundary layer solution in the free stream approaches the inviscid solution evaluated at the wall, since the thickness of the boundary layer is zero from the inviscid point of view. Fourth, the pressure gradient term in the momentum equation can be written in terms of the edge velocity ue using the Bernoulli equation. We have 1 p (x) + ρu2 (x) = const 2 e which implies that − 1 dp due = ue ρ dx dx

**Blasius ﬂow over ﬂat plate
**

Consider the boundary layer equations for a ﬂow over a ﬂat plate with zero pressure gradient, i.e. the outer inviscid ﬂow is ue (x) = u0 . The boundary layer equations become 2 u ∂u + v ∂u = ν ∂ u ∂x ∂y ∂y 2 ∂u ∂v + =0 ∂x ∂y

IC : BC :

u (x = x0 , y) u (x, y = 0) v (x, y = 0) u (x, y → ∞)

= = = =

u0 0 0 u0

We introduce the two-dimensional stream function Ψ u= ∂Ψ , ∂y v=− ∂Ψ ∂x

which implicitly satisﬁes continuity. The boundary layer equations become ∂Ψ ∂ 2 Ψ ∂Ψ ∂ 2 Ψ ∂3Ψ · − · =ν 3 ∂y ∂x∂y ∂x ∂y 2 ∂y with appropriate initial and boundary conditions. Similarity solution We will now look for a similarity solution. Let u= ∂Ψ = u0 f (η) , ∂y η= y δ (x)

which by integration in the y-direction implies Ψ (x, y) = u0 δ (x) f (η) We can now evaluate the various terms in the momentum equation. We give two below v=− ∂Ψ ∂η = −u0 δ f + δf ∂x ∂x y = −u0 δ f − f δ δ = u0 (ηf − f ) δ

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ "#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"#"# "$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$#"$# #################################"$"$

U→ y, v x, u L U δ

u0

u0

L δ ω≈0 U→ y, v x, u η=√ yL νx/u0 U δ u0

}

5

Blasius proﬁle

4

3

2

1

0

0

0.2

0.4

f =

u u0

0.6

0.8

1

Figure 2.28: Similarity solution giving the Blasius boundary layer proﬁle.

∂2Ψ η = −u0 f δ ∂x∂y δ If these and the other corresponding terms are introduced into the momentum equation we have the following equation for f −u2 f f 0 which can be simpliﬁed to u0 δδ ff = 0 ν For a similarity solution to be possible, the coeﬃcient in front of the f f term has to be constant. This implies f + u0 δδ 1 = ν 2 which implies that δ (x) = ⇒ δδ dx = 1 2 ν dx u0 ⇒ 1 2 1 νx δ = 2 2 u0 η f δ f δ + u2 (ηf − f ) = νu0 2 0 δ δ δ

νx , where we have chosen the constant of integration such that δ = 0 u0 when x = 0. We are left with the Blasius equation and boundary conditions f BC : 1 + ff = 0 2

f (0) = f (0) = 0 f (∞) = 1

The numerical solution to the Blasius equation gives the self-similar velocity proﬁle seen in ﬁgure 2.28. The boundary layer thickness can be evaluated as the height where the velocity has reached 99% of the free stream value. We have f (η99 ) = 0.99 ⇒ η99 = δ99

νx u0

= 4.9

⇒ √ x.

δ99 = 4.9

νx u0

where it can be seen that the boundary layer thickness grows as Displacement thickness and skin friction coeﬃcient

Another measure of the boundary layer thickness is the displacement δ ∗ of the wall needed in order to have the same volume ﬂux of the inviscid ﬂow as for the ﬂow in the boundary layer. With the help of ﬁgure 2.29 we can evaluate this as

y0 0 y y

u dy ≡

u0 dy =

δ∗ 0

u0 dy − δ ∗ u0

which implies that

ωz =

√ πνt

Figure 2.29: The displacement thickness δ ∗ ⇐= Pressure force

Figure 2.30: Boundary layer separation caused by a pressure force directed upstream, a so called adverse pressure gradient.

If we evaluate this expression for the Blasius boundary layer we have δ ∗ = 1.72

A quantity which is of large importance in practical applications is the skin friction coeﬃcient. The skin friction at the wall is τw = µ ∂u ∂y = µu0 ρu2 0.332ρu2 0 0 f (0) = √ f (0) = u0 x δ Rex ν

where the last term is the expression evaluated for Blasius ﬂow. This can be used to evaluate the skin friction coeﬃcient as Cf = τw 1 2 2 ρu0 0.664 =√ Rex

where the last terms again is the Blasius value.

**Boundary layer separation
**

∂p A large adverse pressure gradient ∂x > 0 may cause a back ﬂow region close to wall, see ﬁgure 2.30. The boundary layer equations cannot be integrated past a point of separation, since they become singular at that point. This can be seen from the following manipulations of the boundary layer equations. If we take the momentum equation

take the normal derivative and use continuity we have u ∂2u ∂2u ∂3u +v 2 =ν 3 ∂x∂y ∂y ∂y

**If we again take the normal derivative and use continuity we can write this as 1 ∂ 2 ∂x ∂u ∂y
**

2

This expression evaluated at the wall, with the use of the boundary conditions, become

02102102102102102102102102102102110210210210210210210210210210210210211021021021021021111111111111111111111102 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02

u f = u0 Blasius proﬁle δ∗

νx/u0

Separation point

δ∗ =

0

∞

1−

u u0

dy =

y=0

u

∂u ∂u 1 dp ∂2u +v =− +ν 2 ∂x ∂y ρ dx ∂y

+u

∂3u ∂u ∂ 2 u ∂3u ∂4u − +v 3 =ν 4 2 2 ∂x∂y ∂x ∂y ∂y ∂y

) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) (&(&(&(&(&&(&(&(&(&(&(&(&&(&(&(&(&(& &&&&&()&&&&&&&&()&&&&&&()

' ' ' ' ' ' ' ' ' ' ' ' ' ' ' %&%&%&%&&%&%&%&&%&%&%&&%&%&&%&%&%& &&&&%'&&&&%'&&&&%'&&&%'&&&&%'

t L η=√ y νx/u0 }δ u f = u0 ω≈0 Blasius proﬁle U→ y, v x, u L U δ u0 η=√ y

U δ u0

δ∗

x

νx u0

∞ 0

[1 − f (η)] dη

νx u0 .

1 ∂ 2 ∂x

∂u ∂y

2

=ν

y=0

∂4u ∂y 4

y=0

±κ2

**which can be integrated to yield ∂u ∂y
**

2

y=0

= ±κ2 x + C

∂u ∂y

At the point of separation x = xs , i.e. where ∂u ∂y

=0

y=0

this expression is

y=0

√ = κ xs − x

We can see that the boundary layer equations are not valid beyond point of separation since we have a square root singularity at that point. In addition, the point of separation is the limiting point after which there is back ﬂow close to the wall. This means that there is information propagating upstream, invalidating the downstream parabolic nature of the equations assumed by the downstream integration of the equations. In practical situations it is very important to be able to predict the point of separation, for example when an airfoil stalls and experience a severe loss of lift.

2.5

Turbulent ﬂow

**Reynolds average equations
**

Turbulent ﬂow is inherently time dependent and chaotic. However, in applications one is not usually interested in knowing full the details of this ﬂow, but rather satisﬁed with the inﬂuence of the turbulence on the averaged ﬂow. For this purpose we deﬁne an ensemble average as 1 Ui = ui = lim N →∞ N

(n) N

ui

n=1

(n)

where each member of the ensemble ui is regarded as an independent realization of the ﬂow. We are now going to derive an equation governing the mean ﬂow Ui . Divide the total ﬂow into an average and a ﬂuctuating component ui as ui = ui + ui p=p+p

and introduce into the Navier-Stokes equations. We ﬁnd ∂ui ∂ui ∂ui ∂u ∂ui ∂u 1 ∂p 1 ∂p + + uj + uj i + uj + uj i = − − +ν ∂t ∂t ∂xj ∂xj ∂xj ∂xj ρ ∂xi ρ ∂xi We take the average of this equation, useing ui = 0 which gives ∂ui ∂ui 1 ∂p + uj =− +ν ∂t ∂xj ρ ∂xi ∂ui ∂xi = 0 ∂ ui uj ∂xj u j ui = u i · uj = 0

2

ui + ν

2

ui

2

ui −

where the average of the continuity equation also has been added. Now, let U i = ui , as above, and drop the . We ﬁnd the Reynolds average equations

where ui uj is the Reynolds stress. Thus the eﬀect of the turbulence on the mean is through an additional stress Rij which explicitly written out is u2 [Rij ] = [ui uj ] = uv uw uv v2 vw uw vw w2

∂Ui 1 ∂P ∂ ∂Ui + Uj =− + (τij − ui uj ) ∂t ∂xj ρ ∂xi ∂xj ∂Ui ∂xi

The diagonal component of this stress is the kinetic energy of the turbulent ﬂuctuations k = Rii /2 = ui ui /2 = 1 2 u + v2 + w2 2

Turbulence modelling

One may derive equations for the components of the Reynolds stress tensor. However, they would involve additional new unknowns, and the equations for those would in turn involve even more unknowns. This closure problem implies that if one cannot calculate the complete turbulent ﬂow, but is only interested in the mean, the eﬀect of the Reynolds stress components on in the Reynolds average equations have to be modelled. The simplest consistent model is to introduce a turbulent viscosity as a proportionality constant between the Reynolds stress components and the strain or deformation rate tensor, in analogy with the stress-strain relationship in for Newtonian ﬂuids. We have Rij = 2 kδij − 2νT E ij 3

The ﬁrst term on the right hand side is introduced to give the correct value of the trace of R ij . Here the deformation rate tensor is calculated based on the mean ﬂow, i.e. E ij = 1 2 ∂Ui ∂Uj 2 ∂Ur δij + − ∂xj ∂xi 3 r ∂x

where we have assumed incompressible ﬂow. Introducing this into the Reynolds average equations gives ∂Ui ∂ ∂Ui + Uj = − ∂t ∂xj ∂xj P 2 1 ∂P + k δij + 2 (ν + νT ) E ij = − + (ν + νT ) ρ 3 ρ ∂xi

2

Ui

where the last equality assumes that νT is constant, an approximation only true for very simple turbulent ﬂow. It is introduced here only to point out the analogy between the molecular viscosity and the turbulent viscosity. The turbulent viscosity νT must be modelled and most numerical codes which solved the Navier-Stokes equations for practical applications uses some kind of model for νT . For simple shear ﬂows, i.e. velocity ﬁelds of the form U (y), the main Reynolds stress component becomes uv = −νT where, in analogy with kinetic gas theory, νT νT = uT l Here, uT is a characteristic turbulent velocity scale and l is a characteristic turbulent length scale. In kinetic gas theory l would be the mean free path of the molecules and uT the characteristic velocity of the molecules. ∂U ∂y

u f = u0 Blasius proﬁle δ∗ x Separation point ⇐= Pressure force

νx/u0

y

uT l

Figure 2.31: Prantdl’s mixing length l is the length for which a ﬂuid particle displaced in the normal direction can be expected to retain its horizontal momentum. Zero-equation models The simplest model for uT and l is to model them algebraically. In simple free shear ﬂows like jets, wakes and shear layers we can take νT as constant in normal direction or y-direction and to vary in a self-similar manner in x. In wall bounded shear ﬂow we can take l as Prandtl’s mixing length (1920’s). This is a normal distance over which the particle retains its momentum and is depicted in ﬁgure 2.31. We can thus evaluate uT and ﬁnd νT uT = l dU dy ⇒ νT = l 2 dU dy

Deﬁne τw as the shear stress at the wall and uτ = uT = u τ , as Prandtl’s estimation of mixing length we get dU uT uτ = = dy l κy ⇒

κ is the von Karman constant, approximately equal to 0.41. Thus the mixing length is proportional to the distant from the wall and the region close to the wall in a turbulent ﬂow has a logarithmic velocity proﬁle, called the log-region. One-equation models Instead of modelling the characteristic turbulent velocity algebraically, the next level of modelling uses the assumption that 1 ui ui 2 and a diﬀerential equation is used to calculate the turbulent kinetic energy. We can derive such an equation by taking the previously derived equation for Ui + ui and subtracting the equation for U , the Reynolds average equation. We ﬁnd uT = k= ∂ui ∂ui ∂Ui ∂ui ∂ + Uj = −uj − uj + ∂t ∂xj ∂xj ∂xj ∂xj p ∂ui − δij + ν + u i uj ρ ∂xj

We now take the average of the inner product between the turbulent ﬂuctuations u i and the above equation, and ﬁnd ∂ ∂t 1 ui ui 2 ∂ ∂xj 1 ui ui 2 ∂Ui ∂ + ∂xj ∂xj puj ν ∂ 1 − ui ui uj + ui ui ρ 2 2 ∂xj ∂ui ∂ui ∂xj ∂xj

+ Uj

where is the mean dissipation rate of the turbulent kinetic energy. We have now introduced a number of new unknowns which have to be modelled.

5 5 5 5 5 5 5 35 34434344343443434443 43544435444354443544543

l =κ·y = −ui uj −

x

τw /ρ as the skin friction velocity. With

U 1 = ln y + C. uτ κ

−ν

**Second, the turbulent dissipation is modelled purely based on dimension using k and l. We have = cD
**

L2 T .

2 First, we model the Reynolds stress as previously indicated, i.e. ui uj = 3 kδij − νT

∂Ui ∂xj

+

∂Uj ∂xi

.

k

3/2

l

since the dimension of is [ ] = Here cD is a modelling constant. Third, a gradient diﬀusion model is used for the transport terms, i.e. 1 1 νT ∂k − ui ui uj − puj = 2 ρ P rk ∂xj where P rk is the turbulent Prandtl number. For simplicity we write the resulting equation for the turbulent kinetic energy for constant ν T . We ﬁnd Dk i ui uj ∂Uj = (ν + νT ) 2 k − ∂x Dt rate of increase diﬀusion rate generation rate

−

dissipation rate

Two-equation models The most popular model in use in engineering applications is a two equation model, the so called k- -model. This is based on the above model for k but instead of modelling a partial diﬀerential equation is derived governing the turbulent dissipation. This equation has the form as the equation for k, with a diﬀusion term, a generation term and a dissipation term. Many new model constants need to be introduced in order to close the equation. Once k and is calculated the turbulent viscosity is evaluated as a quotient of the two, such that the dimension is correct. We have ⇒ νT = Cµ · where Cµ is a modelling constant. k

2

L T0 U0 Gradient ﬁelds wi

∂p ∂xi

ui Divergence free vectorﬁelds // to boundary L h x1 x2 p0 > p 1 p1 u n dS u · n∆t x 3.1 Finite Volume method on arbitrary grids y u0 Equations with 1st order derivatives: the continuity equation p0 y η = √4νt In ﬁnite volume methods one makes approximations of the diﬀerential equations in integral form. We start u with an equation with only√ u0 order derivatives, the continuity equation. Recall that the integral from of ﬁrst y=η this equation can be written 4νt t η ∂ (uk ) dV = uk nk dS = 0 y ∂xk ω (η)·u0 ˜ V S √ ωz = πνt

Chapter 3

Finite volume methods for incompressible ﬂow

Evaluating the normal vector ni to the curve l for two-dimensional ﬂow, see ﬁgure 3.1, we have the t normal with length dl as n dl = ( dy, − dx). This implies that L }δ ω≈0 uk nk dl = (u dy − v dx) = 0 U→ S S y, v We now apply this to an x, u arbitrary two-dimensional grid deﬁned as in ﬁgure 3.2. L We make use of the following deﬁnitions and approximations U uab δ ui+ 1 ,j ≈ (ui+1,j + ui,j ) /2 = ∆xab = xb − xa 2 u0 y η=√

u f =ab 0= vi+ 1 ,j ≈ (vi+1,j + vi,j ) /2 vu 2 Blasius proﬁle δ∗ y x Separation point ⇐= Pressure force x dx y l uT νx/u0

∆yab = yb − ya

x dl = ( dx, dy) dy

− dx dy n dl = ( dy, − dx) Figure 3.1: Normal vector to the curve l. 59

i − 1, j + 1 c

i, j + 1

i + 1, j + 1

i − 1, j

i, j Aabcd d

i − 1, j − 1

Figure 3.2: Arbitrary grid. Note that the grid lines need not be parallel to the coordinate directions. with analogous expressions for the quantities on the bc, cd and da faces. We can now evaluate the integral associated with the two-dimensional version of the continuity equation for the surface deﬁned by the points abcd. We have

(u dy − v dx)

S

1 ≈ ui+ 2 ,j ∆yab − vi+ 1 ,j ∆xab + ui,j+ 1 ∆ybc − vi,j+ 1 ∆xbc + 2 2 2 1 1 ui− 2 ,j ∆ycd − vi− 1 ,j ∆xcd + ui,j− 1 ∆yda − vi,j− 2 ∆xda 2 2

On a cartesian grid we have the relations ∆xab = ∆xcd = ∆ybc = ∆yda = 0 and ∆ycd = −∆yab , ∆xbc = −∆xda which implies that the approximation of the continuity equation can be written

(u dy − v dx)

S

≈

1 ui+ 2 ,j − ui− 1 ,j ∆yab + vi,j+ 1 + vi,j− 1 ∆xda = 2 2 2

1 1 (ui+1,j − ui−1,j ) ∆yab + (vi,j+1 − vi,j−1 ) ∆xda 2 2 Note that dividing with ∆yab · ∆xda gives a standard central diﬀerence discretization of the continuity equation, i.e. ui+1,j − ui−1,j vi,j+1 − vi,j−1 + =0 2∆xda 2∆yab

**Equations with 2nd derivatives: the Laplace equation
**

We use the Gauss or Greens theorem to derive the integral from of the two-dimensional Laplace equation. We ﬁnd 0=

V

∂2Φ dV = ∂xj ∂xj

S

∂φ nj dS = {2D} = ∂xj

If we approximate this integral in the same manner as for the continuity equation we have ∂Φ ∂x ∂Φ ∂x ∂Φ ∂y ∂Φ ∂y ∂Φ ∂x ∂Φ ∂x ∂Φ ∂y ∂Φ ∂y

1 i+ 2 ,j

∆yab − ∆ycd −

∆xab +

1 i+ 2 ,j

∆xcd +

1 i− 2 ,j

1 i− 2 ,j

87777777876 6 6 6 6 6 6 6 78787878787876 87777777876 6 6 6 6 6 6 6 78787878787876 87777777876 6 6 6 6 6 6 6 78787878787876 87777777876 6 86 86 86 86 86 86 6 87777777876 6 6 6 6 6 6 6 78787878787876 87777777876 6 86 86 86 86 86 86 6 87777777876 6 6 6 6 6 6 6 78787878787876 87777777876 6 6 6 6 6 6 6 78787878787876 878787878787876 7777777876 6 6 6 6 6 6 6

c b d a i, j − 1

S

1 i,j+ 2 1 i,j− 2

b

i + 1, j

a

i + 1, j − 1

∂Φ ∂Φ dy − dx ∂x ∂y

∆ybc − ∆yda −

∆xbc +

1 i,j+ 2

∆xda

1 i,j− 2

First derivatives are evaluated as mean value over adjacent control volumes (areas in the two-dimensional case). We use Gauss theorem over the area deﬁned by the points a b c d , see ﬁgure 3.2. For the threedimensional case we have 1 V

V

∂Φ 1 dV = ∂xj V

Φnj dS ⇒

S

**which in the two-dimensional case is approximated by ∂Φ ∂x where Φ dy ≈ Φi+1,j ∆ya b + Φb ∆yb c + Φi,j ∆yc d + Φ∆yd a
**

abcd

,

1 i+ 2 ,j

∂Φ ∂y

1 i+ 2 ,j

≈

1 Aa b c d

abcd

(Φ dy, −Φ dx)

and the area evaluated as half magnitude of cross products of diagonals, i.e. Aab ≡ Aa b c d = and the value Φa is taken as the average Φa = 1 (Φi,j + Φi+1,j + Φi,j−1 + Φi+1,j−1 ) 4 1 ∆xd b · ∆ya c − ∆yd b · ∆xa c 2

Evaluating the derivatives ∂Φ , ∂Φ along the other cell faces and substituting back we obtain a 9 point ∂x ∂y formula which couples all of the nine points around the point i, j, of the form Ai,j Φi+1,j+1 Fi,j Φi,j−1 + Bi,j Φi+1,j + Gi,j Φi−1,j+1 + Ci,j Φi+1,j−1 + Hi,j Φi−1,j + Di,j Φi,j+1 + Ii,j Φi−1,j−1 + Ei,j Φi,j = 0 +

If we have a total of N = m × n interior nodes, we obtain N equations which in matrix form can be written Ax = b Here xT = (Φ11 , Φ12 , Φ13 , . . . , Φmn ) and b contains the boundary conditions. The matrix A is a N × N with the terms Aij , Bij , etc. as coeﬃcients. Rather than showing the general case in more detail, we instead work out the expression for cartesian grids in two-dimensions. We ﬁnd ∂2Φ ∂2Φ + ∂x2 ∂y 2 ∂Φ ∂x − ∂Φ ∂x dx dy = ∂Φ ∂Φ dy − dx ∂x ∂y ∂Φ ∂y − ∂Φ ∂y ≈ ∆x

1 i,j− 2

∆y +

1 i− 2 ,j

1 i+ 2 ,j+

1 i,j+ 2

We can evaluate the derivatives as ∂Φ ∂x = (Φi+1,j − Φi,j ) ∆y · 1 ∆x · ∆y

1 i+ 2 ,j

and so on for the rest of the terms, which implies that ∂Φ ∂Φ dy − dx ∂x ∂y ≈ (Φi+1,j − 2Φi,j + Φi−1,j ) ∆y ∆x + (Φi,j+1 − 2Φi,j + Φi,j−1 ) ∆x ∆y

Dividing with the area ∆x · ∆y we obtain the usual 5-point Laplace formula, i. e. Φi+1,j − 2φi,j + Φi−1,j Φi,j+1 − 2φi,j + Φi,j−1 + =0 2 ∆x ∆y 2 which of course also can be written in the matrix form shown for the general case above.

a a i − 1, j − 1 i, j − 1 i + 1, j − 1 j

ui,j vi,j pi,j

i

i+1

Figure 3.3: Cartesian ﬁnite-volume grid where the velocity components and the pressure are co-located.

3.2

Finite-volume discretizations of 2D NS

**Co-located Cartesian grid
**

We will apply the ﬁnite volume discretizations to the Navier-Stokes equations in integral form, which in two-dimensions can be written (u dy − v dx) = 0 ∂u dS = − ∂t ∂v dS = − ∂t u2 dy − uv dx + p dy − uv dy − v 2 dx − p dx − 1 Re 1 Re ∂u ∂u dy − dx ∂x ∂y ∂v ∂v dy − dx ∂x ∂y

where we have used the earlier derived result n = ( dy, − dx). We start with a co-located grid which means that the ﬁnite volume used for both velocity components and the pressure is the same, see ﬁgure 3.3. In addition we choose a Cartesian grid in order to simplify the expressions. Recall that the ﬁnite volume discretization of the continuity equation becomes 1 1 (ui+1,j − ui−1,j ) ∆y + (vi,j+1 − vi,j−1 ) ∆x = 0 2 2 which is equivalent to the ﬁnite diﬀerence discretization or ui+1,j − ui−1,j vi,j+1 − vi,j−1 + =0 2∆x 2∆y

The discretization of the streamwise momentum equation needs both ﬁrst and second derivatives. We have seen that the ﬁnite volume discretizations of both are equal to standard second order ﬁnite diﬀerence approximations. Thus we obtain the following result for the u-component

2 (uv)i,j+1 − (uv)i,j−1 u2 ∂ui,j pi+1,j − pi−1,j i+1,j − ui−1,j + + + ∂t 2∆x 2∆y 2∆x

−

1 Re

ui+1,j − 2ui,j + ui−1,j ui,j+1 − 2ui,j + ui,j−1 + ∆x2 ∆y 2

=0

**In a similar manner the discretization for the v-component becomes
**

2 2 (uv)i+1,j − (uv)i−1,j vi,j+1 − vi,j−1 pi,j+1 − pi,j−1 ∂vi,j + + + ∂t 2∆x 2∆y 2∆y

−

1 Re

vi+1,j − 2vi,j + vi−1,j vi,j+1 − 2vi,j + vi,j−1 + 2 ∆x ∆y 2

=0

This simple discretization cannot be used in practice without some kind of modiﬁcation. The reason is that one solution to these equations are in the form of spurious checkerboard modes. Consider the following solution to the discretized equations ui,j = vi,j = (−1)i+j · g (t) , p = (−1)i+j , ∆x = ∆y

ui−1/2,j

i−

control volume for divergence condition. control volume for streamwise momentum equation (u) . control volume for normal momentum equation (v) .

Figure 3.4: Cartesian ﬁnite-volume grid where the velocity components and the pressure uses staggered grids. which satisﬁes the divergence constraint. If it is input into the equation for u and v we ﬁnd the expression 8 dg + g=0 dt Re · ∆x2 ⇒ g = e− Re·∆x2

8t

Thus we have a spurious solution consisting of checkerboard modes which are damped slowly for velocity and constant in time for the pressure. This mode will corrupt any true numerical solution and yields this discretization unusable as is. These modes are a result of the very weak coupling between nearby ﬁnite volumes or grid points, and can be avoided if they are coupled by one sided diﬀerences or if they are damped with some type of artiﬁcial viscosity.

**Staggered cartesian grid
**

To eliminate the problem with spurious checkerboard modes, we can use a staggered grid as in ﬁgure 3.4, where the control volumes for the streamwise, spanwise and pressure are diﬀerent. The control volume for the continuity equation is centered abound the pressure point and the discretization becomes

1 ui+ 2 ,j − ui− 1 ,j ∆y + vi,j+ 1 − vi,j− 1 ∆x = 0 2 2 2 1 ui+ 1 ,j − ui− 2 ,j 2 1 vi,j+ 2 − vi,j− 1 2

or

The control volume for the streamwise velocity is centered around the streamwise velocity point and the discretization becomes

2 (uv)i+1/2,j+1/2 − (uv)i+1/2,j−1/2 u2 ∂ui+1/2,j pi+1,j − pi,j i+1,j − ui,j + + + − ∂t ∆x ∆y ∆x

1 ui+3/2,j − 2ui+1/2,j + ui−1/2,j 1 ui+1/2,j+1 − 2ui+1/2,j + ui+1/2,j−1 − =0 2 Re ∆x Re ∆y 2

@9@9@@9@9@9@@ A A A A A 9A@@@9A@@@@9A@9A @9@9@@9@9@9@@ A A A A A 9A@@@9A@@@@9A@9A @9@9@@9@9@9@@ A A A A A 9A@@@9A@@@@9A@9A 9A 9A 9A 9A 9A 9A @@@9A@@@@9A@9A A A A A A A @ G G G FG G 9@9@9@@9@9@9@@ @F@F@F@@@ @@@9A@@@@9A@9A FG@@@@@FG@FG F@ A@@@9A@@@9A@9A@9A FG@@@@@FG@FG A A A A @ G G G FG G 9@9@9@@9@9@@@ @F@F@F@@@ F@ A A A @ G G G FG G 9@9@9@@9@9@@@ @F@F@F@@@ @A@@9A@A@@9A@9A@9A FG@@@@@FG@FG F@ 9A 9A 9 9A 9 9 @ FG FG FG BCFG @@A@9A@@A@A@9A@9A @C@C@C@@@ FG @@B@B@B@@@ BC@BCFG@@@@@BCFG@BCFG BC@BC@@@@@BC@BC BC BC BC BC @@B@B@B@@@ C C C BC BC@BC@@@@@BC@BC @@B@B@B@@@ C C C DEBC BC@@E@E@E@@@ @@D@D@D@@@ DE@DEBC@@@@@DEBC@DEBC @@D@D@D@@@ E E E DE DE@DE@@@@@DE@DE @@D@D@D@@@ E E E DE DE@DE@@@@@DE@DE DE DE DE DE DE @DE@@@@@DE@DE

vi,j+1/2 pi,j vi,j−1/2

1 2

QPRQPR QPQPRR QPRQPR RR PP

I H¡ I¡I H¡ ¡HIH

T S¡ T¡ST S¡ ¡ST

i, j −a 1 i − 1, j − a 1 i + 1, j −a 1 i, j − 1 j i − 1, j − 1 i + 1, j − 1 i i, j − 1 j i+ i + 1, j − 1 i ui,j j i+1 vi,j i ui,j pi,j i+1 v pi,j i,j ui,j pi,j vi,j+1/2 v pi,j i,j vi,j−1/2 pi,j vi,j+1/2 ui−1/2,j pi,j vi,j−1/2 i− 1 vi,j+1/2 2 ui−1/2,j vi,j−1/2 1 i i− 1 i+ 2 ui−1/2,j 2 i i+2 1 i− 1 i+ 1 3 2 i+ 2 i +1 1 ji + 1 i+ 3 2 2 i+ 2 j +1 1 ji + 3 1 j−2 i+ 2 2 j j −1 1 j+1 j−2 2 j j −1 j−1 2 ence condition. j −1 ence condition. m equation (u) .

j i i+1 ui,j vi,j pi,j

j+

1 2

j j−

1 2

j −1

i

i+

1 2

i+1

i+

3 2

∆x

+

∆y

=0

j−1 2j j−1 2

Inﬂow A

v1, 3 2 1 B 0 0 1 2 i p1,1

3 u 2 ,1

**Figure 3.5: Staggered grid near the boundary.
**

2 where the u2 i+1,j , ui,j , (uv)i+1/2,j+1/2 , (uv)i+1/2,j−1/2 terms need to be interpolated from the points where the corresponding velocities are deﬁned. The control volume for the normal velocity is centered around the normal velocity point and the discretization becomes 2 2 (uv)i+1/2,j+1/2 − (uv)i−1/2,j+1/2 vi,j+1 − vi,j ∂vi,j+1/2 pi,j+1 − pi,j + + + − ∂t ∆x ∆y ∆y

**1 vi+1,j+1/2 − 2vi,j+1/2 + vi−1,j+1/2 1 vi,j+3/2 − 2vi,j+1/2 + vi,j−1/2 − =0 2 Re ∆x Re ∆y 2
**

2 where the (uv)i+1/2,j+1/2 , (uv)i−1/2,j+1/2 , u2 i,j+1 , ui,j terms need to be interpolated from the points where the corresponding velocities are deﬁned. For this discretization no checkerboard modes possible, since there is no decoupling in the divergence constraint, the pressure or the convective terms.

Boundary conditions

To close the system we need to augment the discretized equation with discrete versions of the boundary conditions. See ﬁgure 3.5 for a deﬁnition of the points near a boundary. The boundary conditions on the solid wall, BC in ﬁgure 3.5, consists of the no ﬂow and the no slip conditions, i.e. u = v = 0. The normal points are located on the boundary so that the no ﬂow condition simply become

1 1 v1, 2 = v2, 2 = · · · = 0

3 3 The evaluation of the u-equation at the point ( 2 , 1), for example, requires u 2 ,0 . We can ﬁnd this value as follows

3 0 = u2,1 = 2

The boundary conditions on the inﬂow, AB in ﬁgure 3.5, consist given values of u and v, i.e. so called Dirichlet conditions. The streamwise points are located on the boundary so that the boundary condition is simply consist of the known values

1 1 u 2 ,1 , u 2 ,2 , . . .

The evaluation of the v-equation at the point (1, 3 ), for example, requires v0, 3 . We can ﬁnd this value 2 2 as follows

1 v2,3 = 2

1 v 3 + v0, 3 2 2 1, 2

V V V V V V V V V V V V V V V V V U&U&&U&U&U&U&U&U&U&&U&U&U&U&U&U&&U&U& &&UV&&&&&&&&UV&&&&&&&UV&&&UV

C 1 3 3 u 2 ,1 + u 2 ,0 2 ⇒ ⇒

Solid wall

3 3 u 2 ,0 = −u 2 ,1

3 1 v0, 2 = 2v 2 , 3 − v1, 3 2 2

∂v If AB in ﬁgure 3.5 is an outﬂow boundary, the boundary conditions may consist of ∂u = ∂x = 0, i.e. ∂x so called Neumann conditions. These conditions should be evaluated at i = 1 . For the streamwise and 2 normal velocity we ﬁnd

3 u− 1 ,2 = u 2 ,2 , 2 3 v0, 3 = v1, 2 2

Note that no reference is made to the pressure points outside the domain, so that in this formulation no boundary conditions for p is needed so far. However, depending on the particular discretization of the time derivative and other the details of the solution method one may need to augment the the above with boundary conditions for the pressure. If this is the case particular care has to be taken so that those conditions do not upset the divergence free condition.

3.3

Summary of the equations

The discretized equations derived for the staggered grid can be written in the following form ∂ui+ 1 ,j pi+1,j − pi,j 2 1 + Ai+ 2 ,j + =0 ∂t ∆x ∂vi,j+ 1 pi,j+1 − pi,j 2 1 + Bi,j+ 2 + =0 ∂t ∆y D =0

i,j

**where the expression for the Ai+ 1 ,j can be written 2
**

1 Ai+ 2 ,j =

2 (uv)i+ 1 ,j+ 1 − (uv)i+ 1 ,j− 1 u2 i+1,j − ui,j 2 2 2 2 + ∆x ∆y 1 1 ui+ 1 ,j+1 − 2ui+ 2 ,j + ui+ 1 ,j−1 1 ui+ 3 ,j − 2ui+ 1 ,j + ui− 1 ,j 2 2 2 2 2 − − Re ∆x2 Re ∆y 2

=

= {expand and interpolate using grid values} = 1 1 2 1 1 1 1 − v (u 3 − ui− 2 ,j ) + (v + ui+ 1 ,j i,j+ 1 − vi+1,j− 1 − vi,j− 1 ) + 2 2 2 2 2∆x i+ 2 ,j 4∆y i+1,j+ 2 Re ∆x2 ∆y 2 1 1 1 1 1 + u 3 − ui+ 3 ,j + u 1 − ui− 2 ,j 2 4∆x i+ 2 ,j Re∆x2 4∆x i− 2 ,j Re∆x2 1 1 1 1 1 + v 1 + v + (v ui+ 1 ,j+1 + (v ui+ 1 ,j−1 i,j+ 1 ) − i,j− 1 ) − 2 2 2 2 4∆y i+1,j+ 2 Re∆y 2 4∆y i+1,j− 2 Re∆y 2

1 = a(u, v)i+ 2 ,j ui+ 1 ,j + 2

a(u, v)nb unb

nb

The last line summarizes the expressions, where the sum over nb indicates a sum over the nearby nodes. The expression for Bi,j+ 1 can in a similar way be written 2

1 Bi,j+ 2 =

∆x 1 vi+1,j+ 1 − 2vi,j+ 1 2 2 − R ∆x2

=

(uv)i+ 1 ,j+ 1 − (uv)i− 1 ,j+ 1

2 2 2 2

2 2 vi,j+1 − vi,j ∆y + vi−1,j+ 1 1 vi,j+ 3 − 2vi,j+ 1 + vi,j− 1 2 2 2 2 − R ∆y 2

+

**= {expand and interpolate using grid values} = b(u, v)i,j+ 1 vi,j+ 1 + 2 2
**

nb

b(u, v)nb vnb

where the expressions for b(u, v)i,j+ 1 and b(u, v)nb are analogous to a(u, v)i+ 1 ,j and a(u, v)nb , and 2 2 therefore not explicitly written out. The expression originating from the continuity equation is Di,j = ui+ 1 ,j − ui− 1 ,j 2 2 ∆x +

1 vi,j+ 2 − vi,j− 1 2

∆y

Here we we have used the deﬁnitions u v p A(u, v) B(u, v) Gx Gy Dx Dy f -

These equations, including the boundary conditions, can be assembled into matrix form. We have u A(u, v) 0 Gx u fu d v + 0 B(u, v) Gy v = fv dt 0 Dx Dy 0 p 0 vector of unknown streamwise velocities vector of unknown normal velocities vector of pressure unknowns non-linear algebraic operator from advective and viscous terms of u-eq. non-linear algebraic operator from advective and viscous terms of v-eq. linear algebraic operator from stremwise pressure gradient linear algebraic operator from normal pressure gradient linear algebraic operator from x-part of divergence constraint linear algebraic operator from y-part of divergence constraint vector of source terms from BC

With obvious changes in notation, we can write the system of equations in an even more compact form as d dt u 0 + N (u) G D 0 u p = f 0

If this is compared to the ﬁnite element discretization in the next chapter, it is seen that the form of the discretized matrix equations are the same.

3.4

Time dependent ﬂows

**The projection method
**

For time-dependent ﬂow a common way to solve the discrete equations is the projection or pressure correction method. Sometimes this method and its generalizations are also called splitting methods. We illustrate the method using the system written in the compact notation deﬁned above. We start to discretize the time derivative with a ﬁrst order explicit method n+1 − un u + N (un ) un + G pn+1 = f (tn ) ∆t Dun+1 = 0

Then we make a prediction of the velocity ﬁeld at the next time step u∗ not satisfying continuity, and then a correction involving p such that Dun+1 = 0. We have ∗ n u − u + N (un ) un = f (tn ) ∆t n+1 u − u∗ + G pn+1 = 0 ∆t Dun+1 = 0 Next, apply the divergence operator D to the correction step DG pn+1 = 1 D u∗ ∆t Here DG represents divgrad = Laplacian, i.e. discrete version of pressure Poisson equation. The velocity at the next time level thus becomes

un+1 = u∗ − ∆tG pn+1 Note here that un+1 is projection of u∗ on divergence free space. This is a discrete version of the projection of a function on a divergence free space discussed in the section about the role of the pressure in incompressible ﬂow. In order to be able to discuss the boundary condition on the pressure Poisson equation we write the prediction and correction step more explicitly. Using the staggered grid discretization the prediction step can be written ∗ n ui+ 1 ,j − ui+ 1 ,j 2 2 + An 1 ,j = 0 i+ 2 ∆t n ∗ 1 vi,j+ 1 − vi,j+ 2 n 2 + Bi,j+ 1 = 0 2 ∆t n+1 ui+ 1 ,j − u∗ 1 ,j pn+1 − pn+1 i+ 2 i+1,j i,j 2 + =0 ∆t ∆x n+1 ∗ n+1 n+1 vi,j+ 1 − vi,j+ 1 pi,j+1 − pi,j 2 2 + =0 ∆t ∆y

2

and the projection step as

n+1 By solving for un+1,j and vi,j+ 1 above and using this, and the corresponding expressions for un+1,j and i+ 1 i− 1 n+1 vi,j− 1 , in the expression for the divergence constraint Di,j we ﬁnd the following explicit expression for the 2 discrete pressure Poisson equation ∗ pn+1 − 2pn+1 + pn+1 pn+1 − 2pn+1 + pn+1 Di,j i,j+1 i,j i,j−1 i+1,j i,j i−1,j + = ∆x2 ∆y 2 ∆t ∗ where Di,j is Di,j evaluated using the discrete velocities at the prediction step, u∗ and v ∗ . When the discrete pressure Poisson equation is solved we need values of the unknowns which are outside of the domain, i.e. we need additional boundary conditions. If we take the vector equation for the correction or projection step and project it normal to the solid boundary in ﬁgure 3.5 we have

2 2

pn+1 − pn+1 1 2,1 2,0 ∗ =− v n+1 − vΓ 1 ∆y ∆t 2, 2

∗ ∗ where vΓ = v2, 1 is an unknown value of the velocity in the prediction step at the boundary. In general 2 it is important to choose this boundary condition such that discrete velocity after the correction step is divergence at the boundary, see the discussion in the end of the section on the role of the pressure. However, ∗ for this particular discretization it turns out that we can choose vΓ arbitrarily since it completely decouples from the rest of the discrete problem. To see this we write the pressure Poisson equation, centered abound the point (2,1), close to the boundary. We ﬁnd

pn+1 − 2pn+1 + pn+1 pn+1 − 2pn+1 + pn+1 1 3,1 2,1 1,1 2,2 2,1 2,0 + = 2 2 ∆x ∆y ∆t

u∗ ,1 − u∗ ,1 5 3

2 2

∆x

+

∗ ∗ v2, 3 − vΓ

2

∆y

if we substitute the value of pn+1 − pn+1 from the boundary condition obtained for the pressure into the 2,1 2,0 ∗ above equation we can se that vΓ cancels in the right hand side of the pressure Poisson equation. Since the discretization of the time derivative is explicit there is no other place where the values of v ∗ or u∗ on the ∗ boundaries enter. Thus we ﬁnd that we can choose the value of vΓ arbitrarily. In particular, we can choose n+1 ∗ vΓ = v2, 1 , in which case we obtain a Neumann boundary condition for the pressure. Note, however, that 2 this is a numerical artifact and does not imply that the real pressure gradient is zero at the boundary. A number of other splitting methods have been devised for schemes with higher order accuracy for the time discretization. In general it is preferred to make the time discretization ﬁrst for the full Navier-Stokes equations and afterward split the discrete equation into one predictor and one corrector step. See Peyret and Taylor for details.

**Time step restriction
**

For stability it is necessary that prediction step is stable, numerical evidence indicates that this is suﬃcient. In the predictor case the momentum equations decouple if we linearize them abound the solution U 0 , V0 . It thus suﬃces to consider the equation ∂u ∂u ∂u 1 + U0 + V0 = ∂t ∂x ∂y Re For simplicity we will here consider one-dimensional version ∂u 1 ∂2u ∂u + U0 = ∂t ∂x Re ∂x2 which in discretized form can be written un+1 = un + ∆t −U0 j j un − u n 1 un − 2un + un j+1 j−1 j+1 j j−1 + 2∆x Re ∆x2

2

u

where we have chosen to call the discrete points where the u-velocity is evaluate for x j . We introduce un = Qn · ei·ωxj j into the discrete equation, together with the deﬁnitions U0 ∆t 2∆t , α= , ξ = ω · ∆x ∆x Re ∆x2 This implies that the ampliﬁcation factor Q can be written λ= Q = 1 − iλ sin (ξ) − 2α sin2 We ﬁnd that the absolute value of Q satisﬁes |Q| = 1 − 4 λ2 − α2 s2 + 4 λ2 − s2 s = 1 − 4s where s = sin2

ξ 2

ξ 2

λ2 − α 2 s − λ 2 + α ≤ 1

. This implies that Φ (s) = − λ2 − α2 s + λ2 − α ≤ 0 ∀s : 0 ≤ s ≤ 1

Φ (s) is linear function of s and thus the the limiting condition on Φ (s) is given by its values at the two end points of the s-interval. We ﬁnd Φ (0) = λ2 − α ≤ 0, which implies that ⇒ λ2 ≤ α ≤ 1 Φ (1) = α2 − α ≤ 0

For the two dimensional version of the predictor step Peyret and Taylor [8] ﬁnd the following stability limits 1 U 2 + V 2 ∆t · Re ≤ 1 0 4 0 4∆t ≤ 1 Re∆x2 Note that this is a stronger condition than in the one-dimensional case.

substituting back the deﬁnitions of λ and α we ﬁnd the following stability conditions 1 U 2 ∆tRe ≤ 1 2 0 2∆t ≤ 1 Re∆x2

3.5

General iteration methods for steady ﬂows

Distributive iteration

For steady ﬂows we can still use the methods introduced for unsteady ﬂows, and iterate the solution in time to obtain a steady state. However, in many cases it is more eﬀective to use other iteration methods. We will here introduce a general distributive iteration method and apply it to the discretized steady Navier-Stokes equations. For the steady case the discretized equations can be written in the form N (u) G D 0 u p = f 0

which is of the form A y = b. Now, let y = B y which implies that ˆ AB y = b ˆ and use the iteration scheme on this modiﬁed equation. We divide the matrix AB in the following manner AB = M − T and use this to deﬁne the iteration scheme M B −1 y k+1 = T B −1 y k + b = M B −1 y k + b − A y k which can be simpliﬁed to y k+1 = y k + BM −1 b − A y k

rk

⇒

M y = Ty + b ˆ ˆ

where rk is the residual error in the k:th step of the iteration procedure. If B = I we obtain a regular iteration method where residual r k is used to update y k by inverting M , a simpliﬁed part of A, as in the Gauss-Seidell method for example. For this method we have Ax = b A=L−U (L − U ) x = b L xn+1 = U xn + b

**Application to the steady Navier-Stokes equations
**

When the distributive iteration method is applied to the steady Naiver-Stokes we have A= N (u) G D 0

we can obtain AB in the following block triangular form AB = if the matrix B has the form B= I 0 −N (u)−1 G I ≈ I 0 ˜ −N (u)−1 G I N (u) D 0 −DN (u)−1 G

˜ where N(u)−1 is a simple approximation of the inverse of the discrete Navier-Stokes operator. We now split AB = M − T where M= Q D 0 R

and where Q is an approximation of N (u) and R is an approximation of −DN −1 G. Note that R is a discrete Laplace like operator.

The SIMPLE (Semi-Implicit Method for Pressure Linked Equations) by Pantankar and Spalding (1972) can be thought of as a distributive iteration method. Wesseling [10] indicates that by choosing ˜ N (u)−1 = [diagN (uk )]−1

Q = N (uk ) R = −D[diagN (uk )]−1 G we have a method which is essentially the original version of the SIMPLE method. Thus we have the following iteration steps 1. First, we calculate the residual b − A yk = f 0 − N (uk ) G D 0 uk pk =

k r1 k r2

**2. Second, we calculate M −1 rk which implies
**

k Q δ˜ = r1 u

⇒

k δ˜ = Q−1 r1 u

k R δ p = r2 − D δ˜ ˜ u

3. Third, we have the distribution step BM −1 rk δu δp =B δ˜ u δp ˜

which becomes = ˜ δ˜ − N −1 G δ p u ˜ δp ˜

4. Fourth, there is an under relaxation step, where the velocities and the pressure is updated uk+1 = uk + ωu δu pk+1 = pk + ωp δp The under relaxation is needed for the method to converge, thus ωu and ωp is usually substantially lower that one.

Chapter 4

Finite element methods for incompressible ﬂow

u=0 on Γ0 , (4.1b) i i+ 1 2 ∂u =0 on Γ1 . (4.1c) i+1 ∂n 3 i+ 2 ∂Uj +1 2 The coeﬃcient ν > 0 is the thermal jdiﬀusivity. Incompressibility yields that ∂xj = 0. We assume that j Γ1 is a solid wall, so that nj Uj = 0 on Γ1 . This simpliﬁes the analysis but is not essential. The Dirichlet j−1 2 isothermal condition, whereas the Neumann condition ∂u = 0 boundary condition u = 0 on Γ0 is called an ∂n j−1 on Γ1 is an adiabatic condition. A Deriving the integral form that is used to form the ﬁnite-element discretization requires some vector B calculus. The product rule of diﬀerentiation yields C u3 ∂ 2 ,1 ∂u ∂v ∂u v1, 3 v = + v 2 u. (4.2) 2 ∂xi ∂xi ∂xi ∂xi p1,1 0 n 1 Γ0 2 i j Ω Inﬂow Solid wall Γ1

dx dy − dx dl = ( dx, dy) n dl = ( dy, − dx) i − 1, j + 1 i, j + 1 i + 1, j + 1 c c b b i − 1, j i + 1, j i, j Aabcd d d a The ﬁnite-element method (FEM) approximates an integral form of the governing equations. However, a the integral form—called the variational form—is usually diﬀerent from the one used for the ﬁnite-volume i − 1, j − 1 method. Moreover, diﬀerent integral forms may be used for the same equation depending on circumstances such as boundary conditions. We i, j − 1 introduce FEM ﬁrst for a scalar advection–diﬀusion problem before i + 1, j − discussing the Navier–Stokes equations. 1 j i i+1 4.1 FEM for an advection–diﬀusion problem ui,j vi,j We consider the steady ﬂow of an incompressible ﬂuid in a bounded domain Ω, and we assume that its pi,j velocity ﬁeld Uj is known (from a previous numerical solution, for instance). We assume that the boundary pi,j Γ of the domain can be decomposed into two portions Γ0 and Γ1 , as in ﬁgure 4.1 for instance. We are vi,j+1/2 interested to compute the temperature ﬁeld u in the domain, modeled by the advection–diﬀusion problem vi,j−1/2 ∂u ui−1/2,j in Ω, (4.1a) Uj 1 − ν 2 u = f i −∂xj 2

Figure 4.1: A typical domain associated with the advection–diﬀusion problem (4.1). 71

**Recall the divergence theorem
**

Ω

∂wi dΩ = ∂xi

ni wi dΓ.

Γ

(4.3)

**Integrating expression (4.2) and applying the divergence theorem on the left side yields v
**

Γ

∂u dΓ = ∂n

Ω

∂v ∂u dΩ + ∂xi ∂xi

v

Ω

2

u dΩ.

(4.4)

(Note that ∂u ∂u = ni .) ∂n ∂xi Let v be any smooth function such that v = 0 on Γ0 . Multiplying equation (4.1a) with v, integrating over Ω, and utilizing expression (4.4) yields ∂u vUj dΩ − ν ∂xj ∂u vUj dΩ + ν ∂xj ∂v ∂u dΩ − ν ∂xj ∂xj 0 ∂u dΓ, v ∂n

f v dΩ =

Ω Ω

v

Ω

2

u dΩ =

Ω

Ω

Γ

where the last term vanishes on Γ0 due to the requirements on v, and on Γ1 due to boundary condition (4.1c) We have thus shown that a smooth (twice continuously diﬀerentiable) solution to problem (4.1) (also called a classical solution) satisﬁes the varational form vUj

Ω

∂u dΩ + ν ∂xj

Ω

∂v ∂u dΩ = ∂xj ∂xj

f v dΩ

Ω

(4.5)

for each smooth function v vanishing on Γ0 We will now “forget” about the diﬀerential equation (4.1), and directly consider the variational form (4.5). For this, we need to introduce the function space ∂v ∂v V = v: dΩ < +∞, v = 0 on Γ0 . ∂xi ∂xi

Ω

The space V is a linear space, that is, u, v ∈ V ⇒ αu + βv ∈ V ∀α, β ∈ R. The requirement that the gradient-square of each function in V is bounded should be intepreted as a requirement that the energy is bounded. The variational problem will thus be Find u ∈ V such that ∂u ∂v ∂u vUj dΩ + ν dΩ = ∂xj ∂xj ∂xj

Ω Ω

f v dΩ

Ω

∀v ∈ V.

(4.6)

A solution to variational problem (4.6) is called a weak solution to advection–diﬀusion problem (4.1). The derivation leading up to expression (4.5) shows that a classical solution is a weak solution. However, a weak solution may not be a classical solution since functions in V may fail to be twice continuously diﬀerentiable. Note that the boundary conditions are incorporated into the variational problem. The Dirichlet boundary condition on Γ0 is explicitly included in the deﬁnition of V and is therefore referred to as an essential boundary condition. The Neumann condition on Γ1 is deﬁned implicitly through the variational form and is therefore called a natural boundary condition.

4.1.1

Finite element approximation

Let us triangulate the domain Ω by dividing it up into triangles, as in ﬁgure 4.2. Let M be the number of triangles and N the number of nodes in Ω ∪ Γ1 , that is, the nodes on Γ0 are excluded in the count. Let h be the largest side of any triangle. Let Vh be the space of all functions that are - continuous on Ω,

p1,1 0 1 2 i j Inﬂow Solid wall n Γ0 Γ1 Ω

Γ0

Γ1

3 u 2 ,1 3 v1, 2 p1,1 0 1 2 i j Inﬂow Solid wall n Γ0 Γ1 Ω

Figure 4.2: A triangulation. Here, the number N are Figure 4.3: The “tent or “hat basis function associated with continuous, piece-wise linear functions. the black nodes. - linear on each triangle, - vanishing on Γ0 . Note that Vh ⊂ V . We get a ﬁnite-element approximation of our advection-diﬀusion problem by simply replacing V by Vh in variational problem (4.6), that is: Find uh ∈ Vh such that ∂vh ∂uh ∂uh v h Uj dΩ + ν dΩ = ∂xj ∂xj ∂xj

Ω Ω

f vh dΩ

Ω

∀vh ∈ Vh .

(4.7)

Restricting the function space in a variational form to a subspace is called a Galerkin approximation. The ﬁnite-element method is a Galerkin approximation in which the subspace is given by piecewise polynomials.

4.1.2

The algebraic problem. Assembly.

Once the value of a function uh in Vh is known at all node points, it is easy to reconstruct it by linear interpolation. This interpolation can be written as a sum of the “tent” basis functions Φ (n) (ﬁgure 4.3), which are functions in Vh that are zero at each node point except node n, where it is unity. Thus,

N

uh (xi ) =

n=1

u(n) Φ(n) (xi ) ,

(4.8)

where u(n) denotes the value of uh at node n. Note that expansion (4.8) forces uh = 0 on Γ0 since it excludes the nodes on Γ0 in the sum. Substituting expansion (4.8) into FEM approximation (4.7) yields

N

u(n)

n=1 Ω

v h Uj

∂Φ(n) dΩ + ν u(n) ∂xj n=1

N

Ω

∂vh ∂Φ(n) dΩ = ∂xj ∂xj

f vh dΩ

Ω

∀ v h ∈ Vh .

(4.9)

**Since each basis function is in Vh , we may choose vh = Φ(m) ∈ Vh for m = 1, . . . , N in equation (4.9), which then becomes
**

N

u(n)

n=1 Ω

Φ(m) Uj

∂Φ(n) dΩ + ν u(n) ∂xj n=1

N

Ω

∂Φ(m) ∂Φ(n) dΩ = ∂xj ∂xj

f Φ(m) dΩ, m = 1, . . . , N .

Ω

(4.10)

**This is a linear system system Au = b, where A = C + νK and Kmn =
**

Ω

∂Φ(m) ∂Φ(n) dΩ, ∂xj ∂xj

Cmn =

Ω

Φ(m) Uj

∂Φ(n) dΩ, ∂xj

(4.11)

h Figure 4.4: An example mesh that gives a particular simple stiﬀness matrix K

u(1) . u= . . u(n)

The union of all triangles constitutes the domain Ω and that the triangles do not overlap. Thus, any integral over Ω can be written as a sum of integrals over each triangle, ∂Φ(m) ∂Φ(n) dΩ = ∂xj ∂xj p=1

M

Kmn =

Ω

Note that Kmn vanishes as soon as m or n is not a corner node of triangle p. Thus Kmn contributes to the sum only when m and n both are corner nodes of triangle p. Matrix K can thus be computed by looping (p) all M triangles, calculating the element contribution Kmn for m, n being the three nodes of triangle p and adding these 9 contribution to matrix K. Note that the derivatives of Φ(n) are constant on each triangle, (p) so Kmn is easy to compute exactly. Matrix C and vector b can also be computed by element assembly, but numerical quadrature is usually needed, except if Uj and f happen to be particularly “easy” functions (such as constants).

(p)

4.1.3

An example

As an example, we will compute the elements of K, the “stiﬀness matrix” (a term borrowed from structural mechanics), for the particular case of the structured mesh depicted in ﬁgure 4.4. Let M be the number of internal nodes in the horizontal as well as the vertical direction (the solid nodes in ﬁgure 4.4.) The width of each “panel” is then h = 1/(M + 1), and the area of each triangle is |Tk | = h2 /2. The components of K are Kmn =

Ω

Note that Kmn vanishes for most m and n. For instance, Km,m+2 = 0, since basis functions Φ(m) and Φ(m+2) do not overlap (ﬁgure 4.4). In fact, Kmn = 0 only when m and n are associated with nodes that are nearest neighbors. To compute matrix element (4.12), we need expressions for the gradients of basis functions Φ (m) and (m+1) Φ . Since the functions are piecewise linear, the gradient is constant on each triangles and may thus easily be computed by ﬁnite-diﬀerences in the coordinate direction (ﬁgure 4.5). By the expressions in

X`XXX`X`XX XXWXXXWXWX a `a a a WY Y WY Y Y `aXXX`aXXX`aX`a XWYXXXWYXXXWY a`aX`a `XaX`a`a`a `a `a WYWYWY WY WYWYWY WY WY X`XX`aXaXaX`aX`a XWYXYXXWYXYXYXWY XaXXX`X`XX XXWXXXWXWX `aX`XX`aXaXaX`aX`a XWYXYXXWYXYXYXWY XaXXX`X`XX XXWXXXWXWX `aX`XX`aXXX`aX`a XWYXXXWYXXXWY X`XX`aX`X`X`aX`a XWYXWXXWYXWXWXWY aXXXaXaXX XXYXXXYXYX a`a a a WYWYWY Y WYWYWY Y Y `XX`X`XX XXWXXXWXWX a`aXaXXXaXaXX XXYXXXYXYX `X`XX`aX`X`X`aX`a XWYXWXXWYXWXWXWY aXaXXXaXaXX XXYXXXYXYX `X`XXX`X`XX XXWXXXWXWX XXX`a`aXXX`a`aX`a`a XWYWYXXXWYWYXXXWYWY `a`a WY WY

m m+2 1 M h f Φ(1) dΩ

Ω

i j Inﬂow Solid wall n Γ0 Γ1 Ω

M2

b=

. . . f Φ(N ) dΩ

Ω

Tp

**∂Φ(m) ∂Φ(n) (p) dΩ = Kmn . ∂xj ∂xj p=1
**

(p)

M

∂Φ(m) ∂Φ(n) dΩ ∂xj ∂xj

(4.12)

m 2 m+2 T3 i on T1 , 1 j on T2 , h T4 Inﬂow T2 on T3 , Solid wall Φ(m) on T4 , n T1 T5 on T5 , Γ0 Γ1 on T6 , T6 Ω on all other triangles. 2 M (m) M Figure 4.5: The gradient of test function Φ for m being the (black) middle node. m m+2 1 1 Φ 1 (m) −h, h on T1 , (m) 1 h Φ = − ,0 on T2 , T2 Φ(m+1) h 0 on all other triangles T1 1 ,0 on T1 , h Φ(m+1) = 1 1 on T2 . h, −h 1 h, 0 1 ,−1 h h 0, − 1 h 1 = −h, 0 1 1 − , h h 0, 1 h 0 Figure 4.6: The support for basis functions Φm and Φm+1 are marked with vertical and horizontal stripes respectively. ﬁgure 4.5, we have Kmm =

Ω

∂Φ(m) ∂Φ(m) dΩ = ∂xj ∂xj

6

k=1T k

∂Φ(m) ∂x

1 1 1 1 1 1 |T1 | + 2 2 |T2 | + 2 |T3 | + 4 |T4 | + 2 2 |T5 | + 2 |T6 | 2 h h h h h h 1 h2 =8 2 =4 h 2 = The common support for basis functions Φm and Φm+1 is only the two triangles marked in ﬁgure 4.6; that is, it is only on these two triangles that both functions are nonzero at the same time. Utilizing the gradient expressions given in ﬁgure 4.6 yields

2

Km,m+1 =

Ω

Φ(m) ·

Φ(m+1) dΩ =

k=1T

k

Φ(m) ·

Φ(m+1) dΩ = −

and in the same manner, Altogether, K becomes the block-tridiagonal M 2 -by-M 2 matrix T −I −I T −I .. .. .. K = . . . −I T −I −I T Km,m−1 = Km,m+M = Km,m−M = −1.

where I is the M -byM identity matrix, and T the M -byM matrix 4 −1 −1 4 −1 .. .. .. T = . . . −1 4 −1 −1 4 Thus, a typical row in the matrix–vector product Ku becomes

4um − um+1 − um−1 − um+M − um−M ,

b b f@f cb f b b f b b b b f b b b b b b b bc e@e@@f cccccccccccccccccbb @f@@e@f@f@@f@ ef@e@ef@f@e@e@ef@e@ef d b e@e@@e@ @f@ef@f@ccccccccccccccccccccccccccccdbdbdbdbdbdbdbdbdbdbdbdbdbdbdbe@f@f@ef@f@ef b cb fef@e@@e@e@e@@e@ e@f@ef@@f@f@ef@f@ef d f@e@@f cbdcc e@e@@e@ e@f@@e@f@f@@f@ @e@ef@f@ccccccccccccccccccccccccccccdbdbdbdbdbdbdbdbdbdbdbdbdbdbdc e@e@ef@e@ef b cb @f@@e@f@f@@f@ ef@e@ef@f@e@e@ef@e@ef @e@@f@e@e@@e@ f@@e@f@f@@f@ ef@f@ef@f@ccccccccccccccccccccccccccccdbdbdbdbdbdbdbdbdbdbdbdbdbdbdbe@f@f@ef@f@ef b cb f@e@@e@e@e@@e@ e@@efef@@ccccccccccccccccccccccccccccdbdbdbdbdbdbdbdbdbdbdbdbdbdbdc @@efef@@efef d c b cb fef@ef e@f@@ef cbd ef ef ef @e@ef@@f@f@ef@f@ef @f@@f cc e@e@@e@ ef@e@ef@f@ccccccccccccccccccccccccccccdbdbdbdbdbdbdbdbdbdbdbdbdbdbdbe@f@f@ef@f@ef b cb @f@@e@e@e@@e@ d ef@e@ef@f@f@f@ef@f@ef @f@@e@e@e@@e@ ef@e@ef@@ccccccccccccccccccccccccccccdbdbdbdbdbdbdbdbdbdbdbdbdbdbdc @@ef@@ef b cb @e@ef@e@f@e@ef@f@ef f@@f@e@f@@e@ df e@f c c e@ cbe@f@e@@f@ df d d f d fef@f@@f@dcb e@dbc b e@dbdf@b ef@dbd@b e@ e@e@ef@e@ccccccccccccccccccccccccccccdbdbdbdbdbdbdbc ef d cd d ef@ef@@ef@d ef@d ef@@d ef@ d d d d d @@efef@@c@c@ced fef@c@ced fef cccccd

2

+

∂Φ(m) ∂y

2

dΩ

1 1 2 h2 |T1 | − 2 |T2 | = − 2 = −1, 2 h h h 2

,

. (4.13)

that is, after dividing expression (4.13) with h2 , we recognize the classical ﬁve-point stencil for the negative Laplacian − 2 . Thus our ﬁnite-element discretization is for this particular mesh equivalent to a ﬁnite-volume or ﬁnitediﬀerence discretization. However, for a general unstructured mesh, the ﬁnite-element discretization will not give rise to any obvious ﬁnite-diﬀerence stencil.

4.1.4

Matrix properties and solvability

Matrix A is sparse: the elements of martix A are mostly zeros. At row m, matrix element A mn = 0 only in the columns where n represents a nearest neighbor to node m. The number of nonzero elements on each row does not grow (on average) when the mesh is reﬁned (since the number of nearest neighbors in a mesh does not grow). Thus, the matrices become sparser and sparser as the mesh is reﬁned. A common technique in ﬁnite-element codes is therefore to use sparse representation, that is, to store only the nonzero elements and pointers to matrix locations. It is immediate from expression (4.11) that Kmn = Knm , that is, K is a symmetric matrix (K T = K). Integration by parts also shows that matric C is skew symmetric (C T = −C). Theorem 1. K is positive deﬁnite, that is, v T Kv > 0 for each v = 0. Proof. Let vh ∈ Vh . Then vh =

N N N n=1

v (N ) Φ(n) . Denote v = (v (1) , . . . , v (N ) )T .

v T Kv =

m=1 n=1 N

v (n)

Ω

**∂Φ(n) ∂Φ(m) dΩ v (m) ∂xi ∂xi ∂ v (m) Φ(n) dΩ = ∂xi n=1
**

N

=

∂ v (m) Φ(m) Ω m=1 ∂xi

∂vh ∂xi

Ω

∂vh ∂vh dΩ ≥ 0, ∂xi ∂xi

with equality if and only if

= 0, that is, if vh = Const. But since vh |Γ0 = 0, vh ≡ 0.

From Theorem 1 also follows that A is positive deﬁnite. To see this, note that the skew-symmetry of C yields that T v T Cv = v T Cv = v T C T v = −v T Cv that is, v T Cv = 0 for each v. Thus v T Av = v T (K + C) v = v T Kv + v T Cv = v T Kv > 0 for each v = 0, that is, A is positive deﬁnite. However, A is not symmetric whenever U j = 0. Recall that a linear system Au = b has a unique solution for each b if the matrix is nonsingular. A positive-deﬁnite matrix is a particular example of a nonsingular matrix. We thus conclude that the linear system resulting from the ﬁnite-element discretization (4.7) has a unique solution.

4.1.5

Stability and accuracy

We have deﬁned the ﬁnite-element approximation and showed that the ﬁnite-element approximation yields a linear system Au = b that has a unique solution. Now the question is whether the approximation is any good. We will study the stability and accuracy of the numerical solution. FEM usually uses integral norms for analysis of stability and accuracy. We start by reviewing some mathematical concepts that will be used in the analysis. Basic deﬁnitions and inequalities The most basic integral norm is

1/2

v

L2 (Ω)

=

Ω

v 2 dΩ

**For the current problem, a more fundamental quantity is the energy norm
**

1/2

v and associated inner product

V

=

Ω

| v|2 dΩ

=

Ω

∂v ∂v dΩ ∂xi ∂xi

1/2

(u, v)V =

Ω

u·

v dΩ =

Ω

∂u ∂v dΩ. ∂xi ∂xi

(4.14)

**The Cauchy–Schwarz inequality for vectors a = (a1 , . . . , an ) is |a · b| ≤ |a||b| = |a · a|1/2 |b · b|1/2 , whereas for functions it reads
**

1/2 Ω 1/2

f g dΩ ≤

f 2 dΩ

Ω Ω

g 2 dΩ

,

(4.15)

**which can be denoted (f, g)L2 (Ω) ≤ f
**

1/2 L2 (Ω)

g

L2 (Ω) .

**The Cauchy–Schwarz inequality (4.15) implies that, for each nonzero, square-integrable g, f 2 dΩ
**

Ω

≥

Ω Ω

f g dΩ

1/2

g 2 dΩ

.

**Choosing g = f yields equality in above expression. Thus
**

1/2

f

L2 (Ω)

=

Ω

f 2 dΩ

= max

g=0

Ω Ω

f g dΩ

1/2

g 2 dΩ

= max

g=0

Ω

f g dΩ

L2 (Ω)

g

The rightmost expression is a “variational characterization” of the L2 (Ω) norm, that is, it expresses the norm in terms of something that is optimized. By inserting derivatives in the nominator and/or denominator of the variational characterization, various exotic norms can be deﬁned. A norm, diﬀerent from the L 2 (Ω) norm and of importance for the Navier–Stokes equations is p p

exotic

= max

wi

Ω

∂wi ∂xi

2

dΩ

1/2

.

(4.16)

Ω

| wi | dΩ

This norm will be of crucial importance when we analyze the stability of FEM form the Navier–Stokes equations. The Poincar´ inequality relates the energy and L2 (Ω) norms: e Theorem 2 (Poincar´ inequality). There is a C > 0 such that e v 2 dΩ ≤ C v for each v ∈ V . Remark 1. Theorem 2 says that the energy norm is stronger than then L2 (Ω) norm. The smallest possible value of C grows with the size of Ω. The Poincar´ inequality holds only if Ω is bounded in at least one e direction. Also, the inequality does not hold if Γ0 dΓ = 0. However, the inequality v 2 dΩ ≤ C v 2 dΩ +

Ω Ω 2 L2 (Ω)

Ω

Ω

**∂v ∂v dΩ, ∂xi ∂xi
**

2 V

that is, ≤C v

Ω

∂v ∂v dΩ ∂xi ∂xi

holds also when Stability

Γ0

dΓ = 0.

**Choosing vh = uh in equation (4.7) yields uh U j
**

Ω

∂uh dΩ + ν ∂xj

Ω

∂uh ∂uh dΩ = ∂xj ∂xj

f uh dΩ.

Ω

(4.17)

Note that uh U j

Ω

∂uh dΩ = uT Cu = 0, ∂xj

m m+2 1 h

u V Vh uh

Figure 4.7: An orthogonal projection on a line. Note that the vector u − uh is the shortest of all vectors u − vh for vh ∈ Vh , that is, u − uh ≤ u − vh ∀vh ∈ Vh . since C = −C T (using the notation of § 4.1.4). Expression (4.17) thus becomes ν

Ω

**∂uh ∂uh dΩ = ∂xi ∂xi
**

uh

2 V

f uh dΩ

Ω

1/2

1/2 Ω

(Cauchy–Schwarz) ≤ (Poincar´) ≤ C e

f 2 dΩ

Ω 1/2

**u2 dΩ h ∂uh ∂uh dΩ ∂xi ∂xi
**

1/2

f 2 dΩ

Ω L2 (Ω) Ω

**≤C f Dividing with uh
**

V

uh

V

.

shows that uh

V

C f L2(Ω) ν where the constant C does not depend on h. That is, we have shown that uh cannot blow up as the discretization is reﬁned, which is the same as saying that the numerical solution is stable. ≤ Error bounds Choosing v = vh ∈ Vh ⊂ V in variational problem (4.6) and subtracting equation (4.7) from(4.6) yields the “Galerkin orthogonality”, ν

Ω

∂vh ∂ (u − uh ) dΩ + ∂xi ∂xi

v h Uj

Ω

∂ (u − uh ) dΩ = 0 ∂xj

∀vh ∈ Vh .

**This is a “true” orthogonality condition when Uj ≡ 0: ν
**

Ω

∂vh ∂ (u − uh ) dΩ = 0 ∂xi ∂xi

∀vh ∈ Vh ;

(4.18)

that is, the error u − uh is orthogonal to each vh ∈ Vh with respect to the inner product (4.14). Cf. two orthogonal vectors: a · b = 0 . Orthogonality relation (4.18) means that u h is an orthogonal projection of u on Vh . As illustrated in ﬁgure 4.7, the orthogonal projection on a subspace yields the vector that is closest to the given element in the norm derived from the inner product in which orthogonality is deﬁned. Orthogonality property (4.18) (cf. ﬁgure 4.7) implies that the FE approximation is optimal in the energy norm when Ui ≡ 0: u − uh V ≤ u − v h V ∀vh ∈ Vh . (4.19) The matrix A is symmetric in this case (since C = 0). When Uj = 0, the FEM approximation is no longer optimal in this sense. However, there is a C > 0 such that Cea’s lemma holds, u − uh

V

≤

C u − vh ν

V

∀vh ∈ Vh .

(4.20)

Matrix A is nonsymmetric in this case. Note that the approximation may be far from optimal when ν is small, as the next example illustrates.

M 0 M 0.8 1 m 0.7 m + 2 0.6 i 0.5 j h 0.4 Inﬂow V Solid wall 0.3 Vh n 0.2 u Γ0 uh 0.1 Γ1 0 0.9 0.92 0.94 0.96 0.98 1 Ω x 2 M M Figure 4.8: Under-resolved standard Galerkin ﬁnite-element approximations may yield oscillations for m advection-dominated ﬂows. m+2 v1, 3 1 2 u V Vh u uh ˆ v02 P t = 1, 3 h

Figure 4.9: An interpolant Πh v interpolates v at the nodal points. Example 1. Let us consider the above FEM applied to the 1D problem − 1 u +u =1 500 u(0) = u(1) = 0 in (0, 1). (4.21)

The problem is advection dominated, that is, the viscosity coeﬃcient in equation (4.21) is quite small: ν = 1/500. A boundary layer with sharp gradients forms close to x = 1. The solid line in ﬁgure 4.8 shows the FE solution h = 1/1000, whereas the dashed line shows the FE approximation for h = 1/200. The result of example 1 is typical: under-resolved standard Galerkin methods may generate oscillations for advection-dominated ﬂows in areas of sharp gradients. An explanation of this phenomenon is that the directions of ﬂow is not accounted for in the standard Galerkin approximation. Galerkin approximations correspond to ﬁnite-diﬀerence central schemes. Upwinding is used to prevent oscillations for ﬁnite-diﬀerence methods. Various methods exist to obtain similar eﬀects in FEM, for instance streamline diﬀusion and discontinuous Galerkin. Both these methods modify the basic Galerkin idea to account for ﬂow-direction eﬀects. Approximation and mesh quality Estimates (4.19) and (4.20) are independent of the choice of Vh ! We only used that Vh ⊂ V . Next step in assessing the error is an approximation problem: How well can functions in V h approximate those in V ? This step is independent of the equation in question. Interpolants are used to study approximation properties. The interpolant is the function Πh v ∈ Vh that interpolates the function v ∈ V at the node points of the triangulation, as illustrated for a 1D case in ﬁgure 4.9. Approximation theory yields that, for suﬃciently smooth v, v − Πh v

L2 (Ω) V

v − Πh v

≤ Ch

≤ Ch2

“second order error” “ﬁrst order error”

(4.22)

The constant C contains integrals of the second derivatives of v. A mesh quality assumption is needed for (4.22) to hold. The essential point is: beware of “thin” triangles; do not let the shape deteriorate as the mesh is reﬁned. A strategy to preserve mesh quality when reﬁning is to subdivide each triangle into four new triangles by joining the edge midpoints (ﬁgure 4.10). Each of the conditions below is suﬃcient for approximation property (4.22) to hold:

1 2 i j Inﬂow Solid wall n Γ0 Γ1 Ω Figure 4.10: A subdivision that preserves mesh quality. M2 M m (i) (“Maximum angle condition”). The largest anm+2 gle of any triangle should not approach 180◦ as 1 h → 0. h (ii) (“Chunkiness parameter condition”). The V quotient between the diameter of the largest Vh circle that can be inscribed and the diameter u of the triangle should not vanish as h → 0. uh Condition (ii) implies (i), but not the reverse. Accuracy of the FE solution Error bounds (4.19) and (4.20) together with approximation property (4.22) implies u − uh Further analysis (not provided here) yields u − uh

L2 (Ω) V

M m m+2 1 h V Vh u uh

≤ Ch

≤ Ch2 .

This second-order convergence rate requires a mesh quality assumption and a suﬃciently smooth solution u (since the constant C contains second derivatives of u). The convergence rate is of optimal order : it agrees with approximation property (4.22) and is the best order that in general can be obtained for, in this case, piecewise-linear functions. However, the solution may still be bad for a particular, too-crude mesh; recall example 1. Improving Accuracy Accuracy is improved by reﬁning the mesh (“h-method”). This may be done adaptively, where it is needed (say, in areas of large gradients), to prevent the matrices to become too large. Automatic methods for adaptation are common. An alternative is to keep the mesh ﬁxed and increase the order of the polynomials at each triangle (“p-method”). For instance, continuous, piecewise quadratics yields a third-order-accurate solution (in the L2 (Ω) norm). These strategies may be combined in the “h–p method”, where the mesh is reﬁned in certain regions and the order of approximations is increased in other.

4.1.6

Alternative Elements, 3D

A quadrilateral is a “skewed” rectangle: four points connected by straight lines to form a closed geometric object. Nonoverlapping quadrilaterals can be used to “triangulate” a domain, as in ﬁgure 4.11. An example of a ﬁnite-element space Vh on quadrilaterals is globally continuous functions varying linearly along the edges. As in the triangular case, the functions are deﬁned uniquely by specifying the function values at each node. The functions are deﬁned by interpolation into the interior of each quadrilateral. This space reduces to piecewise bilinear functions for rectangles with edges in the coordinate directions: vh (x, y) = a + b x + c y + d xy Quadrilaterals on a “logically rectangular domain” (a distorted rectangle, as in ﬁgure 4.11) yields a regular structure to the matrix A. This may

2

1

2

p1,1 p1,1 Ω 2 M 0 0 M 1 1 m 2 2 m+2 i i j j 1 h Inﬂow Inﬂow V Solid wall Solid wall Vh n n Γ0 Γ0 u uh Γ1 Γ1 Ω Ω M 2 Figure 4.11: A logically rectangular domain triangulated with quadrilaterals. M2 M M m m m+2 m+2 1 1 h h V V Vh Vh u u uh uh Figure 4.12: In 3D, triangular and quadrilateral elements generalize to tetrahedral (left) and hexahedral (right) meshes. • give high accuracy (particularly if the mesh is aligned with the ﬂow), and • allow eﬃcient solution of the linear system (particularly for uniform, rectangular meshes). Compared to triangular mesh, it is harder to generate quadrilateral meshes automatically for complicated geometries. Triangular and quadrilateral meshes generalize in 3D to tetrahedral and hexahedral meshes (ﬁgure 4.12). Advantages and limitations with tetrahedral and hexahedral meshes are similar to corresponding meshes in 2D.

Figure 4.14: The black nodes are the degrees of freedom for the velocity components, which are fewer than the number of triangles.

4.2

FEM for Navier–Stokes

We now turn to the Navier–Stokes equations for the steady ﬂow of an incompressible ﬂuid, and consider a situation with ﬂow in a bounded domain Ω. A simple example domain is depicted in ﬁgure 4.13. The mathematical problem in diﬀerential form is uj ∂ui ∂p 1 + − ∂xj ∂xi Re

2

The boundary condition on Γ1 = ∂Ω \ Γ0 is of no obvious physical signiﬁcance, but is useful as an artiﬁcial outﬂow condition. The Navier–Stokes equations is a system of nonlinear advection–diﬀusion equations together with the incompressibility condition (4.23b). This condition introduces new complications not present in the advection–diﬀusion problem of section 4.1. Naive approximations of the Navier–Stokes equations are bound to produce disappoing results, as the following example indicates. Example 2 (“the counting argument”). Consider a case with only Dirichlet boundary conditions—that is, Γ1 = ∅—and the mesh of ﬁgure 4.14. Assume that we use continuous, piecewise-linear approximations u h , vh for the x- and y-components of the velocity ﬁeld. There are I number of panels in each direction and therefore 2(I − 1)2 degrees of freedom (two times the number of black dots) for the velocity vector ﬁeld. The left side of equation (4.23b), that is ∂uh ∂vh + ∂x ∂y is here constant on each triangle. Thus, demanding equation (4.23b) to be satisﬁed on each triangle yields 2I 2 equations, which is more than the degrees of freedom for the velocity! Thus u h = vh = 0 is the only FE function satisfying the incompressibility condition (4.23b)!

h gh h h gh h gh h h gh h gh h h gh h gh h h h h gh h h h h h h h h gXXXgXgXXXgXXXgXgXXXgXXXgXgXXXgXXXgXgXXgXgXXXgXgXgXgXgXgXXgXgX XXghXXXXghXXXghXXXXghXXXghXXXXghXXXghXXXghXXXXghXXXXXXXghXXXgh p iXippi pi pi ippi pi ippi pi pi ippi pi ippi pi pi ippi pi ippi pi pi pi pi ippi pi pi pi pi pi pi pi pi XXXpXpXXXpXXXpXpXXXpXXXpXpXXXpXXXpXpXXpXpXXXpXpXpXpXpXpXXpXpX ipXXpiXiXiXXpiXiXXpiXiXiXXpiXiXXpiXiXiXXpiXiXXpiXiXiXpiXiXiXXpiXiXiXiXiXiXiXpiXiXiXip XXipXXXXipXXXipXXXXipXXXipXXXXipXXXipXXXipXXXXipXXXXXXXipXXXip pipXipipip p p ipipip p ipipip p p ipipip p ipipip p p ipipip p ipipip p p p p ipipip p p p p p p p p iXXXiXiXXXiXXXiXiXXXiXXXiXiXXXiXXXiXiXXiXiXXXiXiXiXiXiXiXXiXiX XXipXpXpXXipXpXXipXpXpXXipXpXXipXpXpXXipXpXXipXpXpXipXpXpXXipXpXpXpXpXpXpXipXpXpXip XXXiXiXXXiXXXiXiXXXiXXXiXiXXXiXXXiXiXXiXiXXXiXiXiXiXiXiXXiXiX ipXXipXpXpXXipXpXXipXpXpXXipXpXXipXpXpXXipXpXXipXpXpXipXpXpXXipXpXpXpXpXpXpXipXpXpXip XXXiXiXXXiXXXiXiXXXiXXXiXiXXXiXXXiXiXXiXiXXXiXiXiXiXiXiXXiXiX ipXXipXpXpXXipXXXipXXXXipXXXipXXXXipXXXipXXXipXXXXipXXXXXXXipXXXip ip XXipXiXiXXipXiXXipXiXiXXipXiXXipXiXiXXipXiXXipXiXiXipXiXiXXipXiXiXiXiXiXiXipXiXiXip pXXiXiXXXiXXXiXiXXXiXXXiXiXXXiXXXiXiXXiXiXXXiXiXiXiXiXiXXiXiX iXXpXpXXXpXXXpXpXXXpXXXpXpXXXpXXXpXpXXpXpXXXpXpXpXpXpXpXXpXpX ipip p ipip p p ipipip p ipipip p p ipipip p ipipip p p p p ipipip p p p p p p p p pipXXXpXpXXXpXXXpXpXXXpXXXpXpXXXpXXXpXpXXpXpXXXpXpXpXpXpXpXXpXpX iXXipXiXiXXipXiXXipXiXiXXipXiXXipXiXiXXipXiXXipXiXiXipXiXiXXipXiXiXiXiXiXiXipXiXiXip pXXXpXpXXXpXXXpXpXXXpXXXpXpXXXpXXXpXpXXpXpXXXpXpXpXpXpXpXXpXpX iXXXiXiXXXiXXXiXiXXXiXXXiXiXXXiXXXiXiXXiXiXXXiXiXiXiXiXiXXiXiX XXipipXXXXipipXXXipipXXXXipipXXXipipXXXXipipXXXipipXXXipipXXXXipipXXXXXXXipipXXXipip ipip ipip ipip ip ip ip ip ip

ui = f i ∂ui =0 ∂xi ui = g i 1 ∂ui nj + pni = 0 − Re ∂xj Γ1

1 h V Γ Vh 0 u uh

0 Γ0 1 2 i j Ω Inﬂow Solid wall Γ0 n Γ0 Γ1 Ω Figure 4.13: A simple example of a domain. M2 Γ0 M m m+2 1 h V Γ0 Γ Vh 0 u uh Γ0 Γ1 Ω Γ0

ww xx yy

uu vv dd

ss tt ee ff

rr qqq

in Ω, in Ω. on Γ0 on Γ1 .

(4.23a) (4.23b) (4.23c) (4.23d)

Following strategies can be used to tackle the problem of example 2. (i) Use more degrees of freedom for the velocity—for instance continuous, piecewise quadratics on each triangle—to balance the large number of equations associated with the incompressibility condition. (ii) Accept that the discrete velocities only is approximately divergence-free. (iii) Associate the velocity with the midpoints of edges instead of the vertices. This also gives more degrees of freedom for the velocities. (There are ∼ 3I 2 edges, but only ∼ 2I 2 vertices). Strategies (i) and (ii) are the “standard” methods, whereas (iii) leads to “non-conforming” approximations related to ﬁnite-diﬀerence and ﬁnite-volume approximation with “staggered mesh”.

4.2.1

A variational form of the Navier–Stokes equations

To derive the variational form, we need new integration-by-part formulas. The product rule of diﬀerentiation yields ∂ ∂ui ∂zi ∂ui zi = + z i 2 ui (4.24) ∂xj ∂xj ∂xj ∂xj and ∂ ∂p ∂zj (zj p) = zi +p . ∂xj ∂xi ∂xj (4.25)

Integrating expressions (4.24) and (4.25) and applying the the divergence theorem (4.3) on the left sides yields ∂ui ∂zi ∂ui nj z i dΓ = dΩ + zi 2 ui dΩ (4.26) ∂xj ∂xj ∂xj

Γ Ω Ω

and nj zj p dΓ =

Γ Ω

zi

∂p dΩ + ∂xi

p

Ω

∂zi dΩ, ∂xi

(4.27)

which are the integration-by-parts formulas needed to derive our variational form. Now, let zi be a smooth vector-valued function on Ω that vanishes on Γ0 . Multiply both sides of equation (4.23a) with zi , integrating over Ω, and utilizing (4.26) and (4.27), we ﬁnd that zi fi dΩ =

Ω Ω

z i uj

∂ui dΩ + ∂xj

zi

Ω

∂p 1 dΩ − ∂xi Re

zi

Ω

2

**ui dΩ nj z i ∂ui dΓ − ∂xj p
**

Ω

=

Ω

∂ui 1 z i uj dΩ + ∂xj Re z i uj 1 ∂ui dΩ + ∂xj Re

Ω

∂zi ∂ui 1 dΩ − ∂xj ∂xj Re ∂zi ∂ui dΩ − ∂xj ∂xj

Γ

∂zi dΩ + ∂xi

nj zj p dΓ (4.28)

Γ

=

Ω

p

Ω

Ω

∂zi dΩ, ∂xi

where, in the last equality, the boundary integrals vanish on Γ0 since zi = 0 on Γ0 and on Γ1 due to boundary condition (4.23d). Moreover, multiplying equation (4.23b) with a smooth function q and integrating yields q

Ω

∂ui dΩ = 0 ∂xi

(4.29)

Thus, we conclude that a classical solution (a twice continuously diﬀerentiable solution) to the Navier– Stokes equations satisﬁes the variational forms (4.28) and (4.29) for all smooth functions z i and q such that zi vanishes on Γ0 . Similarly as for the advection–diﬀusion problem, the variational form will be used to deﬁne weak solutions, without explicit reference to the diﬀerential equations. A diﬀerence with advection–diﬀusion problem is that we here need diﬀerent function spaces for the velocity components and the pressure. Also note that ui = gi on Γ0 whereas zi vanishes on Γ0 . To avoid this last complication and simplify the exposition, we will only consider homogeneous boundary conditions, gi = 0. Thus, the equations will be “driven” only by the right-hand side fi , which, for instance, could be gravity or magnetic forcing for a electrically conducting ﬂuid. The pressure space is H = L2 (Ω) = q| q 2 dΩ < +∞

Ω

,

and the space of velocity components is V = u| ∂u ∂u dΩ < +∞ and u = 0 on Γ0 ∂xi ∂xi .

Ω

For the analysis in section 4.2.4, it will be convenient to work with the space V of the velocity vector u = (u1 , u2 ). Saying that u ∈ V is equivalent to saying that that each velocity component u i ∈ V . The energy norm for elements of V is denoted u whereas the L2 (Ω) norm is

1/2

V

=

Ω

∂ui ∂ui dΩ ∂xj ∂xj

1/2

,

u

L2 (Ω)

=

Ω

ui ui dΩ

.

We assume that both boundary conditions really are active. In the case of only Dirichlet boundary condition, that is, that Γ0 constitute the whole boundary, the pressure is only possible to specify uniquely up to an additive constant. The variational problem deﬁning weak solutions to equation (4.23) is Find ui ∈ V and p ∈ H such that ∂ui 1 ∂zi ∂ui z i uj dΩ + dΩ − ∂xj Re ∂xj ∂xj

Ω Ω

p

Ω

∂zi dΩ = ∂xi

zi fi dΩ

Ω

∀zi ∈ V

Ω

∂ui q dΩ = 0 ∂xi

∀q ∈ H

This is a variational problem of mixed type: two diﬀerent spaces are involved. As for the advection–diﬀusion problem, the boundary conditions are incorporated in the variational formulation. The essential boundary conditions are here that ui = 0 on Γ0 and is explicitly assigned through the deﬁnition of V . The natural boundary conditions are the boundary conditions on Γ1 , implicitly speciﬁed through the variational form. Note that the natural boundary condition is diﬀerent than for the advection–diﬀusion problem. Remark 2. The mathematical properties of the Navier–Stokes equations are complicated, and important issues are still unresolved. In fact, there is a $ 1 000 000 prize for the person that can develop a theory that can satisfyingly explain their behavior in three space dimensions! A short summary of what is known is the following. For the unsteady problem on a given, ﬁnite time interval (0, T ), and • in two space dimensions: – a unique weak solution exists for any square-integrable f and for any Re, – smooth data yield smooth solutions, so that weak solutions are classical solutions if data is smooth enough; • in three space dimensions: – a weak solution exists for any square-integrable f and for any Re. – It is not known whether the weak solution always is unique (besides during some initial interval (0, T ∗ ), where the size of T ∗ is unknown). – It is not known whether smooth data always yields a smooth solution, or if singularities can evolve from smooth data. The steady problem has at least one weak solution, in 2D as well as 3D. Note that the steady problem may have several solutions, even in 2D. Note that the natural boundary condition here not just is a Neumann condition, but the more complicated condition (4.23d).

4.2.2

Finite-element approximations

To obtain a FEM, we choose subspaces Vh ⊂ V , Hh ⊂ H of piecewise polynomials, and replace V with Vh and H with Hh in the variational form. In 2D and in components, we obtain Find uh , vh ∈ Vh and ph ∈ Hh such that z h uh

Ω

∂uh ∂uh + vh ∂x ∂y

dΩ +

1 Re

Ω

∂zh ∂uh dΩ − ∂xj ∂xj

ph

Ω

∂zh dΩ = ∂x

Ω

zh f1 dΩ zh f2 dΩ

Ω

∀zh ∈ Vh , ∀zh ∈ Vh ,

(4.30a) (4.30b) (4.30c)

zh

Ω

∂vh ∂vh uh + vh ∂x ∂y ∂uh ∂vh + ∂x ∂y

1 dΩ + Re

Ω

∂zh ∂vh dΩ − ∂xj ∂xj

∂zh ph dΩ = ∂y

Ω

qh

Ω

dΩ = 0.

We postpone for a while the question on how to choose the subspaces Vh and Hh .

4.2.3

**The algebraic problem in 2D
**

Np

**Expanding ph in basis for Hh , we obtain ph (x, y) =
**

n=1

p(n) ψ (n) (x, y) .

(4.31)

**Likewise, expanding uh , vh in basis for Vh yields
**

Nu Nu

uh (x, y) =

n=1

u(n) ϕ(n) (x, y) ,

vh (x, y) =

n=1

v (n) ϕ(n) (x, y) .

(4.32)

Now insert expansions (4.31) and (4.32) into equations (4.30). In (4.30a) and (4.30b), choose z h = ϕ(m) for m = 1, . . . , Nu ; in (4.30c), choose qh = ψ (m) for m = 1, . . . , Np . Then, equations (4.30) become

Nu

u(n)

n=1 Ω

ϕ(m) uh

Np

∂ϕ(n) ∂ϕ(n) + vh ∂x ∂y ∂ϕ ψ (n) dΩ = ∂x

(m)

dΩ +

1 u (n) u Re n=1

N

∂ϕ(m) ∂ϕ(n) ∂ϕ(m) ∂ϕ(n) + ∂x ∂x ∂y ∂y

Ω

dΩ

−

Nu

p(n)

n=1 Ω

ϕ(m) f1 dΩ

Ω

m = 1, . . . , Nu ∂ϕ(m) ∂ϕ(n) ∂ϕ(m) ∂ϕ(n) + ∂x ∂x ∂y ∂y

Ω

v (n)

n=1 Ω

ϕ(m) uh

Np

∂ϕ ∂ϕ + vh ∂x ∂y

(n)

(n)

dΩ +

1 u (n) v Re n=1

N

dΩ

(4.33)

−

Nu

p(n)

n=1 Ω (n)

∂ϕ(m) (n) ψ dΩ = ∂y

Ω Nu

ϕ(m) f2 dΩ ∂ϕ(n) dΩ = 0 ∂y

m = 1, . . . , Nu

u(n)

n=1 Ω

ψ (m)

∂ϕ dΩ + v (n) ∂x n=1

ψ (m)

Ω

m = 1, . . . , Np

**Placing the coeﬃcients into column matrices, u = u(1) , . . . , u(N u) p = p(1) , . . . , p(Np )
**

T T

, ,

v = v (1) , . . . , v (N u)

T

,

we can write the system (4.33) in the matrix–vector form (x) 1 C(uh , vh ) + Re K 0 G(x) u b 1 0 C(uh , vh ) + Re K G(y) v = b(y) , p 0 −G(x)T −G(y)T 0

(4.34)

**where the matrix and vector components are Kmn =
**

Ω

∂ϕ(m) ∂ϕ(n) ∂ϕ(m) ∂ϕ(n) + ∂x ∂x ∂y ∂y ∂ϕ(m) ψ (n) dΩ, ∂x

Ω

dΩ,

Cmn (uh , vh ) =

Ω

ϕ(m) uh

∂ϕ(n) ∂ϕ(n) + vh ∂x ∂y b(y) = m

Ω

dΩ,

G(x) = − mn

G(y) = − mn

Ω

∂ϕ(m) ψ (n) dΩ, ∂y

b(x) = m

Ω

ϕ(m) f1 dΩ,

ϕ(m) f2 dΩ. (4.35)

Matrix K represents an approximation of the discrete negative Laplacian − 2 ; C(uh , vh ) the discrete ∂ ∂ advection operator u ∂x + v ∂y ; G(x) and G(y) the discrete gradient ; and G(x)T and G(y)T the discrete divergence. That is, the structure of the Navier–Stokes equations are directly reﬂected in the nonlinear system (4.34). We obtained the same structure of the nonlinear equation for the ﬁnite-volume discretization, except that here, it is guaranteed the the last block row of the matrix is the negative transpose of the last block column.

4.2.4

Stability

So far, the approximating spaces Vh and Hh have not been speciﬁed. The counting argument of example 2 indicates that not any choice is good. We will limit the discussion to the Stokes equations, whose ﬁniteelement formulation reads Find uh ∈ Vh and ph ∈ Hh such that i

Ω h ∂zi ∂uh i dΩ − ∂xj ∂xj h ∂zi dΩ = ∂xi

ph

Ω

h zi fi dΩ Ω

h ∀ z i ∈ Vh ,

(4.36a)

Ω

∂uh qh i dΩ = 0 ∂xi

∀ q h ∈ Hh .

(4.36b)

The Stokes equations is a limit case of the Navier–Stokes equations for very low Reynolds numbers and assume that inertia can be neglected compared to viscous forces. The Stokes equations may be used to model very viscous ﬂow, creeping ﬂow, lubrication, but also ﬂow of common ﬂuids (like air and water) at the micro scale. Due to the lack of nonlinear advection, these are linear equations. Mathematically, the Stokes equations are much easier to analyze than the full Navier–Stokes equations. Indeed, there exists a unique weak solutions for each square-integrable fi both in two and three space dimension. The reason to start the analysis on the Stokes equation is that a stable Stokes discretization is necessary to obtain a stable Navier–Stokes discretization. As for the Navier–Stokes equations, we require that Vh ⊂ V and Hh ⊂ H. Recall that each function in V and H is required to be bounded in the energy and the L2 (Ω) norm, respectively. Thus, in makes sense to require that the velocity and pressure approximations satisfy the same bounds, that is,

1/2 1/2

ph

L2 (Ω)

=

Ω

p2 dΩ h

≤C

1/2

fi fi dΩ

Ω

=C f

L2 (Ω) ,

(4.37a) (4.37b)

uh

V

=

Ω

∂uh ∂uh i i dΩ ∂xj ∂xj

≤C f

L2 (Ω) ,

where the constant C should not depend on h. This implies that the approximations will not blow up when the mesh is reﬁned. h Choosing zi = uh in equation (4.36a), we obtain i ∂uh ∂uh i i dΩ − ∂xj ∂xj ph

Ω

Ω

∂uh i dΩ = ∂xi

Ω

∂uh ∂uh i i dΩ = ∂xj ∂xj

uh fi dΩ, i

Ω

**where the ﬁrst equality follows from equation (4.36b). Thus, uh
**

2 V

=

Ω

uh fi dΩ ≤ u i

L2 (Ω)

f

L2 (Ω)

≤C u

V

f

L2 (Ω)

(4.38)

where the Cauchy–Schwarz inequality yields the ﬁrst inequality, and the Poincar´ inequality the second. e We thus conclude from expression (4.38) that uh

V

≤C f

L2 (Ω) ,

(4.39)

that is, any Galerkin approximation of the velocity in the Stokes equations is stable! For bad choices of subspaces, we will of course obtain inaccurate velocity approximations, but they will not blow up as the mesh is reﬁned, since their energy norm is bounded by the given data fi . Regarding the pressure, the situation is more complicated. Not any combination Hh and V h yield stable pressure approximations in the sense of (4.37a). We will derive a compatibility condition between the velocity and pressure spaces, the LBB condition, necessary to yield stable pressure approximations.

4.2.5

**The LBB condition
**

h ∂zi dΩ = ∂xi h ∂zi ∂uh i dΩ − ∂xj ∂xj h h ∂zi ∂zi dΩ ∂xj ∂xj V V

h Rearranging equation (4.36a) implies that for each zi ∈ Vh ,

ph

Ω

Ω

Ω 1/2

h zi fi dΩ

(Cauchy–Schwarz) ≤

Ω

Ω L2 (Ω) V

**∂uh ∂uh i i dΩ ∂xj ∂xj f f
**

L2 (Ω) L2 (Ω) L2 (Ω) .

1/2

1/2

1/2

+

Ω

h h zi zi dΩ

fi fi dΩ

Ω

(by (4.39)) ≤ C z h

= zh (Poincar´) ≤ z h e

uh uh

V

V V

+ zh + zh

f

V

f

L2 (Ω)

+ zh

(4.40)

V

Dividing expression (4.40) with z h

yields that ph

Ω h ∂zi dΩ ≤ (C + 1) f ∂xi L2 (Ω)

1 zh

V

h for each zi ∈ Vh , from which it follows that there is a C > 0 such that

0=zi ∈V

max

1 zh

V Ω

ph

h ∂zi dΩ ≤ C f ∂xi

L2 (Ω) .

(4.41)

We have thus found a bound on the discrete pressure in the “exotic” norm (4.16). However, we were aiming for the bound (4.37a). Thus, if it happens that α ph

L2 (Ω)

≤ max

1 zh

V Ω

0=zi ∈Vh

ph

h ∂zi dΩ ∂xi

(4.42)

for some α > 0 independent of h, then it is immediate that ph L2 (Ω) ≤ C f L2 (Ω) for some C > 0. Condition (4.42) says that our exotic norm is stronger than the L2 (Ω)-norm for ph . The LBB-condition is a condition on the pairing of spaces Vh and Hh that requires the exotic norm to be stronger than the L2 (Ω)-norm for each function qh ∈ Hh . The derivation above proves thus that the LBB-condition is suﬃcient for the pressure to be stably determined by data: Theorem 3. Assume that the LBB-condition holds for the spaces H h , Vh . That is, assume that there is an α > 0 independent of h, such that, for each qh ∈ Hh ,

1/2

α

Ω

2 qh dΩ

≤ max

1 zh

V Ω

0=zi ∈Vh

qh

h ∂zi dΩ, ∂xi

(4.43)

**then there is a C > 0 such that
**

1/2 Ω 1/2

p2 dΩ h

≤C

Ω

|f |2 dΩ

holds, independently of h. The LBB condition (4.43) is satisﬁed for Vh = V and Hh = H. That is, the pressure (before discretization) is stably determined by data. The expression Ω qh ∂xii dΩ represents the discrete gradient of qh (see § 4.2.3). The LBB condition thus means that a zero discrete pressure gradient implies a zero value of the pressure. Conversely, if the LBB condition is not satisﬁed, it means that the discrete pressure gradient may be small even when the pressure is not small: The discrete gradient cannot “feel” that the pressure is large, an eﬀect that usually manifests itself as wild pressure oscillations.

∂z h

Ω M2 M m m+2

1

−1

0

1

0

1

−1

0

h V Vh −1 u uh Γ0 Γ1 1 Ω

0

1

−1

−1

0

1

Figure 4.15: A checkerboard mode for equal-order interpolation on a regular triangular mesh.

4.2.6

Mass conservation

The integral Γ ρ ni ui dΓ yields the total ﬂux of mass(in kg/s) through the boundary. There is no net mass-ﬂux through the boundary for a constant-density ﬂow, so Γ ρ ni ui dΓ = 0. A velocity approximation uh is is globally mass conservative if i

Γ

ni uh dΓ = 0, i ∂uh i dΩ = 0, ∂xi

or, equivalently, if

Ω

**by the divergence theorem (4.3). Recall from equation (4.30c) that qh
**

Ω

∂uh i dΩ = 0 ∂xi

∀qh ∈ Hh .

A Galerkin approximation is thus globally mass conservative if 1 ∈ Hh , that is, if constant functions can be represented exactly in Hh . All reasonable FE approximation satisfy this. A ﬁnite-element approximation is locally mass conservative if ∂uh i dΩ = 0, ∂xi

K

for each element K in the mesh. Not all ﬁnite-element approximations are locally mass conservative. If the pressure approximations consist of piecewise constants on each element, the FE approximation will be locally mass conservative. However, it will not be locally mass conservative if the pressure approximations are continuous and piecewise linear.

4.2.7

Choice of ﬁnite elements. Accuracy

Now we consider a regular triangulation of a 2D domain Ω, that is, a triangulation satisfying a mesh quality condition of the sort discussed in section 4.1.5. “Equal-order interpolation” Let us start by trying the approximation space we used for the advection–diﬀusion problem (section 4.1.1). That is, we use continuous functions that are linear on each triangle both for the functions in H h and Vh . This pair does not satisfy the LBB condition and yields an unstable discretization. In particular, the system matrix (4.34) is singular : the discrete gradient matrix G has linearly dependent columns. Thus, there is at least one nonzero Np -vector pc such that Gpc = 0. Corresponding function pc ∈ Hh , whose h nodal values are equal to the elements of pc is called a checkerboard mode. Figure 4.15 depicts the nodal values of a checkerboard mode for equal-order interpolation. To see this, consider the element contribution, associated with a triangle Tn , to the ith row of vector Gpc (notice from the expressions in ﬁgure 4.5 that ∂φi is constant on each triangle): ∂x −

Tn

pc h

∂φi ∂φi dΩ = − ∂x ∂x

Tn Tn

pc dΩ ≡ 0, h

h V Vh u uh Γ0 Γ1 Ω Figure 4.16: The evaluation points at each triangle for the continuous, piecewise quadratic velocity approximations of the Taylor–Hood pair. since the mean value of the function pc vanishes on each triangle, as can be seen from ﬁgure 4.15. The right h side of the LBB condition (4.43) vanishes for qh = pc , which immediately implies that the LBB cannot h hold for a space Hh that includes pc . This (and other) checkerboard mode will pollute the solution of the h Stokes as well as Navier–Stokes equations.. Constant pressure, (bi)linear velocities Now we consider a triangulation either of triangles or rectangles. Let the functions in H h be piecewise constant on each element. In case of triangular elements, the space Vh will be continuous functions, piecewise linear on each element, whereas for quadrilateral elements, continuous functions that are piecewise bilinear on each rectangle comprise Vh . Remark 3. The pressure approximations need not to be continuous in order for Hh ⊂ H. However, if Vh ⊂ V , the velocity approximations cannot be discontinuous, since v is not a function for discontinuous v. This choice leads to a locally mass-conserving approximation. However, for triangular elements, the counting argument of example 2 applies here and indicates problems with this choice of elements. Indeed, the LBB conditions is not satisﬁed for this unstable discretization. Checkerboard modes similar as in ﬁgure 4.15 exist. Still, the constant pressure—bilinear velocity space on rectangles is in widespread engineering use! “Fixes” have been developed to take care of pressure oscillations. For instance, the checkerboard mode can be ﬁltered out from the discrete pressures. Moreover, common iterative methods to solve the resulting discrete equations (such as the projection schemes) implies a stabilization of the discretization. The Taylor–Hood pair Here, the functions in Hh and Vh are globally continuous and, on each triangle, • ph ∈ Hh is linear, whereas • uh , vh ∈ Vh is quadratic. Thus, on each triangle, we have uh = a + b x + c y + d xy + e x2 + f y 2 The six coeﬃcients a–f are uniquely determined by uh ’s values at the triangle vertices and midpoints (ﬁgure 4.16). Nu A basis φ(m) (xi ) m=1 of continuous, piecewise quadratic functions interpolates the nodal values of ﬁgure 4.16, so that uh ∈ Vh can be expressed as uh (xi ) = Nu ui φ(m) (xi ). There are two kinds of basis m=1 functions here, illustrated in ﬁgures 4.17 and 4.18. The LBB condition is satisﬁed for the Taylor–Hood pair, and the following error estimates holds for a smooth solution to the Navier–Stokes equation: u − uh p − ph

L2 (Ω) L2 (Ω)

≤ Ch3

≤ Ch2

The Taylor–Hood pair is globally, but not locally mass conservative. The Taylor–Hood pair can be generalized to higher order. Then, the pressure and velocity approximations will be continuous functions, consising of piecewise polynomials of degree k for pressure and k + 1 for velocity (k ≥ 1) on each triangle.

m+2 m+2 0 d 11d 1 d 1 2h a h a ia V V a i − 1, j j 1 − i Vh − i − 1, j j 1 i, jVh 1 − Inﬂowu u j−1 i, i + 1, j − 1 Inﬂow Solid wallh uh u j i + 1, j − 1 Solid wall n0 Γ0 Γ j ΓΓ i 0 Γ1 n i 1 i+ Γ0 Γ1 1 Ω Γ+ 1 Ω i1 ui,j Ω u vi,j Ω i,j The basis function φ(m) (x ) when M 2 4.18: The basis function φ(m) (xi ) when Figure 4.17: Figure vi,j i pi,j M2 M m corresponds to an edge-midpoint node m corresponds to a corner node p pi,j M i,j m pi,j vi,j+1/2 m+2 v m vi,j−1/2 O(h2) m i,j+1/2 +2 1 vi,j−1/2 ui−1/2,j 1 h 1 ui−1/2,j i− 2 V ih 1 −2 i V Vh 1 i+ 2 i Vh 1 u i+ 2 i+1 u uh 3 i+ i+1 uh 3 Γ0 2 i+ 2 j+1 Γ0 Γ1 2 j+1 j Γ1 2 j jΩ 1 −2 Ω 1 j−2 j− Figure 4.20:1Isoparametric elements use polynomials j− Figure 4.19:1Approximating a curved boundary with of the sameA order as used for the FE approximations A a triangle sides leads to an O(h2 ) error of the bound- to approximate the shape of the boundary. B B ary condition. C 3 u 2 ,1 C 3 u 2 ,1 v1, 3 2 v1, 3 Remark 4. 2 Inaccuracies in approximations of the domain boundary may reduce the convergence rate of p1,1 p1,1 higher-order approximations. For instance, a straight-forward triangulation of a domain with a curved 0 boundary will lead to a boundary-condition error at the mid nodes of O(h2 ) (ﬁgure 4.19). This is ﬁne 0 1 for piecewise-linear approximations since the error anyway is O(h2 ). However, this inaccuracy at the 1 2 boundary results in a half-order reduction of the convergence rateiof piecewise quadratic approximations. 2 Using curved elements restores convergence rate. The most common strategy for curved elements is known i j j as Isoparametric Elements: piecewise polynomials of the same degree as used in the FE approximation are Inﬂow used toInﬂow approximate the shape of the boundary (ﬁgure 4.20). wall Solid Solid wall n n Γ0 A stable approximation with piecewise-linear velocities and pressures Γ0 Γ1 Instead of Γ1 using diﬀerent approximation orders, as in the Taylor–Hood pair, we here consider diﬀerent Ω meshes for Ω velocity and pressure. Both the velocity and pressure approximations consist of continuous, M2 piecewise-linear functions on triangles, but each pressure triangle is subdivided into four velocity triangles M2 M as in ﬁgure 4.21. M m This is a stable approximation satisfying the LBB condition, + 2 the following error estimates holds for m m and m+2 1 1 h h V V Vh Vh u u uh uh Γ0 Γ0 Γ1 Γ1 Ω Ω Figure 4.22: Subdividing a pressure rectangle into Figure 4.21: Subdividing a pressure triangle into four four velocity rectangles yields a stable approximation velocity triangles yields a stable approximation with with continuous piecewise-bilinear functions for both continuous, piecewise-linear functions for both veloc- velocity and pressure. ity and pressure.

1 h V Vh u uh Γ0 Γ1 Ω

j

1 h V Vh u uh Γ0 Γ1 Ω Figure 4.24: Approximations spanned by the basis functions (4.23) yields piecewiselinear functions that are continuous only along at the edge midpoints, in general.

Figure 4.23: The basis functions associated with the velocity approximation for a nonconforming but stable discretization.

**a smooth solution to the Navier–Stokes equation: u − uh
**

L2 (Ω) L2 (Ω)

p − ph

≤ Ch2 ≤ Ch

These approximations use the same number of unknowns and the same nodal locations as for the Taylor– Hood pair, but are one order less accurate. Calculating the matrix elements (4.35) is however somewhat easier than for the Taylor–Hood pair. There is also a version of this discretization for rectangular meshes: Each pressure rectangle is then subdivided into four velocity rectangles, as in ﬁgure 4.22. Both the velocity and pressure approximations consist of continuous and piecewise-bilinear functions. A nonconforming method The discussion after example 2 suggested a strategy to address the problem with the overdetermined incompressibility condition, namely to use nodal points on the edge mid points for the velocity. To accomplish this, consider basis functions, associated with the edges, of the type depicted in ﬁgure 4.23. Basis function φj is linear on each triangle and continuous along edge j (but discontinuous along four edges!). If x i , y i are the coordinates of the mid point of edge i, we have φj (xi , y i ) = 1 for i = j, 0 for i = j.

**Now deﬁne the space of the velocity components Vh from all possible expansions
**

N N

uh (x, y) =

j=1

uj φj (x, y),

vh (x, y) =

j=1

vj φj (x, y),

Here, N is the total number of edges, excluding those on Γ1 . Note that uh will be continuous only at edge mid points in general. Thus, Vh ⊂ V since uh is not a square-integrable (in fact, constant) function. Violating a property like Vh ⊂ V is called a variational crime. However, uh restricted on each element is a square-integrable function. Letting Hh be piecewise constant on each triangle and Vh deﬁned as above, we deﬁne the nonconforming FE approximation: Find uh , vh ∈ Vh and ph ∈ Hh such that

M

z h uh

n=1T n M

∂uh dΩ + ∂x n=1 ∂vh dΩ + ∂x n=1

M

M

zh vh

Tn

∂uh 1 dΩ + ∂y Re n=1 ∂vh 1 dΩ + ∂y Re n=1 ∀ q h ∈ Hh ,

M

M

Tn

**∂zh ∂uh dΩ − ∂xj ∂xj n=1 ∂zh ∂vh dΩ − ∂xj ∂xj n=1
**

M

M

ph

∂zh dΩ = ∂x

Ω

zh f1 dΩ

∀ z h ∈ Vh , ∀ z h ∈ Vh ,

Tn

z h uh

n=1T n M

zh vh

ph

∂zh dΩ = ∂y

Ω

zh f2 dΩ

Tn

Tn

Tn

qh

n=1T n

∂uh ∂vh + ∂x ∂y

dΩ = 0

where M is the number of triangles.

The fact that Vh ⊂ V complicates the analysis of this discretization. Nevertheless, the approximation is stable, although not as accurate as the Taylor–Hood pair, for instance. Other nice properties with this method is that it is locally mass conservative, and quite easy to implement. But maybe the most interesting with this discretization is that it allows for a construction of a divergence free velocity basis. In all types of discretizations discussed so far, including the present, we have used the same basis for the uh and vh components. By a clever combination of basis functions of the type illustrated (j) (j) in ﬁgure 4.23, it is possible to construct a new vector-valued basis functions η j = (zh , wh ) for the velocity having the property · ηj = Expanding the velocity in this basis,

M

∂zh ∂wh + ∂x ∂y

(j)

(j)

= 0.

uh = (uh , vh ) =

j=1

uj η j ,

(4.44)

yields that · uh = 0. Inserting expansion (4.44) into equations (4.30), and choosing (zh , wh ) = η i , the continuity equation and the pressure variable vanishes, and equations (4.30) reduce to

M M dNu

uj

j=1 Ω

η i · (uh ·

) η j dΩ + ν

j=1

uj

Ω

ηi ·

η j dΩ +

j=1

uj

Ω

ηi · (uh ·

) η j dΩ =

Ω

η i · f dΩ, (4.45)

for i = 1, . . . , M . The Navier–Stokes equations then reduces to a vector-valued, non-linear advection– diﬀusion problem, quite similar to equation (4.10). This leads to a substantial simpliﬁcation when designing an algorithm to solve the discretized equations; in fact, one of the great challenges for any solver of the Navier–Stokes equations is how to treat the incompressibility condition (4.30c). Unfortunately, the construction of the basis functions η j is not straightforward.

Stabilization of unstable discretizations Stability of the choices of elements presented in sections 4.2.7–4.2.7 relied on manipulations of the local polynomial order, the meshes, or the conformity of the method. These manipulations complicate the implementation. An alternative is to use the same approximations for velocity and pressure (equal-order interpolation) combined with a stabilization method. This circumvents the LBB condition. We will present a very simple idea of this sort for the Stokes equations. First choose the same type of approximations for the pressure and the velocity components, for instance continuous, piecewise-linear functions on a triangular mesh, and replace equations (4.36) with Find uh ∈ Vh and ph ∈ Hh such that i

Ω h ∂zi ∂uh i dΩ − ∂xj ∂xj

ph

Ω

h ∂zi dΩ = ∂xi

h zi fi dΩ Ω

h ∀ z i ∈ Vh ,

(4.46a)

Ω

∂qh ∂ph dΩ + ∂xi ∂xi

Ω

∂uh qh i dΩ = 0 ∂xi

∀ qh ∈ Hh .,

(4.46b)

for some small > 0. Equation (4.46b) formally inconsistent by O( ) with corresponding variational form. More advanced stabilization methods avoids this inconsistency by using a term that vanishes as h → 0. The main advantage with stabilization methods is the simpliﬁed implementation, but the method introduces an extra parameter that may need judicious tuning. Note also that the increased number of pressure unknowns, as compared with the discretization in section 4.2.7, does not increase the accuracy. The added resolution introduced is ﬁltered away by the introduction of the stability term. h To see that approximation (4.46) is stable, choose zi = uh , qh = ph and add equations (4.46a) i

**and (4.46b), ∂uh ∂uh i i dΩ + ∂xi ∂xi ∂ph ∂ph dΩ = ∂xi ∂xi uh fi dΩ i
**

1/2 1/2

Ω

Ω

Ω

(Cauchy–Schwarz) ≤ (Poincar´) ≤ C e ≤C

Ω

uh uh dΩ i i ∂uh i

Ω

fi fi dΩ

Ω 1/2 1/2

∂uh i

∂xi ∂xi

dΩ

Ω

**fi fi dΩ ∂uh ∂uh i i dΩ ∂xi ∂xi
**

1/2 1/2

Ω

∂ph ∂ph dΩ + ∂xi ∂xi

fi fi dΩ

Ω

.

Ω

Dividing and squaring yields the stability estimate ∂uh ∂uh i i dΩ + ∂xi ∂xi ∂ph ∂ph dΩ ≤ C ∂xi ∂xi fi fi dΩ.

Ω

Ω

Ω

Since C > 0 does not depend on h, both the velocity and pressure approximations cannot blow up as h → 0, even though the discretization is unstable for = 0.

j j−1 2 j−1 A B C 3 u 2 ,1 v1, 3 2 p1,1 0 1 2 i j Inﬂow Solid wall A.1 Iterative solutions to linear systems n Γ0 Γ1 Gauss-Seidel method Ω The Laplace equation on a Cartesian grids can be discretized as M2 M Φi+1,j − 2φi,j + Φi−1,j + β 2 (Φi,j+1 − 2Φi,j + Φi,j−1 ) = 0 m m+2 where β = ∆x/∆y. We can solve for Φi,j , 1 Φi+1,j + Φi−1,j + β 2 (Φi,j+1 + Φi,j−1 ) h Φi,j = V 2 (1 + β 2 ) Vh Gauss-Seidel uses this expression to update Φi,j (using the already updated values calculated in same u iteration), i.e.u h Γ0 k+1 k+1 2 Φk Φk i+1,j + Φi−1,j + β i,j+1 + Φi,j−1 Γ1 Φk+1 = i,j 2 (1 + β 2 ) Ω

Appendix A

Background material

i−1

i

i+1

j−1

This iteration method has a slow convergence, it is possible to show Φ − Φk ≤ ρ m Φ − Φ0 where ρ = 1 − O h2 h = ∆x = ∆y

e.i. the error is reduced by O h2 each iteration. Requiring an error reduction to O h2 the number of iterations must satisfy ρ m = O h2 which implies that m=O ln(h) ln(ρ) =O ln(h) ln(1 − O(h2 )) = O −h2 ln(h)

To proceed we note that we have N = n2 unknown interior nodes, see Figure A.1. Thus 95

ite

: old values

ra

tio

: new values

j

n

di

re ct

: to be updated

io

n

j+1

Ω 1 M2 h = ∆x = ∆y = ≈ n−1 = N −1/2 M n+1 m which implies that m+2 1 m = O (N ln(N )) h and that the total work is V Vh W = O N 2 ln(N ) u uh since the work per iteration is O (N ). Γ0 Γ1 y Ω : to be updated : new values1 : old values n i−1 i i+1 j−1 j j+1 iteration direction 1 1 n 1 Figure A.1: Domain for the solution of the Laplace equation. x

Multigrid method

A general way to accelerate the convergence of e.g. the Gauss-Seidel method is provided by the Multigrid method. We note that the Gauss-Seidel provides good smoothing of the local error, but converges slowly since it takes time for boundary information to propagate into the domain. The idea behind the multigrid method is that a slowly converging low-frequency on a ﬁne grid is a fast converging high-frequency on a coarse grid. Let LΦi,j = Φi,j+1 − 2Φi,j + Φi,j−1 Φi+1,j − 2Φi,j + Φi−1,j + ∆x2 ∆y 2

and deﬁne the correction to the solution to the true discrete solution in the following manner Φi,j Φk i,j ∆Φi,j This implies that L∆Φi,j = −LΦk ≡ −Ri,j i,j since LΦi,j = 0. Thus we are left with an equation implying that the Laplace operator on the correction is equal to the negative of the residual. This residual equation is solved on a coarser grid and the correction is interpolated onto ﬁner grid. In order to deﬁne the algorithm we use the following deﬁnitions I transfer operator between grids: 2 An example of a restriction operator I1 is to 2 I1 - restriction operator from grid 1 to grid 2 1 I2 - prolongation operator from grid 2 to grid 1 1 deﬁne it as an injection, i.e. we use every other value in each direction. The prolongation operator I 2 may be deﬁned by linear interpolation for the intermediate values, i.e. an average of right and left values = Φk + ∆Φi,j i,j − solution at iteration level k − correction to Φk to true discrete solution i,j

: old values i−1 i i+1 j−1 j j+1 iteration direction x y 1 n

y

ﬁne grid (1) coarse grid (2)

j x i Figure A.2: Two level grid. and top and bottom values. The values marked with × in ﬁgure A.2 are then taken as an average of the averages of the intermediate values. Now we can deﬁne the following two level scheme: 1. LΦi,j iterated k times

2 2. L∆Φi,j = −I1 Ri,j 1 3. Φi,j = Φk + I2 ∆Φi,j i,j

⇒

LΦk = Ri,j i,j

iterated k times, ∆Φ0 = 0 (starting value) (here ∆Φi,j is deﬁnied on grid 2) i,j and goto 1.

at the end of 1. convergence is checked. Wesseling [10] estimated the work of the multigrid method to be W = O (N ln N ), a substantial improvement over the Gauss-Seidel method.

i y = x2 i i+1 j j−1 ﬁne grid (1) j j coarse grid (2) +1 iteration direction r x y e1 e1 1 x = x1 n e3 e2 x z = x3 y i Figure A.3: Cartesian coordinate system. j ﬁne grid (1) x2 coarse grid (2) x2 e2 e2 e1 e1 x1 a e1 e3 x3 x1

e3 x3

Figure A.4: Two diﬀerent coordinate systems.

A.2

Cartesian tensor notation

Tensor notation is a compact way to write vector and tensor formulas. Vectors and tensors are objects which are independent of the coordinate system they are represented in. In the Cartesian coordinate system in Figure A.3 we have the unit vectors |ek | = 1 k = 1, 2, 3

The scalar product of two unit vectors is the Kronecker’s delta, δkl ek · el = δkl = and the position vector r is

3

1 k=l 0 k=l

r = x 1 e1 + x2 e2 + x3 e3 =

k=1

xk e k = x k e k .

The last expression make use of Einstein’s summation convention: when an index occurs twice in the same expression, the expression is implicitly summed over all values for that index. Example δkk = δll = δ11 + δ22 + δ33 = 3 δij aj = δi1 a1 + δi2 a2 + δi3 a3 = ai .

A.2.1

Orthogonal transformation

Consider two arbitrary oriented coordinate systems as in Figure A.4. Assume identical origin, a = 0. The unit vectors are orthogonal ek · el = δkl

n x y i j ﬁne grid (1) coarse grid (2)

e 1 x1

e1 e2

e2 · e1 x1 = cos (x2 , x1 ) x1 = c21 x1 Figure A.5: Projection of e1 x1 on e2 . and a position vector is independent of coordinate system, r = x l el = xl el . Component k of r can be written as ek · r = xl ek · el = xl el · ek = xl δkl = xk where we have the transformation tensor ck l = ek · el = cos(xk , xl ). Since cos(x2 , x1 ) = cos(x1 , x2 ) in general, we have (for k = 1, l = 2) cl k = el · ek = cos(xl , xk ) = ck l From ek · r = x l ek · el = xl ek · el Example xk = ckl xl = ck1 x1 + ck2 x2 + ck3 x3 = cos(xk , x1 )x1 + cos(xk , x2 )x2 + cos(xk , x3 )x3 = e k · e 1 x1 + e k · e 2 x2 + e k · e 3 x3 which for component 2 reads x2 = e 2 · e 1 x1 + e 2 · e 2 x2 + e 2 · e 3 x3 x2 is projection of e1 x1 , e2 x2 , e3 x3 on e2 . A visualization of the projection of e1 x1 on e2 can be seen in Figure A.5. The relation between ckl and cml can be studied through xk

δkm xm

⇒

x k = e k · e l xl = c k l xl

⇒

xk = el · ek xl = cl k xl = clk xl

we conclude that ﬁrst index refer to primed system!

= ckl xl

cml xm

= ckl cml xm

⇒

(ckl cml − δkm ) xm = 0.

**In the same manner we have xk
**

δkm xm

= clk

xl

clm xm

= clk clm xm

⇒

(clk clm − δkm ) xm = 0.

A.2.2

Cartesian Tensors

Def. a Cartesian tensor of rank R is a set of quantities Tklm... which transform as a vector in each of its R indices. Tklm... = ckr cls cmp . . . Trsp...

**Example For scalars the rank is R = 0 a · b = ak bk = ckl ckm al bm = al bl ,
**

δlm

for vectors R = 1 ak = ckl al and for a tensor of second order we have R = 2 δkl = ckm cln δmn = ckm clm = δkl . δkl is isotropic, i.e. identical in all coordinate systems. An outer product generates a tensor Tkl Srs = ckm cln crp csq Tmn Spq and an inner product generates a tensor Tkl Sls = ckm cln clp csq Tmn Spq = ckm csq Tmn Snq

δnp

A.2.3

Permutation tensor

εklm 1 even −1 odd = if klm is permutation of (1, 2, 3) 0 no even permutation odd permutation no permutation ε123 = 1 ε213 = −1 ε221 = 0

As examples of permutations of 1,2,3 we have (1,2,3), (3,1,2), (2,3,1) (2,1,3), (3,2,1), (1,3,2) at least 2 subscripts equal

A.2.4

**Inner products, crossproducts and determinants
**

a · b = δij ai bj = ai bi = a1 b1 + a2 b2 + a3 b3 .

The inner product of the vectors a and b is

The cross product of a and b can be written as a×b= e1 a1 b1 e2 a2 b2 e2 a3 b3 = (a2 b3 − a3 b2 )e1 + (a3 b1 − a1 b3 )e2 + (a1 b2 − a2 b1 )e3 = εijk ei aj bk

and the determinant of the tensor aij is |{aij }| = = a11 a21 a31 a12 a22 a32 a13 a23 a33 = εijk a1i a2j a3k

1 1 εijk (ai1 aj2 ak3 + ai3 aj1 ak2 + ai2 aj3 ak1 − ai2 aj1 ak3 − ai3 aj2 ak1 − ai1 aj3 ak2 = εijk εpqr aip ajq akr . 6 6 We have the permutation relation εklm εksp = δls δmp − δlp δms . If we set s = l in (A.1) we get and if we also use p = m we end up with εklm εklp = (3 − 1)δmp = 2δmp εklm εklm = 6 The permutation tensor can also be used to derive the following vector relation (a × b) · (c × d) = εijk aj bk εilm cl dm = (δjl δkm − δjm δlk )aj bk cl dm = aj cj bk dk − aj dj bk ck = (a · c)(b · d) − (a · d)(b · c) (A.1)

A.2.5

Second rank tensors

All second rank tensors can be decomposed into symmetric and antisymmetric parts, T kl = T(kl) + T[kl] , where the antisymmetric part is 1 T[kl] = −T[lk] = (Tkl − Tlk ) 2 and the symmetric part 1 T(kl) = T(lk) = (Tkl + Tlk ) . 2 The symmetric part can be decomposed into isotropic and traceless part, T(kl) = T kl + T kl , the isotropic part being 1 T kl = Tmm δkl 3 and the traceless part is 1 T kl = T(kl) − Tmm δkl . 3 If we project the isotropic, traceless and entisymmetric parts we get S(kl) T[kl] = S kl T kl = 1 1 S(kl) T[kl] + S(lk) T[lk] = S(kl) T[kl] − S(lk) T[kl] = 0 2 2 = 1 1 Smm Tkk − Tkk δll 3 3 =0

1 1 Smm δkl T(kl) − Tmm δkl 3 3 ⇒ Skl T[kl] Skl T(kl) Skl T kl Skl T kl

= S[kl] T[kl] = S(kl) T(kl) = S kl T kl = S kl T kl ,

and each part is invariant under transformation T[kl] = ckr cls T[rs] = −ckr cls T[sr] = −clr cks T[rs] = −T[lk] Trr = crp crq Tpq = δpq Tpq = Tpp 1 T kl = ckp clq T pq = ckp clq T(pq) − Trr δpq 3 1 = T(pq) − Trr δkl . 3

Since we only need three components to completely describe an antisymmetric tensor of rank R = 2 it can be represented by its dual vector di = εijk Tjk = $$$$ + εijk T[jk] εijk T(jk) The ﬁrst term on the right hand side is zero since εijk is antisymmetric in any two indecies. Multiply both sides by εilm εilm di = εilm εijk Tjk = (δlj δmk − δlk δmj ) Tjk = Tlm − Tml = 2T[lm] Note: There are 3 independent components in T[lm] and di . In order to sum up, an arbitrary tensor of rank 2 can be divided into the following parts Tkl = T(kl) + T[kl] = T kl + T kl + T[kl] = 1 1 Trr δkl + T(kl) − Trr δkl + 3 3

isotropic traceless

⇒

T[lm] =

1 εilm di 2

1 εikl di 2

antisymmetric

Properties of the above parts are invariant under transformation. In the Navier-Stokes equations we have ∂ui the velocity gradient ∂xj . The antisymmetric part of this tensor describes rotation, the isotropic part the volume change and the traceless part describes the deformation of a ﬂuid element.

Example We have the second rank tensor

which can be divided into a symmetric part

1 2 3 {Tkl } = 4 5 6 7 8 9 1 3 5 = 3 5 7 5 7 9

T(kl) and an antisymmetric part T[kl]

The symmetric part can then be divided into a isotropic 5 T kl = 0 0 and a traceless part T kl The dual vector of the antisymmetric tensor is

0 −1 −2 = 1 0 −1 . 2 1 0 part 0 0 5 0 0 5

−4 3 5 = 3 0 7 . 5 7 4

−2 {di } = 4 . −2

A.2.6

Tensor ﬁelds

We have the tensor ﬁeld Tij (x1 , x2 , x3 ) ≡ Tij (xk ) with the following properties • partial derivative transforms like a tensor ∂xk ∂ ∂ ∂ = = cpk ∂xp ∂xp ∂xk ∂xk • gradient adds one to tensor rank ∂ Tij ∂xk where (xk = cpk xp )

• divergence decreases tensor rank by one ∂ ∂T1l ∂T2l ∂T3l Tkl = + + ∂xk ∂x1 ∂x2 ∂x3 • curl ∂uk ∂xj

(

× u)i = εijk

which can be compared with the expression for the determinant. Example Show that

·(

× Ψ) = 0. ·( × Ψ) = ∂ ∂Ψm ∂2 εklm = εklm Ψm = 0 ∂xk ∂xl ∂xk ∂xl is symmetric in kl.

since εklm is antisymmetric in kl and

∂2 ∂xk ∂xl

y 1 n x y i j ﬁne grid (1) coarse grid (2)

n

S

V

dS = n dS Figure A.6: Volume V bounded by the close surface S. Example Show that

f (r) = 1 f (r) where {r}k = xk and r = |r| = r =

√

xk xk .

{ f (r)}k

∂ df ∂r ∂(xp xp )1/2 f (r) = =f = ∂xk dr ∂xk ∂xk 1 1 1 1 ∂ f (xp xp )− 2 (xp xp ) = f (δpk xp + xp δpk ) = f xk = 2 ∂xk 2r r

1 fr r

k

Example Show that ×(

×( × F)

× F) =

i

(

· F) − ∆F.

= εijk

**∂ ∂ ∂ ∂ ∂ ∂ εklm Fm = εijk εklm Fm = εkij εklm Fm = ∂xj ∂xl ∂xj ∂xl ∂xj ∂xl ( · F) − ∆F
**

i

(δil δjm − δim δjl ) Example

∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ Fm = Fj − Fi = Fj − Fi = ∂xj ∂xl ∂xj ∂xi ∂xj ∂xj ∂xi ∂xj ∂xj ∂xj ×u ∂ . This means, show that εijk uk = 2Ωi . 2 ∂xj

Show that if u = Ω × r then Ω = ¯ ¯ ¯ εijk

∂ ∂ ∂ uk = {uk = εklm Ωl rm } = εijk (εklm Ωl rm ) = εijk εklm Ωl rm = ∂xj ∂xj ∂xj ∂ ∂ Ωl rm = (δil δjm − δim δjl ) Ωl rm = ∂xj ∂xj ∂ri = δil } = ∂xl

εkij εklm

∂ ∂ ∂rm ∂ri ∂rm Ωi rm − Ωl ri = {Ω independent of x} = Ωi − Ωl ={ =1+1+1=3 ∂xm ∂xl ∂xm ∂xl ∂xm 3Ωi − Ωl δil = 2Ωi

A.2.7

Gauss & Stokes integral theorems

Let n be the outward pointing normal to the closed surface S bounding the simply-connected volume V , see Figure A.6. Gauss integral theorem then states a · n dS =

S V

· a dV ∂ak dV ∂xk

or ak nk dS =

S V

j ﬁne grid (1) coarse grid (2) S dl

n

C

Figure A.7: The open surface S enclosed by the contour C. for a diﬀerentiable vector ﬁeld a. This can be generalized to Tklm np dS =

S V

∂ Tklm dV . ∂xp

Example f n dS =

S V

f dV .

a × n dS =

S V

× a dV

if

Tklm nm = εklm al nm .

Stokes theorem states that for a diﬀerentiable vector ﬁeld a if an open simply-connected surface S is enclosed by a closed curve C, of which dl is the line element as in Figure A.7, then a · dl =

C S

(

× a) · n dS ∂am dS ∂xk

or ak dxk =

C S

εklm nk

**this can be generalized to Tklm dxp =
**

C S

εijp ni

∂ Tklm dS ∂xj

**Example Let Tklm = f. In the generalized Stokes theorem we then get
**

S

∂f εijp ni ∂xj dS in right hand side

f dl =

C S

n×

f dS

A.2.8

Archimedes principle

Gauss theorem can be used to calculate the buoyancy on an object immersed in a ﬂuid (Archimedes circa 281–212 BC). Consider a water cylinder with volume dV and mass dm, Figure A.8. The force from the water cylinder is dm g = dV ρg = x3 ρg dA

pressure

where g is the gravity constant and ρ the water density. The pressure at a distant x 3 from the surface in a ﬂuid is then p = ρgx3 + p0 where p0 is the atmospheric pressure. Let Π be the buoyancy force resulting from the pressure on the object, m the mass of equivalent volume of water and M the mass of the object in Figure A.8. The force on a small element dS of the body resulting from the pressure is dΠk = −gρx3 nk dS

i + 1 Γ1 p0 i+ 3 2 Ω 1 j updated : to be+ 2 j : new 1 values n j−2 : old values j−1i−1 dV dm A i Bi+1 Cj − 1 3 u 2 ,1 j v1, 3 dA 2 j + 1 p1,1 M x3 iteration direction 0 x 1 y 2 1 Figure A.8: An object with mass M immersed in a ﬂuid illustrating Archimedes principle. i n j mg x Inﬂow y Solid wall i n j ﬁne Γ0 (1) grid coarse Γ1 (2) grid Ω M2 Mg M m Figure A.9: The forces acting on an object immersed in a ﬂuid. m is the mass of equivalent volume of m+2 water and M the mass of the body. 1 h where the minus sign stems from the outward pointing normal n. The buoyancy force can then be calculated V through Vh ∂ x3 dV = −ρgδk3 V. Πk = − ρgx3u k dS = −ρg n ∂xk uh S V Γ0 This means that the object loses some weight as the water is displaced. The buoyancy force is Γ1 Ω Π3 = −ρgV = −g m : to be updated see Figure A.9. : new values The moment of a force F at some arbitrary point with position vector r from the point of application of : old values the force is M = r × F. The moment around the center of gravity G of the homogenous body produced by i−1 the pressure force dΠ acting on an element dS of the body is dMG = (r − rG ) × dΠ as in Figure A.10. i This can then be used to calculate the total moment around G form pressure forces on the body i+1 j−1 ∂x3 ∂ MGk = − ρgεklm (xl − xGl ) x3 nm dS = −ρgεklm (xl x3 ) − xGl dV = j ∂xm ∂xm S V j+1 iteration direction −ρgεklm [δlm x3 + δm3 xl − δm3 xGl ] dV = −ρgεkl3 xl dV − xGl V = 0 x y V V 1 If a part of the ﬂuid is frozen (without changing in density) this body is in equilibrium in the ﬂuid: n F = M = 0. x y i j ﬁne grid (1) G coarse grid (2) r − rG n x y i j ﬁne grid (1) coarse grid (2) n Figure A.10: Moment around G from pressure forces acting on the object.

i−1 i i+1 j−1 j j+1 iteration direction x y 1 n x y i j ﬁne grid (1) coarse grid (2)

z • P (r, θ, z)

z y r θ x Figure A.11: Cylindrical polar coordinates

A.3

Curvilinear coordinates

**Cylindrical Polar Coordinates
**

The cylindrical polar coordinates are (r, θ, z), where θ is the azimuthal angle, see ﬁgure A.11. The velocity can be written as u = u r er + uθ eθ + uz ez , (A.2) where the unit vectors are related to Cartesian coordinates as er = ex cos θ + ey sin θ, eθ = −ex sin θ + ey cos θ, ez = e z . Non-zero derivatives of unit vectors Gradient of a scalar p p= Laplacian of a scalar p

2

(A.3)

∂er = eθ , ∂θ

∂eθ = −er . ∂θ

(A.4)

∂p 1 ∂p ∂p e + e + e . ∂r r r ∂θ θ ∂z z + 1 ∂2p ∂2p + 2. r2 ∂θ2 ∂z

(A.5)

p=

1 ∂ ∂p r r ∂r ∂r

(A.6)

Divergence of a vector u ·u= Advective derivative of a scalar p (u · Curl of a vector u ×u = 1 ∂uz ∂uθ − er + r ∂θ ∂z ∂ur ∂uz 1 ∂(ruθ ) ∂ur − eθ + − ez . ∂z ∂r r ∂r ∂θ (A.9) )p = ur ∂p uθ ∂p ∂p + + uz . ∂r r ∂θ ∂z (A.8) 1 ∂(rur ) 1 ∂uθ ∂uz + + . r ∂r r ∂θ ∂z (A.7)

**Incompressible Navier-Stokes equations with no body force ∂ur + (u · ∂t ∂uθ + (u · ∂t )ur − )uθ + u2 1 ∂p θ =− +ν r ρ ∂r
**

2

ur −

ur 2 ∂u − 2 θ , r2 r ∂θ

(A.10) (A.11) (A.12)

ur uθ 1 ∂p 2 ∂u uθ 2 =− +ν uθ + 2 r − 2 , r ρ r ∂θ r ∂θ r ∂uz 1 ∂p + (u · )uz = − + ν 2 uz . ∂t ρ ∂z

i−1 i i+1 j −1 j j +1 ration direction x y 1 n x y i j ﬁne grid (1) coarse grid (2)

z • P (r, θ, ϕ) r

θ y

ϕ x Figure A.12: Spherical polar coordinates

**Spherical Polar Coordinates
**

The spherical polar coordinates are (r, θ, ϕ), where ϕ is the azimuthal angle, see ﬁgure A.12. The velocity can be written as u = u r er + uθ eθ + uϕ eϕ , (A.13) where the unit vectors are related to Cartesian coordinates as er = ex sin θ cos ϕ + ey sin θ sin ϕ + ez cos θ, eθ = ex cos θ cos ϕ + ey cos θ sin ϕ − ez sin θ, eϕ = −ex sin ϕ + ey cos ϕ. Non-zero derivatives of unit vectors ∂er = eθ , ∂θ ∂er = eϕ sin θ, ∂ϕ ∂eθ = −er , ∂θ ∂eθ = eϕ cos θ, ∂ϕ (A.15) (A.16) (A.14)

**∂eϕ = −er sin θ − eθ cos θ. ∂ϕ Gradient of a scalar p p= Laplacian of a scalar p
**

2

∂p 1 ∂p 1 ∂p e + e + e . ∂r r r ∂θ θ r sin θ ∂ϕ ϕ

(A.17)

p=

1 ∂ ∂p r2 2 ∂r r ∂r

+

1 ∂ ∂p sin θ 2 sin θ ∂θ r ∂θ

+

1 ∂2p . r2 sin2 θ ∂ϕ2

(A.18)

Divergence of a vector u ·u= Advective derivative of a scalar p (u · Curl of a vector u ×u= 1 r sin θ ∂(uϕ sin θ) ∂uθ ∂(ruϕ ) 1 1 ∂ur − er + − eθ ∂θ ∂ϕ r sin θ ∂ϕ ∂r + 1 ∂(ruθ ) ∂ur − eϕ . r ∂r ∂θ (A.21) )p = ur ∂p uθ ∂p uϕ ∂p + + . ∂r r ∂θ r sin θ ∂ϕ (A.20) 1 ∂(r2 ur ) 1 ∂(uθ sin θ) 1 ∂uϕ + + . r2 ∂r r sin θ ∂θ r sin θ ∂ϕ (A.19)

**Incompressible Navier-Stokes equations with no body force ∂ur + (u · ∂t =− 1 ∂p +ν ρ ∂r
**

2

)ur −

u2 + u 2 ϕ θ r

(A.22)

ur −

2ur 2 ∂(uθ sin θ) 2 ∂uϕ − 2 − 2 , r2 r sin θ ∂θ r sin θ ∂ϕ )uθ + (A.23)

∂uθ + (u · ∂t =− 1 ∂p +ν ρ r ∂θ

u2 cot θ ur uθ ϕ − r r 2 cos θ ∂uϕ 2 ∂u u 2 uθ + 2 r − 2 θ 2 − 2 2 , r ∂θ r sin θ r sin θ ∂ϕ )uϕ +

∂uϕ + (u · ∂t =− 1 ∂p +ν ρr sin θ ∂ϕ

ur uϕ u uϕ cot θ + θ r r uϕ 2 ∂ur 2 cos θ ∂uθ 2 uϕ + 2 + 2 2 − 2 2 . r sin θ ∂ϕ ∂ϕ r sin θ r sin θ

(A.24)

Appendix B

Recitations 5C1214

B.1

Scalar p=p Vector (¯)i = ui u u If u = v then u2 = v ¯ w a11 If A = a21 a31 1 if i = j 0 if i = j a12 a22 a32 a13 a23 then A23 = a23 a33

Tensors and invariants

Tensor Notation

Matrix (A)ij = Aij

Kronecker delta δij = (I)ij =

**Einstein summation convention
**

3

ui ui =

i=1

ui ui

**Kinetic energy per unit volume
**

1 u2 2 ρ|¯| 1 1 = 2 ρ(u2 + v 2 + w2 ) = 2 ρui ui

Matrix operations a · ¯ = a1 b1 + a2 b2 + a3 b3 = ai bi = δij ai bj = aj bj ¯ b (A¯ i = Aij bj b) (AB)ij = Aik Bkj tr(A) = Aii (A)ij = Aij ⇐⇒ (AT )ij = Aji Permutation symbol 109

εijk

1 if ijk in cyclic order. ijk = 123, 231 or 312 = 0 if any two indices are equal −1 if ijk in anticyclic order. ijk = 321, 213 or 132 e1 ¯ a1 b1 e2 ¯ a2 b2 e3 ¯ a3 b3 = e1 (a2 b3 − a3 b2 ) − e2 (a1 b3 − a3 b1 ) + e3 (a1 b2 − a2 b1 ) = εijk ei aj bk ¯ ¯ ¯ ¯

a×¯= ¯ b

(¯ × ¯ i = εijk aj bk a b) εijk εilm = δjl δkm − δjm δkl εijk εijm = {l = j} = 3δkm − δjm δkj = 3δkm − δmk = 2δkm εijk εijk = {m = k} = 2 · 3 = 6 Rewrite without the cross product: (¯ × ¯ · (¯ × d) = (¯ × ¯ i (¯ × d)i = εijk aj bk εilm cl dm = εijk εilm aj bk cl dm = a b) c ¯ a b) c ¯ (δjl δkm − δjm δkl ) aj bk cl dm = aj cj bk dk − aj dj bk ck = (¯ · c)(¯ · d) − (¯ · d)(¯ · c) a ¯ b ¯ a ¯ b ¯

**Invariant parts of Tensors
**

S A Tij = Tij + Tij S Tij = 1 2 1 2

**Symmetric and Anti-symmetric parts.
**

S = Tji

Tij + Tji Tij − Tji

Symmetric Anti-symmetric

A Tij =

A = −Tji

**The symmetric part of the tensor can be divided into two parts: ¯ ¯ TS = T + T
**

ij ij ij 1 S ¯ Tij = Tij − 3 Tkk δij

trace less

¯ Tij = 1 Tkk δij 3 This gives:

isotropic

S A A ¯ ¯ Tij = Tij + Tij = Tij + Tij + Tij

∂ui . The anti-symmetric part describes rotation, the isotropic ∂xj part describes the volume change and the trace-less part describes the deformation of a ﬂuid element. In the NS equations we have the tensor

Operators

( p)i = ∆p = ∂ p ∂xi

∂2 p ∂xi ∂xi ∂ ui ∂xi ∂ uk ∂xj

·u= ¯ (

× u)i = εijk ¯

**Gauss theorem (general)
**

Gauss theorem:

S

¯ ¯ F · n dS =

V

¯ · F dV ∂Fi dV ∂xi

or, Fi ni dS =

S V

**In general we can write, Tijk nl dS =
**

S V

∂ Tijk dV ∂xl

**Example: Put Tijk nl = Tij ul nl Tij ul nl dS =
**

S V

∂ (ul Tij ) dV ∂xl

or,

S

T(¯ · n) dS = u ¯

(

V

· u)T + (¯ · ¯ u

)T dV

Identities

Derive the identity ¯ ¯ ¯ × (F × G) = (G · ¯ ¯ × (F × G)

i

¯ )F + (

¯ ¯ · G)F − (

¯ ¯ ¯ · F )G − ( F ·

¯ )G

= εijk

∂ ∂ ∂ εklm Fl Gm = εijk εklm Fl Gm = εkij εklm Fl G m = ∂xj ∂xj ∂xj

(δil δjm − δim δjl ) Gj

∂ ∂ ∂ ∂Fi ∂Gj ∂Fj ∂Gi Fl G m = Fi G j − Fj G i = Gj + Fi − Gi + Fj = ∂xj ∂xj ∂xj ∂xj ∂xj ∂xj ∂xj ¯ )F + ( ¯ ¯ · G)F − ( ¯ ¯ ¯ · F )G − ( F · ¯ )G]i

∂Fi ∂Gj ∂Fj ∂Gi ¯ + Fi − Gi − F j = [(G · ∂xj ∂xj ∂xj ∂xj

Show that: ·( ·( ∂ ∂ ∂ ∂ ¯ εijk Fk = εijk Fk × F) = ∂xi ∂xj ∂xi ∂xj ∂ ∂ ∂ ∂ Fk = Fk and thus all terms will cancel. ∂xi ∂xj ∂xj ∂xi ¯ × F) = 0

Remember that εijk = −εjik and that εijk ∂ ∂ Fk = 0 ∂xi ∂xj

Example

¯ Show that if u = Ω × r then Ω = ¯ ¯ ¯ εijk ×u ¯ ∂ . This means, show that εijk uk = 2Ωi . 2 ∂xj

∂ ∂ ∂ uk = {uk = εklm Ωl rm } = εijk (εklm Ωl rm ) = εijk εklm Ωl rm = ∂xj ∂xj ∂xj ∂ ∂ Ωl rm = (δil δjm − δim δjl ) Ωl rm = ∂xj ∂xj ∂ri = δil } = ∂xl

εkij εklm

∂ ∂ ∂rm ∂ri ∂rm Ωi rm − Ωl ri = {Ω independent of x} = Ωi − Ωl ={ =1+1+1=3 ∂xm ∂xl ∂xm ∂xl ∂xm 3Ωi − Ωl δil = 2Ωi Derive the identity ¯ ¯ · (F × G) = ( ¯ ¯ ¯ × F) · G − F · ( ¯ × G)

¯ ¯ · (F × G) = εkij

∂ ∂ ∂Fj ∂Gk ∂Fj ∂Gk εijk Fj Gk = εijk Fj Gk = εijk Gk + εijk Fj = εkij Gk + Fj εijk = ∂xi ∂xi ∂xi ∂xi ∂xi ∂xi ¯ ¯ ¯ × F) · G − F · ( ¯ × F) = ¯ × G) ¯ ¯ · F ) − ∆F

∂Fj ∂Gk Gk − Fj εjik =( ∂xi ∂xi

Show that: ×( ¯ × F)

(

×(

i

= εijk

**∂ ∂ ∂ ∂ ∂ ∂ εklm Fm = εijk εklm Fm = εkij εklm Fm = ∂xj ∂xl ∂xj ∂xl ∂xj ∂xl ( ¯ ¯ · F ) − ∆F
**

i

(δil δjm − δim δjl )

∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ Fm = Fj − Fi = Fj − Fi = ∂xj ∂xl ∂xj ∂xi ∂xj ∂xj ∂xi ∂xj ∂xj ∂xj

B.2

Euler and Lagrange coordinates

**Particle path (Lagrange’s representation)
**

Following a ﬂuid particle as time proceeds the path is described by ∂¯ r = u, ¯ ˆ ∂t r (t = 0) = x0 ¯ˆ ¯

**Streamline (Euler’s representation)
**

Instantaneously a ﬂuid particles displacement d¯ is tangential to its velocity u: x ¯ d¯ x u ¯ = |d¯| x |¯| u or if we put ds = |d¯|/|¯|: x u d¯ x =u ¯ ds

Exercise 1

Show that the streamlines for the unsteady ﬂow u = u0 , v = kt, w = 0, u0 , k > 0

is straight lines, while any ﬂuid particle follows a parabolic path as time proceeds. Streamline (ﬁx t): dx = u = u0 ⇒ x = u0 s + a ds dy = u = kt ⇒ y = kts + b ds Then we see that y(x) is a linear function, y= Particle path (ﬁx x0 ): ¯ kt x + d(t) u0 ˆ x = u 0 t + x0 x= 1 ˆ2 k t + y0 2

∂x = u = u0 ˆ ∂t ∂y ˆ = u = kt ˆ ∂t

⇒ ⇒

Then we see that y(x) is a parabola, y= 1 k 2 (x − 2x0 x + x2 ) + y0 0 2 u2 0

Exercise 2

a) Separate the shear ﬂow u = (βy, 0, 0) into its (i) local translation, (ii) rotation and (iii) pure straining ¯ parts.

**Lets denote the diﬀerent parts as u = u T + uR + uD ¯ ¯ ¯ ¯ In tensor notation it can be written as ui 2 = ui 1 + ξij dxj + eij dxj
**

P P

(i) Translation is simply uT = {ui 1 } = (βy, 0, 0) ¯

P

1 j−2 (ii) j−1 A B C 3 u 2 ,1 v1, 3 2 p1,1

0 β ∂ui = 0 0 ∂xj 0 0

0 0 0

ξij =

1 ∂ui ∂uj − 2 ∂xj ∂xi

uR = ξd¯ = ¯ x

1 (βdy, −βdx, 0) 2

0 1 = −β 2 0

β 0 0

0 0 0

1 (iii) 2 0 β 0 1 ∂ui ∂uj 1 i + = β 0 0 eij = j 2 ∂xj ∂xi 2 0 0 0 Inﬂow 1 Solid wall uD = ed¯ = (βdy, βdx, 0) ¯ x n 2 Γ0 Γ1 Ω M 2 b) Find the principal axes of the rate of strain tensor eij and make a schematic sketch of the decomposition M of the ﬂow. 1 ∂ui ∂uj ∂ui m + symmetric part of . eij = 2 ∂xj ∂xi ∂xj m+2 1 0 β 0 1 h eij = β 0 0 2 V 0 0 0 Vh uEigenvalues: uh −λ β/2 0 Γ0 0 = −λ3 +β 2 λ/4 = −λ(λ−β/2)(λ+β/2) = 0. ⇒ λ1 = β/2, λ2 = −β/2, λ3 = 0. det(eij −λδij ) = β/2 −λ Γ1 0 0 −λ Ω : to be updatedWe get the eigenvectors from eij kj = λki : : new values 1 1 ¯ ¯ ¯ : old values k1 = √ (1, 1, 0), k2 = √ (−1, 1, 0), k3 = (0, 0, 1) 2 2 i−1 iNow sketch the decomposition of the ﬂow i+1 j−1 1 1 (βdy, 0, 0) = (βdy, −βdx, 0) + (βdy, βdx, 0) j 2 2 j+1 iteration direction x 1 1 1 y n x y i j ﬁne grid (1) coarse grid (2)

0.5 0.5 = 0.5 +

0

−0.5 −1 −1

0

−0.5 −1 −1

0

−0.5 −1 −1

0

1

0

1

0

1

1 Exercise 3 2 ia) Sketch the streamlines for the ﬂow [u, v, w] = [αx, −αy, 0], where α > 0. j Streamlines in parametric form (x(s), y(s), z(s)). We have a 2D ﬂow since w = 0. Inﬂow dx dy Solid wall = αx, = −αy. ds ds n Γ0 Integrate w.r.t. s, Γ1 x(s) = a eαs , y(s) = b e−αs . Ω M 2 Then we see that, x(s) y(s) = ab = c ( constant ). M mWe have, c c m+2 x(s) = , y(s) = 1 y(s) x(s) h Streamlines: V Vh dx dy > 0, <0 x > 0, y > 0 ⇒ u ds ds uh dx dy Γ0 x < 0, y > 0 ⇒ < 0, <0 ds ds Γ1 dx dy Ω x < 0, y < 0 ⇒ < 0, >0 ds ds to be updated dx dy : new values x > 0, y < 0 ⇒ > 0, >0 : old values ds ds i−1 dx(s) dy(s) = αx and = −αy. This gives the direction of the ﬂow: iRemember that ds ds i+1 j −1 1 j j +1 ration direction 0.5 x y 0 n x y −0.5 i j ﬁne grid (1) −1 coarse grid (2) −1

−0.8

−0.6

−0.4

−0.2

0

0.2

0.4

0.6

0.8

1

b) The concentration of a scalar is c(x, y, t) = βx2 ye−αt . Does this concentration change for a particular ﬂuid particle in time? Use the material time-derivative ∂c ∂c ∂c Dc = +u +v = −αβx2 ye−αt + αx2βxye−αt − αyβx2 e−αt = (−1 + 2 − 1)βx2 ye−αt = 0. Dt ∂t ∂x ∂y Thus the concentration c(x, y, t) is constant for a ﬂuid particle.

c) Using the alternative (Lagrangian) description of the ﬂow, show that ∂u ¯ D¯ u = ˆ Dt ∂t ˆ and write the concentration as c = c(x0 , y0 , t).

Particle path:

∂x ˆ = u = αx ⇒ x = x0 eαt ˆ ∂t ∂y ˆ = v = −αy ⇒ y = y0 e−αt ˆ ∂t ∂u ∂ ∂x ˆ = = α 2 x0 e α t = α 2 x ˆ ˆ ∂t ˆ ∂t ∂t ∂v ∂ ∂y ˆ = = α2 y0 e−αt = α2 y ˆ ∂t ∂t ˆ ˆ ∂t Du ∂u ∂u ∂u = +u +v = α2 x Dt ∂t ∂x ∂y Dv ∂v ∂v ∂v = +u +v = α2 y Dt ∂t ∂x ∂y

∂u ¯ D¯ u = ˆ Dt ∂t Both terms describe the acceleration of a ﬂuid particle. Now rewrite the concentration ¯ c(x, y) = βx2 ye−αt = β x2 e2αt y0 e−αt e−αt = βx2 y0 = c(X) 0 0

x2 y ˆ ˆ ˆ ˆ

So we have

ˆ independent of t!

The concentration foes not change as we follow a ﬂuid particle.

B.3

Reynolds transport theorem and stress tensor

Exercise 1

a) Consider a ﬁxed closed surface S in a ﬂuid. Show that conservation of mass implies ∂ρ ∂ + (ρui ) = 0. ∂t ∂xi The change of mass in the volume is described by the contribution from the mass ﬂux and the change in density and conservation means that this should be zero ∂ρ dV + ∂t ρui ni dS = 0

S

V

**Using Gauss Theorem we can rewrite the surface integral ρui ni dS =
**

S V

∂ (ρui ) dV. ∂xi

We have

V

{

∂ρ ∂ + (ρui )} dV = 0. ∂t ∂xi

This must be true for an arbitrary volume and this implies ∂ ∂ρ + (ρui ) = 0. ∂t ∂xi b) Show that this can be written Dρ ∂ui +ρ = 0, Dt ∂xi ∂ρ ∂ui ∂ρ ∂ui ∂ (ρui ) = ui + ρ = ui +ρ ∂xi ∂xi ∂xi ∂xi ∂xi This gives ∂ρ ∂ρ Dρ ∂ui ∂ui + ui +ρ = +ρ ∂t ∂xi ∂xi Dt ∂xi Dρ Dt c) Explain what the following means ∂ui =0 ∂xi ⇒ Dρ =0 Dt

If we are following a ﬂuid particle, the density and volume are constant. But in a ﬁx position, the density can change ∂ρ ∂ρ = −ui ∂t ∂xi Remember the decomposition of

∂ui ∂xj

**in its invariant parts − 1 ∂uk 1 ∂ui ∂uj δij + − 3 ∂xk 2 ∂xj ∂xi
**

rotation ξij

∂ui 1 ∂ui ∂uj = + ∂xj 2 ∂xj ∂xi .

+

**1 ∂uk δij 3 ∂xk
**

¯ volume change eij

deformation eij ¯

Exercise 2

a) Use Reynolds transport theorem to provide an alternative derivation of the conservation of mass equation ∂ρ ∂ + (ρui ) = 0. ∂t ∂xi Reynolds Transport Theorem: D Dt Put Tijk = ρ, D Dt ρ dV =

V (t) V (t)

V (t)

Tijk dV =

V (t)

∂Tijk ∂ + (u T ) dV ∂t ∂xl l ijk ∂ρ ∂ + (ρui ) dV ∂t ∂xi

No ﬂow through the surface. Valid for all V (t): ∂ρ ∂ + (ρui ) = 0 ∂t ∂xi b) Use this result to show D Dt

V (t)

ρFijk dV =

ρ

V (t)

DFijk dV Dt

Put Tijk = ρFijk in Reynolds transport theorem D Dt ρFijk dV = ∂ ∂ (ρFijk ) + (u ρF ) dV = ∂t ∂xl l ijk dV = dV =

V (t)

V (t)

ρ

V (t)

**∂Fijk ∂Fijk ∂ρ ∂ + Fijk + Fijk (ρul ) + ρul ∂t ∂t ∂xl ∂xl ∂Fijk ∂Fijk ∂ρ ∂ + (ρul ) + ρ + ul ∂t ∂xl ∂t ∂xl
**

=0 =

DF ijk Dt

V (t)

Fijk

ρ

V (t)

DFijk dV Dt

Exercise 3

Show that the stress tensor Tij = −pδij for the ﬂow u = ω × x, ¯ ¯ ¯ where ω is constant. ¯ Tensor notation: ui = εikl ωk xl The stress tensor: Tij = −pδij + µ −pδij + µ ∂uj ∂ui + ∂xj ∂xi = =

∂ ∂ (εikl ωk xl ) + (ε ω x ) ∂xj ∂xi jkl k l ∂xl ∂x + εjkl l ∂xj ∂xi =

−pδij + µωk εikl

−pδij + µωk (εikl δlj + εjkl δli ) = −pδij + µωk (εikj + εjki ) = −pδij + µωk (−εijk + εijk ) = −pδij

**The stress tensor Tij
**

The equation for the stress tensor was deduced by Stokes in (1845) from three elementary hypotheses. Tij = −pδij + 2 µeij Writing this on the form, Tij = −pδij + τij the following statements should be true for the viscous stress tensor τij in a Newtonian ﬂuid. ∂ui (i) each τij should be a linear combination of the velocity gradients . ∂xj (ii) each τij should vanish if the ﬂow involves no deformation of ﬂuid elements. ∂ui (iii) the relationship between τij and should be isotropic as the physical properties of the ﬂuid ∂xj are assumed to show no preferred direction.

Exercise 4

If τij is a linear function of the components of eij then it can be written τij = cijkl ekl It can be shown that the most general fourth order isotropic tensor is of the form cijkl = Aδij δkl + Bδik δjl + Cδil δjk where A, B and C are scalars. a) Use this to show that τij = λekk δij + 2µeij where λ and µ are scalars. τij = cijkl ekl = (Aδij δkl + Bδik δjl + Cδil δjk )ekl = Aδij ekk + Beij + Ceji Say that A = λ and that B = C = µ, then since eij = eji we have τij = λekk δij + 2µeij 1 b) Show that if we have a Newtonian ﬂuid where p = − Tii then 3 ∂ui ∂uj 2 ∂uk Tij = − p + µ δij + µ + 3 ∂xk ∂xj ∂xi We have Tij = −pδij + τij = −pδij + λekk δij + 2µeij Then This gives λ = − 2 µ, 3 Tii = −3p + 3λeii + 2µeii = −3p + (3λ + 2µ)eii and thus 2 Tij = −p − µekk δij + 2µeij 3 2 ∂uk ∂ui ∂uj Tij = − p + µ δij + µ + 3 ∂xk ∂xj ∂xi

or

h V B.4 Vh

Rankine vortex and dimensionless form

The Rankine vortex

n x y i j ﬁne grid (1) coarse grid (2)

0

0

1

r

2

3

0 0 1

ω≈0 1

uh A simple model for a vortex is that it is a rigid body rotation within a core, and a decay of angular velocity Γ0 outside. This can be described by Γ1 Ω ωr, r < a, : to be updated uθ = ωa2 ur = u z = 0 , r > a, : new values r : old values i − 1and is called a Rankine vortex. i i+1 3 j−1 j 2 j+1 2 iteration direction x y 1 u

θ

r

2

3

Figure B.1: Velocity and vorticity in a Rankine vortex with ω = a = 1.

Exercise 1

a) Find the pressure inside and outside of a Rankine vortex: We use Eulers equations for incompressible ﬂow ( neglecting viscous eﬀects ): u D¯ = − 1 p + g ¯ Dt ρ Eulers equations ·u=0 ¯ ∂u ¯ D¯ u = +(¯ · u Dt ∂t

=0

)¯ u

We are working with cylindrical coordinates, use the appendix formulas u· ¯ (¯ · u )¯ = u = ur ∂ uθ ∂ ∂ + + uz ∂r r ∂θ ∂z

**uθ ∂ uθ ∂uθ uθ ∂¯θ e uθ ∂uθ u2 (uθ eθ ) = ¯ eθ + ¯ uθ = eθ − θ er ¯ ¯ r ∂θ r ∂θ r ∂θ r ∂θ r
**

=−¯r e

p= Insert this into the Euler equations:

∂p 1 ∂p ∂p er + ¯ eθ + ¯ ez ¯ ∂r r ∂θ ∂z

uθ ∂uθ u2 1 ∂p 1 ∂p ∂p eθ − θ er = − ¯ ¯ er + ¯ eθ + ¯ ez ¯ r ∂θ r ρ ∂r r ∂θ ∂z

=0

− g¯z e

Look at the diﬀerent components: er : ¯ eθ : ¯ 0=− −

u2 1 ∂p θ =− r ρ ∂r

1 1 ∂p ⇒ p = p(r, z) only. ρ r ∂θ 0=− 1 ∂p −g ρ ∂z

ez : ¯ Solve for the pressure when r < a: er : ¯ −ω 2 r = −

1 ∂p ρ ∂r ⇒

⇒

p = ρω 2

r2 + f (z) 2

ez : ¯ So we have the pressure:

∂p = −ρg ∂z

f (z) = −ρgz + C1 r<a

p(r, z) = ρω 2 Solve for the pressure when r > a: er : ¯ −

r2 − ρgz + C1 2

ω 2 a4 1 ∂p =− r3 ρ ∂r ∂p = −ρg ∂z ⇒

⇒

p=−

ρω 2 a4 + f (z) 2 r2

ez : ¯ So we have the pressure:

f (z) = −ρgz + C2

ρω 2 a4 − ρgz + C2 r > a 2 r2 Now determine the diﬀerence between the constants by evaluation at r = a p(r, z) = − ρω 2 a2 ρω 2 a2 − ρgz + C1 = − − ρgz + C2 2 2 ⇒ C2 − C1 = ρω 2 a2

b) Determine the pressure diﬀerence ∆p between r = 0 and r → ∞ ∆p = p∞ − p0 = −ρgz + C2 − (−ρgz + C1 ) = C2 − C1 = ρω 2 a2 c) Now consider a free surface at atmospheric pressure p0 . Find the diﬀerence in z between r = 0 and r → ∞ r=0: r→∞: p0 =

ρω 2 a2 − ρgz0 + C1 2 ρω 2 a2 p0 = − 2 − ρgz∞ +

C2

⇒

z∞ − z0 =

C2 − C 1 ω 2 a2 = ρg g

**Determine the shape of the free surface: p0 = p0 =
**

ρω 2 r 2 − ρgz + C1 2 ρω 2 a4 − 2r2 − ρgz + C2

r < a z ∼ r2 ⇒ z = r>a z∼

1 r

⇒z

C1 −p0 ω2 r2 2g + ρg 2 4 −p = − ω a2 + C2ρg 0 2gr

Say that z = 0 at r = 0 then C1 = p0 and we get z= and z=− So we have z(r) = ω 2 r2 2g r<a

**ω 2 a4 C2 − C 1 ω 2 a4 ω 2 a2 ω 2 a2 a2 + =− + = 1− 2 2 2 2gr ρg 2gr g g 2r ω2 r2 2g ω ga r<a 1−
**

a2 2r 2

r>a

2 2

r > a.

i i+1 j−1 j j+1 iteration direction x y n x y i j ﬁne grid (1) coarse grid (2)

0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01

z

0 0

0.5

1

1.5 r

2

2.5

3

Figure B.2: The free surface of a Rankine vortex with ω = a = 1 and g = 9.82. ω2 r2 2g ω ga

r<a 1−

a2 2r 2

z(r) =

2 2

r > a.

Exercise 2

∂u ¯ + (¯ · u ∂t with the boundary conditions

**Solving the 2D ﬂow around a cylinder involves solving )¯ = − u 1 p+ν ρ
**

2

u, ¯

· u = 0, ¯

u = 0 on x2 + y 2 = a2 ¯

u → (U, 0, 0) as x2 + y 2 → ∞. ¯

Rewrite this problem in dimensionless form using the dimensionless variables x = x/a u = u/U ¯ ¯ ¯ ¯ Note that the scaling x = x/a implies ¯ ¯ Change to dimensionless variables =a p = p/ρU 2 t = tU/a. ∂ a ∂ = . ∂t U ∂t νU a2

2

and t = tU/a gives

U 2 ∂u ¯ U2 + (¯ · u a ∂t a Divide by U 2 /a ∂u ¯ + (¯ · u ∂t

)¯ = − u

ρU 2 ρa

p +

u ¯

)¯ = − u

p +

νU a2 U2 a

1 Re

2

u ¯

**U2 Inertial forces Ua The Reynolds number is Re = a = = νU Viscous forces ν a2 ∂u ¯ + (¯ · u ∂t The continuity condition ·u=0 ¯ ⇒ U a ·u =0 ¯ ⇒ ·u =0 ¯ )¯ = − u p + 1 Re
**

2

u ¯

**The boundary conditions u = 0 on x2 + y 2 = a2 ⇒ ¯ U u = 0 on a2 x + a2 y ¯
**

2 2

= a2 ⇒

u = 0 on x + y ¯

2 2

2

2

= 1.

u → (U, 0, 0) as x2 + y 2 → ∞ ⇒ ¯

U u → (U, 0, 0) as a2 x + a2 y ¯

2 2

→∞⇒

u → (1, 0, 0) as x + y ¯

→∞

The solutions of this problem will depend on x , Re and t only, and thus the solution of this problem is ¯ the same for a speciﬁc Re independently of the individual values of U , a and ν.

Exercise 3

Show that the net viscous force per unit volume is proportional to the spatial derivative of vorticity, i.e. ∂τij ∂ωk = −µεijk ∂xj ∂xj and discuss its implication for ﬂows with uniform vorticity (as in solid-body rotation). ∂τij ∂ =µ ∂xj ∂xj −µεijk ∂ui ∂uj + ∂xj ∂xi =µ ∂ 2 ui ∂ 2 uj + ∂xj ∂xj ∂xj ∂xi =µ ∂ 2 ui ∂xj ∂xj

∂ωk ∂ ∂um = −µεijk εklm ∂xj ∂xj ∂xl −µ

= −µεkij εklm

∂ 2 um ∂ 2 um = −µ(δil δjm − δim δjl ) = ∂xj xl ∂xj xl ∂ 2 ui ∂xj ∂xj

∂ 2 ui ∂ 2 uj − ∂xj ∂xi ∂xj ∂xj

=µ

Thus

∂τij ∂ωk = −µεijk ∂xj ∂xj

The net viscous force vanishes when the vorticity is uniform, since no deformation exists.

j+

2

j− j − 1Exercise 1 APlane Couette ﬂow: BConsider the ﬂow of a viscous ﬂuid between two parallel plates at y = 0 and y = h. The upper plane is Cmoving with velocity U . Calculate the ﬂow ﬁeld. 3 u 2 ,1 v1, 3 2 Assume the following: p1,1 Steady ﬂow: ∂ 0 =0 1 ∂t 2Parallel ﬂow: ∂ui i =0 v = 0, j ∂x InﬂowTwo-dimensional ﬂow: ∂ Solid wall w = 0, =0 ∂z n Γ0 No pressure gradient: ∂p Γ1 =0 ∂xi Ω M 2 Navier–Stokes streamwise momentum equation: M ∂u 1 m + (¯ · )u = − u p + ν 2u ∂t ρ m+2 1We have h ∂2u ∂u =0 ⇒ = A ⇒ u = Ay + B V ∂y 2 ∂y Vh Boundary conditions: u u(0) = 0 ⇒ B = 0, u(h) = U ⇒ A = U/h uh Γ0 So we get: Uy Γ1 u(y) = h Ω : to be updated : new valuesExercise 2 (i) : old valuesPlane Poiseuille ﬂow: ( Channel ﬂow ) i − 1Consider the ﬂow of a viscous ﬂuid between two solid boundaries at y = ±h driven by a constant pressure igradient p = [−P, 0, 0]. Show that i+1 P 2 j−1 u= (h − y 2 ), v = w = 0. j 2µ j+1 iteration direction Y x y 1 n x X y i j Z ﬁne grid (1) coarse grid (2) Figure B.3: Coordinate system for plane Poiseuille ﬂow Navier–Stokes equations: ¯ ∂ u + (¯ · )¯ = − 1 p + ν u u ∂t ρ · u = 0. ¯

1 2

j B.5

Exact solutions to Navier-Stokes

2

u ¯

j+

2 1 2

u(y = ±h) = 0 ¯ j− j −1 ∂u ¯ • We are considering a stationary ﬂow and thus = 0. A ∂t • The constant pressure gradient implies u = u(y). Changes of u in x, z would require a changing pressure ¯ ¯ ¯ B gradient in x, z. C 3 ∂v u 2 ,1 • The continuity equation · u reduces to ¯ = 0. The boundary condition v(y = ±h) = 0 then implies v1, 3 ∂y 2 p1,1 v = 0. Let’s study the z component of the Navier–Stokes equations: 0 1 ∂w ∂2w v =ν 2 ⇒ w = c1 y + c2 2 ∂y ∂y i =0 j InﬂowThe boundary conditions w(y = −h) = w(y = h) = 0 imply c1 = c2 = 0 and thus w = 0. Solid wall ¯ n• We can conclude that u = [u(y), 0, 0]. Γ0 Γ1 Let’s study the x component of the Navier–Stokes equations: Ω P ∂2u ∂u P P 2 M2 0= +ν 2 ⇒ = − y + d1 {µ = ρν} ⇒ u(y) = − y + d1 y + d 2 M ρ ∂y ∂y νρ 2µ m m + 2The boundary conditions at y = ±h give 1 P 2 P 0=− h + d1 h + d2 and 0 = − h2 − d1 h + d2 h 2µ 2µ V Vh P 2 uWe can directly conclude that d1 = 0 and this gives d2 = 2 µ h and thus uh Γ0 P 2 u= (h − y 2 ), v = w = 0. Γ1 2µ Ω to be updated Exercise 2 (ii) : new values : old valuesPoiseuille ﬂow: ( Pipe ﬂow ) i − 1Consider the viscous ﬂow of a ﬂuid down a pipe with a cross-section given by r = a under the constant ∂p i pressure gradient P = − . Show that i+1 ∂z j −1 P 2 j uz = (a − r2 ) ur = uθ = 0. 4µ j +1

R

j

Boundary conditions:

ration direction x y 1 n x y i j ﬁne grid (1) coarse grid (2)

Z

**Figure B.4: Coordinate system for Poiseuille ﬂow Use the Navier–Stokes equations in cylindrical coordinates ∂ur + (¯ · u ∂t )ur − u2 1 ∂p θ =− +ν r ρ ∂r
**

2

ur −

ur 2 ∂uθ − 2 2 r r ∂θ

∂uθ + (¯ · u ∂t

)uθ +

ur uθ 1 ∂p =− +ν r ρ r ∂θ )uz = −

2

uθ +

2

2 ∂ur uθ − 2 r2 ∂θ r

∂uz + (¯ · u ∂t

1 ∂p +ν ρ ∂z

uz

1 ∂ 1 ∂uθ ∂uz (r ur ) + + = 0. r ∂r r ∂θ ∂z ∂p ∂p We know that = 0 and = 0 and can directly see that ur = uθ = 0 satisfy the two ﬁrst equations. ∂θ ∂r From the continuity equation we get ∂uz =0 ∂z ⇒ uz = uz (r, θ) only.

**Considering a steady ﬂow we get from the Navier–Stokes equations (¯ · u But we know that )uz = uz ∂uz ∂z ⇒ uz ∂uz 1 = P +ν ∂z ρ
**

2

uz

**∂uz = 0 from the continuity equation. We get ∂z
**

2

uz =

1 ∂ ∂uz P (r )=− r ∂r ∂r µ

⇒

∂ ∂uz P (r )=− r ∂r ∂r µ

Integrate r Integrating again we get uz = −

∂uz P = − r 2 + c1 ∂r 2µ

⇒

∂uz P c1 =− r+ ∂r 2µ r

P 2 r + c1 ln(r) + c2 4µ

using the boundary conditions uz (r = 0) < ∞ ⇒ c1 = 0 P a2 and we have 4µ P 2 (a − r2 ) 4µ

We also have uz = 0 at r = a and this gives c2 = uz =

Exercise 3

Calculate the asymptotic suction boundary layer, where the boundary layer over a ﬂat plate is kept parallel of a steady suction V0 through the plate. Assumptions. Two-dimensional ﬂow:

∂ = 0, ∂z ∂ = 0, ∂x

w=0

Parallel ﬂow:

Continuity satisﬁed ∂ =0 ∂t

Steady ﬂow:

Momentum equations:

∂u ∂u ∂u 1 ∂p ∂2u ∂2u +u +v =− +ν + 2 ∂t ∂x ∂y ρ ∂x ∂x2 ∂y ∂v ∂v ∂v 1 ∂p ∂2v ∂2v +u +v =− +ν + 2 ∂t ∂x ∂y ρ ∂y ∂x2 ∂y

Normal momentum equation gives

∂p =0 ∂y

Boundary conditions: y=0: u = 0, v = −V0 y→∞: Continuity gives ∂u ∂v + =0 ∂x ∂y Streamwise momentum equation at y → ∞ −V0 ∂U∞ 1 ∂p ∂ 2 U∞ =− +ν ∂y ρ ∂x ∂y 2 ⇒ Resulting streamwise momentum equation −V0 Characteristic equation λ2 = − V0 λ ν ⇒ λ1 = 0, λ2 = − V0 ν ∂u ∂2u =ν 2 ∂y ∂y ⇒ ∂2u V ∂u =− 0 ∂y 2 ν ∂y ∂p =0 ∂x ⇒ v = −V0 u → U∞

u(y) = A + Be−V0 y/ν

Exercise 4

Two incompressible viscous ﬂuids ﬂow one on top of the other down an inclined plane at an angle α. They both have the same density ρ and the viscosities µ1 and µ2 . The lower ﬂuid has depth h1 and the upper h2 . Assuming that viscous forces from the surrounding air is negligible and that the pressure on the free surface is constant, show that 1 g sin(α) u1 (y) = [(h1 + h2 )y − y 2 ] . 2 ν1 Make the ansatz u1 = [u1 (y), 0, 0] and u2 = [u2 (y), 0, 0]. The continuity equation ¯ ¯ ∂u ∂v + =0 ∂x ∂y • Layer 1: N–S · ey : 0 = − ¯ gives ∂v = 0 ⇒ v = c and the boundary condition at y = 0 gives v = 0. ∂y 1 ∂p1 − g cos(α) ρ ∂y

⇒

p1 = −ρg cos(α)y + f1 (x) ⇒ f1 (x) = c1

1 d 2 u1 N–S · ex : 0 = − f1 (x) + ν1 2 + g sin(α) ¯ ρ dy • Layer 2: N–S · ey : 0 = − ¯ 1 ∂p2 − g cos(α) ρ ∂y ⇒

p2 = −ρg cos(α)y + f2 (x) ⇒ f2 (x) = c2

1 d 2 u2 N–S · ex : 0 = − f2 (x) + ν2 2 + g sin(α) ¯ ρ dy The pressure at the free surface y = h1 + h2 is p0 : p0 = −ρg cos(α)(h1 + h2 ) + f2 (x) The pressure is continuous at y = h1 : p0 + ρgh2 cos(α) = −ρgh1 cos(α) + f1 (x) This gives the pressure: ⇒ ⇒

f2 = p0 + ρg(h1 + h2 ) cos(α)

⇒

f2 = 0

f1 = p0 + ρg(h1 + h2 ) cos(α)

⇒

f1 = 0

p1 (y) = p2 (y) = p(y) = −ρ g cos(α)y + p0 + ρ g cos(α)(h1 + h2 )

We now have two momentum equations in x: 0 = ν1 0 = ν2 And four boundary conditions: BC1: No slip on the plane: BC2: No viscous forces on the free surface: BC3: Force balance at the ﬂuid interface: µ1 u1 (0) = 0 µ2 du2 |y=h1 +h2 = 0 dy d 2 u1 + g sin(α) dy 2 d 2 u2 + g sin(α) dy 2 (1) (2)

du1 du2 |y=h1 = µ2 |y=h1 dy dy g 2 y sin(α) + c11 y + c12 2 ν1 g 2 y sin(α) + c21 y + c22 2 ν2

BC4: Continous velocity at the interface: u1 |y=h1 = u2 |y=h1 (1) ⇒ (2) ⇒ du1 g = − y sin(α) + c11 dy ν1 ⇒ ⇒ u1 = − u2 = −

du2 g = − y sin(α) + c21 dy ν2

BC1 ⇒ c12 = 0 g g BC2 ⇒ µ2 (− (h1 + h2 ) sin(α) + c21 ) = 0 ⇒ c21 = (h1 + h2 ) sin(α) ν2 ν2 g g µ2 g BC3 ⇒ µ1 (− y sin(α)+c11 ) = µ2 (− y sin(α)+c21 ) {µ = νρ} ⇒ c11 = c21 = (h1 +h2 ) sin(α) ν1 ν2 µ1 ν1 g 2 g g 2 g BC4 ⇒ − h1 sin(α) + (h1 + h2 ) sin(α)h1 = − h1 sin(α) + (h1 + h2 ) sin(α)h1 + c22 2 ν1 ν1 2 ν2 ν2 ⇒ c22 = g sin(α) This gives us the velocities, u1 (y) = − u2 (y) = − = g 2 g g sin(α) 1 y sin(α) + (h1 + h2 ) sin(α)y = [(h1 + h2 )y − y 2 ] 2 ν1 ν1 ν1 2 1 1 − ν2 ν1 h2 1 − (h1 + h2 )h1 2 1 1 − ν2 ν1

g sin(α) 2 g sin(α) h2 y + (h1 + h2 )y + g sin(α) 1 − (h1 + h2 )h1 2 ν2 ν2 2

g sin(α) 1 h2 ((h1 + h2 )y − y 2 ) + g sin(α) 1 − (h1 + h2 )h1 ν2 2 2

1 1 − ν2 ν1

The velocity in layer 1 does depend on h2 but not on the viscosity in layer 2. This is due to that the depth is important for the tangential stress boundary condition at the interface but the viscosity is not. There is no acceleration of the upper layer and thus the tangential stress must be equal to the gravitational force on the upper layer which depends on h2 but not on ν2 .

B.6

More exact solutions to Navier-Stokes

Exercise 1

Oscillating Rayleigh–Stokes ﬂow (or Stokes second problem). a) Show that the velocity ﬁeld u = [u(y, t), 0, 0] satisﬁes the equation ¯ ∂u ∂2u =ν 2 ∂t ∂y Consider the Navier–Stokes equation in the x direction: ∂u ∂u ∂u 1 ∂p ∂2u ∂2u ∂2u ∂u +u +v +w =− +ν + 2 + 2 ∂t ∂x ∂y ∂z ρ ∂x ∂x2 ∂y ∂z With the current velocity ﬁeld only terms with y derivatives will remain since there can be no change in the other directions. Further more, the streamwise pressure gradient has to be zero since the streamwise velocity far from the wall is constant, namely zero. ∂u ∂2u =ν 2 ∂t ∂y b) Show that the velocity ﬁeld is u(y, t) = U e−ky cos(ky − ωt), where k = ω 2ν

Make the ansatz: u = Insert into the equation:

f (y)eiω t = f (y) cos(ω t). iωf (y)eiω t = νf (y)eiωt f (y) − λ2 − iω f (y) = 0 ν ⇒ with f (y) = eλy iω =± ν ω ν gives 1+i √ 2

iω =0 ν

λ=±

Introduce k =

ω , 2ν f (y) = Aeyk(1+i) + Be−yk(1+i) , f (y) → 0 as y → ∞ → A = 0

ui−1/2,j 1 i − hWe have 2 V i V1 u= Be−yk(1+i) eiwt = Be−ky ei(ωt−ky) = Be−ky cos(ωt − ky) = Be−ky cos(ky − ωt) i+ h 2 u i+1 uh i + 3 Boundary condition at y = 0 1 Γ2 0 j+2 Γ1 At y = 0 we have u = U cos(ωt) ⇒ B = U j 1 Ω j−2 : to be updatedSo we have j−1 ω : new values u(y, t) = U e−ky cos(ky − ωt), where k = A 2ν : old values B i−1 C 3 i u 2 ,1 5 iv1, 1 +3 j p 1 4.5 −2 1,1 j 4 j+1 1 3.5 iteration direction x 3 i y 2.5 j Inﬂow 2 n Solid wall 1.5 n Γ0 1 i Γ1 0.5 j Ω ﬁne grid (1) 0.2 0.4 0.6 0.8 M2 0 0 1 coarse grid (2) −1 −0.8 −0.6 −0.4 −0.2 x M m c) How thick is the boundary layer thickness? m+2 1 hIf we deﬁne the thickness δ of the oscillating layer as the position where u/U = 0.01 we get that V 2ν ν Vh e−kδ = 0.01 ⇒ kδ ≈ 4.6 ⇒ δ ≈ 4.6 ⇒ δ ≈ 6.5 ω ω u uh d) Consider instead the oscillating ﬂow U = U cos(ωt) over a stationary wall. This will simply result in a ∞ Γ0 change of frame of reference to one following the plate instead. If we consider the solution to the previous Γ1 problem and look at it in this new frame of reference we get Ω : to be updated ω u(y, t) = U cos(ωt) − U e−ky cos(ky − ωt), where k = : new values 2ν : old values i − 1The solution now looks like this i 5 i+1 j − 1 4.5 j 4 j+1 3.5 iteration direction x 3 y 2.5 n 2

1.5

1 i j 0.5 ﬁne grid (1) 0 coarse grid (2) −1.5

y

y

−1

−0.5

0 x

0.5

1

1.5

Exercise 2

Consider a long hollow cylinder with inner radius r1 and a concentric rod with radiusr0 inside it. The rod is moving axially with velocity U0 . a) Find the velocity ﬁeld of a viscous ﬂuid occupying the space between the rod and the cylinder. Assumptions Steady ﬂow:

∂ =0 ∂t u = uz (r)ez , ∂ = 0, ∂z ∂ =0 ∂θ

Parallel ﬂow and symmetry:

No streamwise pressure gradient:

∂p =0 ∂z We can directly see that ur = uθ = 0 satisfy the two ﬁrst equations. The streamwise momentum equation reduces to (u · )uz = ν 2 uz where (u ·

2

)uz = ur

∂uz uθ ∂uz ∂uz + + uz =0 ∂r r ∂θ ∂z + 1 ∂ 2 uz ∂ 2 uz 1 ∂ ∂uz + = r 2 ∂θ 2 r ∂z 2 r ∂r ∂r =0

uz =

1 ∂ ∂uz r r ∂r ∂r

We end up with ∂ ∂uz r ∂r ∂r Integrate twice r Boundary conditions uz (r0 ) = U0 uz (r1 ) = 0 = A ln r0 + B = A ln r1 + B ⇒ ⇒ B = −A ln r1 A= U0 ln r0 r1 ⇒ ∂uz =A ∂r ⇒

uz = A ln r + B

U0 = A(ln r0 − ln r1 ) uz (r) =

ln rr1 U0 U0 ln r − r0 ln r1 = U0 r0 ln r0 ln r1 ln r1 r1

b) With what force does one have to pull a rod with length L? Neglect end eﬀects. Shear stress τrz = µ Force F = 2πr0 Lτrz (r0 ) = ∂uz µU0 = r0 ∂r r ln r1 2πLµU0 r ln r0 1

B.7

Axisymmetric ﬂow and irrotational vortices

Axisymmetric ﬂow

Consider incompressible and rotationally symmetric ﬂow with no mass source at the symmetry axis. The velocity component in the direction of the axis of symmetry and the vorticity is given. uz = γ z and ω = ω(r) ez . ¯ ¯

1) Compute ur (r, z) and uθ (r, z) for the two cases when a) ω(r) = 0 The vorticity is given, ω= ¯ Look at the diﬀerent components: er : ¯ eθ : ¯ 1 ∂uz ∂uθ − = ω · er = 0 ¯ ¯ r ∂θ ∂z ⇒ uθ = uθ (r) ×u= ¯ 1 r

∂ ∂r ur

and

b) ω(r) = ω0 e−r er ¯ r¯θ e ez ¯

2

/a2

.

∂ ∂θ r uθ

∂ ∂z uz

∂ur ∂uz − = ω · eθ = 0 ⇒ ur = ur (r) ¯ ¯ ∂z ∂r 1 ∂(r uθ ) ∂ur ez : ¯ − = ω · ez = ω(r) ¯ ¯ r ∂r ∂θ ∂ = 0. From the ez component we get, ¯ ∂θ ∂ (r uθ ) = r ω(r) ∂r (B.1)

The rotational symmetry implies

**We also have the incompressibility condition: ·u= ¯ From this we get, 1 ∂ 1 ∂uθ ∂uz + =0 (rur ) + r ∂r r ∂θ ∂z
**

=0 γ

∂ (rur ) = −rγ ∂r Equation (2) is valid both for case a) and b). Case a) ω(r) = 0 (1) ⇒ (2) ⇒ ∂ (rur ) = −rγ ∂r ⇒ ∂ (r uθ ) = 0 ∂r ⇒ ⇒ uθ = A r r2 γ +B 2

(B.2)

integrate the rhs rur = − ur = −

**γr B + 2 r There should be no mass source at r = 0. The mass ﬂow is,
**

2π

Q=

0

ur r dθ = 2π B − 2π

γ r2 . 2

When r → 0 then Q → 2π B which gives B = 0. ur = − γr 2 for both a) and b)!

The circulation Γ is, Γ=

0

2π

uθ r dθ = 2πA ⇒ A =

Γ 2π

Case b) ω(r) = ω0 e−r

2

/a

2

, (1) ⇒

2 2 ∂ (r uθ ) = r ω0 e−r /a ∂r

**Integrate the right hand side: r uθ = − Look at the circulation, Γ(r) =
**

0

ω0 a2 −r2 /a2 e +C 2

2π

⇒

uθ = −

ω0 a2 −r2 /a2 C e + 2r r

uθ r dθ = 2π(− Γ0 = 2π(−

ω0 a2 −r2 /a2 e + C) 2 ⇒ C= Γ0 ω a2 + 0 2π 2

⇒ This gives

for r = 0 we get

ω0 a 2 + C) 2

uθ =

2 2 Γ0 ω a2 ++ 0 1 − e−r /a 2πr 2r

c) Consider circles C(t) following the ﬂow with a radius R(t) and position Z(t). Compute R(t) and Z(t). Can any of the ﬂow cases be inviscid? R(0) = R0 1 dR = ur (R) = − γR dt 2 dZ = uz (Z) = γZ dt The circulation is then,

2π

Z(0) = Z0 ⇒ ⇒ R(t) = R0 e− 2 γ t Z(t) = Z0 eγ t

1

Γ(t) =

0

uθ (R)R dθ = 2πRuθ (R) =

Γ a)

The circulation is constant for a) but not for b). This means that the ﬂow in a) can be inviscid.

Γ0 + πω0 a2 1 − e−R

2

/a2

b)

**Inviscid and irrotational vortices
**

Consider a circular ﬂow with u = uθ (r)¯θ . Which vortices are inviscid and which vortices are irrotational? ¯ e The Navier–Stokes equation for uθ ∂uθ 1 1 ∂ ∂uθ =− +ν r ∂t ρr r ∂r ∂r For inviscid ﬂow we require, 1 ∂ ∂uθ r r ∂r ∂r Make the ansatz uθ = rn , 1 ∂ 1 (rnrn−1 ) − rn−2 = n2 rn−1 − rn−2 r ∂r r We get the inviscid ﬂow, uθ (r) = The vorticity is, ω= ¯ ω=0 ¯ Conclusion: Irrotational ⇒ inviscid Inviscid irrotational ⇒ × u = {A.32} = ¯ ∂ (ruθ ) = 0 ∂r Ar

solid body rotation

−

uθ r2

−

uθ =0 r2

→ +

n2 − 1 = 0 B r

⇒ n = ±1

irrotational

1 ∂ (ruθ ) ez ¯ r ∂r ⇒ uθ = C r

Exercise 3

Show that the inviscid vorticity equation Dω ¯ = (¯ · ω Dt reduces to the equation D ω Dt r in the case of axisymmetric ﬂow: ¯ u = ur (r, z, t)¯r + uz (r, z, t)¯z e e The vorticity in an axisymmetric ﬂow ω= ¯ ¯ ×u= ∂ur ∂uz − ∂z ∂r

ω

)¯ u

=0

¯ eθ = ω¯θ e

Study the right hand side of the inviscid vorticity equation (¯ · ω )= ω ∂ r ∂θ ⇒ (¯ · ω )¯ = u ω ∂ ur (r, z, t)¯r + uz (r, z, t)¯z e e r ∂θ =

**ω ∂ur ω ∂¯r ω ∂uz e ω ∂¯z e ω ¯ ¯ ¯ e r + ur e z + uz = ur e θ + r ∂θ r ∂θ r ∂θ r ∂θ r
**

=0 =¯θ e =0 =0

The left hand side of the inviscid vorticity equation gives Dω ¯ ∂ω ¯ = + (¯ · u Dt ∂t )¯ = {¯ = ω¯θ } = ω ω e ∂ω ∂ ∂ ¯ + ur + uz ω eθ ∂t ∂r ∂z

This gives that the inviscid vorticity equation now is ∂ ∂ ω ∂ω + ur + uz ω = ur ∂t ∂r ∂z r Multiply by 1 r

∂ ω ∂t r −

+

1 ∂ ∂ ω ur + uz ω − 2 ur = 0 r ∂r ∂z r and that ωuz ∂ 1 =0 ∂z r 1 =0 r

Notice that

ω ∂ 1 u = ωur 2 r r ∂r r

This means we can write ∂ ω ∂t r And thus we have + 1 ∂ ∂ ∂ ∂ ur + uz ω + ω ur + uz r ∂r ∂z ∂r ∂z 1 ∂ ∂ ur + uz r ∂r ∂z ω D ω = r Dt r

∂ ω ∂t r

+

=0

B.8

Vorticity equation, Bernoulli equation and streamfunction

**Solid body rotation
**

Consider the ﬂow in a uniformly rotating bucket with velocity u = (ωr, 0, 0) a) Use Bernoulli equation to determine the free surface: What is wrong? Bernoulli: p 1 2 + |u| + gz = C along streamlines ρ 2 This gives the surface of constant pressure z= C − p0 ω 2 r2 − ρg 2g

The free surface is highest in the center of the bucket, something is wrong. Bernoulli theorem is valid along a streamline in a steady ideal ﬂuid. If the ﬂow had been irrotational, it would have been valid everywhere. ¯ But now × u = (0, 0, 2ω). b) Use Euler equation to determine the free surface: 1 Du =− p+g Dt ρ this gives er : − u2 1 ∂p θ =− r ρ ∂r eθ : ez : 0 = − Thus p(r, z) = This gives the free surface where p = p0 z= 1 ∂p −g ρ ∂z ∂p = ρω 2 r ∂r 1 ∂p ρr ∂θ ⇒ 1 2 2 ρω r + f (z) 2

⇒ 0=− ⇒

⇒

p=

∂p =0 ∂θ ⇒ f = ρgz + p0

∂f = −ρg ∂z

1 2 2 ρω r − ρgz + p0 2 ω 2 r2 2g

**Flow over a hill (Exam 20020527 question 2)
**

A hill with the height h has the shape of a half circular cylinder as shown in Figure 1. Far from the hill the wind U∞ is blowing parallel to the ground in the x-direction and the atmospheric pressure at the ground is p0 . a) (5) Assume potential ﬂow and show that the stream function in cylindrical coordinates is of the form ψ = f (r) sin θ, where f (r) is an arbitrary function. Calculate the velocity ﬁeld above the hill. b) (3) Derive an equation for the curve with constant vertical wind velocity V . c) (2) Assume that the density ρ and the gravitational acceleration g is constant. Calculate the atmospheric pressure at the top of the hill. a) The stream function satisﬁes continuity: ur = The ﬂow is irrotational: ×u=0 ¯ ⇒ 1 ∂ψ , r ∂θ uθ = − ∂ψ ∂r

∂ψ ∂2ψ 1 ∂2ψ +r 2 + = 0 (1) ∂r ∂r r ∂θ2

ration direction x y 1 n x y i j ﬁne grid (1) coarse grid (2)

y

x

Figure B.5: Streamlines above a hill with h = 100m and U∞ = 5m/s. A paraglider pilot with a sink of 1m/s will ﬁnd lift in the area within the dotted line, while soaring along the hill. Introduce the ansatz ψ = f (r) sin θ into equation (1): 1 f sin θ + rf sin θ − f sin θ = 0 r 1 f + rf − f = 0 r Make the ansatz f = r n : 1 nrn−1 + rn(n − 1)rn−2 − rn = 0 r n + n2 − n − 1 = 0 So we have ψ= We need two boundary conditions. 1. Free stream: Ar sin θ = U∞ r sin θ 2. Streamline on the hill surface: U∞ h + So we have: ψ = U∞ r − Now we can calculate the velocity ﬁeld above the hill: ur = 1 ∂ψ h2 = U∞ 1 − 2 r ∂θ r ∂ψ h2 = −U∞ 1 + 2 ∂r r cos θ B =0 h ⇒ h2 r B = −U∞ h2 sin θ ⇒ A = U∞ Ar + B r ⇒ n = ±1 ⇒ ⇒

sin θ

uθ = −

sin θ

b) Constant vertical wind velocity is described by: V = ur sin θ + uθ cos θ = U∞ 1 − h2 r2 cos θ sin θ − U∞ 1 + r=h −2 h2 r2 sin θ cos θ = −2U∞ h2 sin θ cos θ r2 ⇒

U∞ sin θ cos θ V

c) Use Bernoulli equation with free stream pressure p0 at the ground: 1 2 1 po + ρU∞ = p + ρ(2U∞ )2 + ρgh 2 2 3 2 ⇒ p = po − ρU∞ − ρgh 2

**Stokes stream function:
**

Consider a 2D incompressible ﬂow · u = 0 or ¯ Deﬁne the stream function Ψ such that: u= This means ∂Ψ ∂y v=− ∂Ψ ∂x ∂u ∂v + = 0. ∂x ∂y

∂u ∂v ∂ 2Ψ ∂2Ψ + = − = 0, ∂x ∂y ∂x∂y ∂y∂x

**so continuity is fulﬁlled. Now we can write ¯ ex u= ¯ And ω= ¯ Irrotational ﬂow
**

2

× Ψ¯z = e

∂ ∂x

¯ ey

∂ ∂y

¯ ez

∂ ∂z

=

0 ¯ ez

∂ ∂z

0

Ψ

∂Ψ ∂Ψ ,− ,0 ∂y ∂x

¯ ex ×u= ¯

∂ ∂x ∂Ψ ∂y

¯ ey

∂ ∂y − ∂Ψ ∂x

=

0

0, 0, −

∂Ψ ∂2Ψ − 2 ∂x ∂y 2

=−

2

Ψ¯z e

Ψ=0

In spherical coordinates for axisymmetrical ﬂow, deﬁne Ψ ur = Incompressibility is still valid ·u=0 ¯ Velocity u= ¯ Vorticity ω=− ¯ Irrotational × Ψ ¯ eθ r sin θ 1 ∂Ψ sin θ ∂θ =0 r2 1 ∂Ψ sin θ ∂θ uθ = − 1 ∂Ψ r sin θ ∂r

1 ∂ 2 Ψ sin θ ∂ + 2 r sin θ ∂r2 r ∂θ 1 ∂Ψ sin θ ∂θ

∂ 2 Ψ sin θ ∂ + 2 ∂r2 r ∂θ

**Flow around a sphere
**

Consider a sphere with Ψ = 0 at r = a, compute the irrotational velocity distribution when the velocity of the freestream at inﬁnity is U: 1 r → ∞ Ψ → U r2 sin2 θ 2 ⇒ ur → U cos θ uθ → −U sin θ Make the ansatz: Ψ = f (r) sin2 θ For an irrotational ﬂow we get f − From inﬁnity we get A= On the sphere 1 2 B Ua + =0 2 a Ψ= The slip velocity on the sphere is 2 f =0 r2 ⇒ 1 U 2 ⇒ 1 B = − U a3 2 sin2 θ f = Ar 2 + B r

This gives

1 a3 U r2 − 2 r

3 uθ = − U sin θ 2

The radial velocity is ur = 0.

ΩB.9 Flow around a submarine and other potential ﬂow problems : to be updated : new valuesExercise 1 : old values i − 1The ﬂow around a submarine moving at a velocity V can be described by the ﬂow caused by a source and a sink with strength Q at a distance 2a from each other. i i+1 x V j−1 j j+1 Submarine iteration direction x y Q -Q 1 p1,1 z n y y i j ﬁne grid (1) coarse grid (2)

a

L

Figure B.6: Coordinate system for submarine problem a) If one wants to construct a pressure sensor that will register an approaching submarine at a distance L, what sensitivity is needed for the sensor? Assume an ideal ﬂuid and that 2a = 80 m, Q = 915 m3 /s, U = 8 m/s, L = 200 m and ρ = 1000 kg/m3 .

Use a potential ﬂow description u= ¯ φ, u=

∂φ , ∂x

v=

∂φ ∂y

The ﬂow is always irrotational due to the deﬁnition of the velocity potential ω= For incompressibility we get ·u= ¯ · φ= ∂( φ)i = ∆φ = 0 ∂xi Q −Q + 4π r1 4π r2

source sink

×u= ¯

×

φ = 0,

curl(grad)=0

The equation is linear and thus superposition can be used. We have freestream plus 3D source plus sink φ= Uz

freestream

+

The ﬁrst term is in cylindrical coordinates (R, θ, z) and the two last are in two diﬀerent spherical coordinate systems with origin in z−a and z+a, respectively. Transform the two second terms to cylindrical coordinates φ = Uz + Velocity 4π (z + a)2 + R2 −Q + Q 4π (z − a)2 + R2

eR ¯

∂φ 1 ∂φ ∂φ eR + ¯ eθ + ¯ ez = ¯ ∂R R ∂θ ∂z QR QR Q(z + a) Q(z − a) − +¯z U + e − 4π((z + a)2 + R2 )3/2 4π((z − a)2 + R2 )3/2 4π((z + a)2 + R2 )3/2 4π((z − a)2 + R2 )3/2 u= ¯ φ=

We need to know the distance, b, from the point source to the stagnation point on the submarine nose. Thus we need to know the length of the submarine.

b) How long is the submarine? Compute where uz = 0 for R = 0 U+ Q 4π 1 1 − (z + a)2 (z − a)2 =0

−

Solving this system gives z = ±43.01 m, z = ±36.99 m, where the second solution lies inside the submarine. The length is then 2 ∗ 43.01 ≈ 86 m and b = 3.01 m. Now we continue to solve a) Use Bernoulli equation to determine the pressure ﬂuctuations at z = −L − b − a, R = 0 1 1 p + ρ|¯|2 = p∞ + ρU 2 u 2 2 Evaluate u noticing that uR = 0 ¯ u(z = −L − b − a, R = 0) = ez U + ¯ ¯ Q −Q + 4π(L + b)2 4π((L + b + 2a)2

4πU (z − a)2 − (z + a)2 −4az = = 2 Q (z − a)2 (z + a)2 (z − a2 )2

Inserting the given values gives |¯| = 7.99914 m/s and we get u p − p∞ = 1 2 ρ(U 2 − |¯|2 ) ≈ 6.86 N/m ≈ 0.07 mbar u 2

c) How wide is the submarine? To get this we need to compute the shape of the submarine. The stream function is constant along streamlines and is useful for this. In spherical coordinates the stream function is deﬁned as ur = 1 ∂ψ r2 sin θ ∂θ

1 ∂ψ r sin θ ∂r Transformation between cylindrical and spherical coordinates uθ = − ur = uR sin θ + uz cos θ R = r sin θ, z = r cos θ Our velocity ﬁeld gives 1 ∂ψ Q 1 1 = sin θ r sin θ − 2 + r2 sin θ ∂θ 4π (r2 + a2 + 2ar cos θ)3/2 (r + a2 − 2ar cos θ)3/2 cos θ U + Q 4π Q 4π r cos θ + a r cos θ − a − 2 (r2 + a2 + 2ar cos θ)3/2 (r + a2 − 2ar cos θ)3/2 (r2

= U cos θ + This is diﬃcult to integrate.

r + a cos θ r − a cos θ − 2 2 + 2ar cos θ)3/2 +a (r + a2 − 2ar cos θ)3/2

p1,1 Simplify to a Rankine body by neglecting the sink and say that a = 0 Q 1 1 ∂ψ 1 ur = U cos θ + = 2 ⇒ 2 4π r r sin θ ∂θ 2 i 1 Q j Ψ = U r2 sin2 θ − cos θ + C 2 4π Inﬂow Solid wallDetermine C from the stagnation point n Q 2 Γ0 ur (θ = π, r0 ) = 0 ⇒ r0 = 4πU Γ1 ΩSince Ψ = 0 on the body we get Q M2 C =− . 4π M mThe stream function is then 1 Q m+2 Ψ = U R2 sin2 θ − (cos θ +1) 2 4π 1 h source V d Vh The shape is given by Ψ = 0. As r → ∞, θ → 0 then r sin θ → . This gives 2 u uh Γ0 Γ1 Ω : to be updated 10 : new values : old values i−1 5 i i+1 j−1 j 0 j+1 iteration direction x y −5 1 n x y −10 i j ﬁne grid (1) −15 −15 −10 −5 5 10 15 0 coarse grid (2) −20

x

z

Figure B.7: Rankine body for submarine problem Q 1 d2 U −2 =0 2 4 4π 4Q Uπ Q Uπ

⇒

d2 =

⇒

d=2

There is a simple way of determining the radius as z → ∞ directly. The ﬂow from the source must take up an particular area in the ﬂow at inﬁnity. Since no ﬂuid can cross the streamlines this area must be equal to that of the Rankine body: Q = Uπ d2 2 ⇒ d=2 Q = 12.07 m Uπ

uh We can use the computed stream function for a point source and displace it to z = −a. In cylindrical Γ0 coordinates Γ1 −Q z+a Ψ= Ω 4π (z − a)2 + R2 to be updated : new valuesTransform to spherical coordinates : old values −Q −Q r cos θ + a r cos θ + a i−1 √ Ψ= = 4π 4π 2 r2 + a2 + 2ar cos θ (r cos θ + a)2 + r2 sin θ i i+1 j − 1The stream function for the submarine is then j r cos θ − a 1 Q r cos θ + a j +1 −√ +√ Ψ = U r2 sin2 θ + 2 + a2 + 2ar cos θ 2 + a2 − 2ar cos θ 2 4π r r ration direction x y 1 n x y i j ﬁne grid (1) coarse grid (2)

10

x

0

−10 −80 −60 −40 −20

0

20

40

60

80

z

Figure B.8: Submarine body for submarine problem At r = R and θ = π/2 we get Ψ= For the body Ψ = 0 and we get R2 Multiply by R2 R 2 + a2 + R 2 + a2 − 1 Q a −a U R2 + −√ +√ 2 + a2 2 4π R R 2 + a2 aQ =0 πU aQ πU ⇒

(R2 )3 + (R2 )2 a2 − Computing this

a2 Q2 =0 π2 U 2

d = 2R = 12.0006 m

m The complex potential m+2 1 hThe lines with constant stream function Ψ are the streamlines. They are orthogonal to the lines of constant V velocity potential φ which are equipotential lines. Since both of them satisfy Laplace’s equation we can Vh deﬁne a complex function F (z) = φ(x, y) + i Ψ(x, y) z = x + iy u uh Γ0 Γ1 Ω : to be updated 18 : new values : old values 16 i−1 i 14 i+1 j − 1 12 j j + 1 10 iteration direction x 8 y 1 6 n x 4 y i 2 j ﬁne grid (1) 0 −45 −40 −35 −30 −25 coarse grid (2) −50 Figure B.9: Complex potential for submarine problem, solid: Ψ , dotted: φ This is an analytical function since the Cauchy–Riemann equation holds ∂φ ∂Ψ = ∂x ∂y The velocity is then and ∂φ ∂Ψ =− ∂y ∂x

dF ∂φ ∂Ψ = +i = u − iv dz ∂x ∂x This enables the use of complex analysis, in particular conformal mapping that can be used to compute the ﬂow over airfoil shapes. w(z) =

m Example: m+2 1 h Flow past a rotating cylinder centered at z = λ at an angle of attack α V Vh (a + λ)2 iα iΓ F (z) = U (z + λ)e−iα + e − log(z + λ) u (z + λ) 2π uh 1 Γ0 2 Γ1 Mapping by z = 1 Z + 1 Z 2 − a2 gives an airfoil shape with the potential F(z). A correct ﬂow is not 2 4 Ω to be updatedachieved unless the Kutta–Joukovski condition is satisﬁed requiring : new values Γ = −4πU (a + λ) sin α : old values i−1 i i+1 Z=f(z) j −1 4 j 3 j +1 ration direction 2 2 x 1 y 0 0 n x y i j ﬁne grid (1) coarse grid (2)

−1 −2 −3 −4 −4 −2 −1 z=f (Z) −2 −5

0

5

0

2

Figure B.10: Conformal mapping from circle to airfoil shape, (a = 3,λ = 0.5)

B.10

More potential ﬂow

**Half body over a wall
**

A line source of strength Q is located at (0, a) above a ﬂat plate that coincides with the x-axis. A uniform stream with velocity U ﬂows along the x-axis. Calculate the irrotational ﬂow ﬁeld. Method of images. Put a line source of equal strength at (0, −a) in order to fulﬁll the condition of no ﬂow through the plate. Superposition of a uniform ﬂow and the two line sources gives the complex potential F = Uz + Complex velocity W = dF Q =U+ dz 2π Q 2π 1 1 + z − ia z + ia ⇒ ⇒ ⇒ Q Q ln(z − ia) + ln(z + ia) 2π 2π

W =U+ W =U+ W =U+

1 1 + x + i(y − a) x + i(y + a) x − i(y − a) x − i(y + a) + 2 2 + (y − a)2 x x + (y + a)2

Q 2π

The velocity ﬁeld now becomes

Q x x y−a y+a + 2 −i 2 + 2 2 + (y − a)2 2 2 2π x x + (y + a) x + (y − a) x + (y + a)2 Q 2π x x + 2 x2 + (y − a)2 x + (y + a)2

u=U+ v=

Q 2π

y−a y+a + 2 x2 + (y − a)2 x + (y + a)2

Γ0 Γ1 Flow past a symmetric airfoil Ωa) Use conformal mapping to calculate the irrotational ﬂow ﬁeld around a symmetric airfoil. to be updated : new values Joukowski transformation : old values c2 ζ(z) = z + i−1 z i i+1 j −1 j 1 2 j +1 ration direction 1 x 0.5 y 0 0 n x −1 y −0.5 i j −1 −2 ﬁne grid (1) −3 −2 −1 2 1 −1 −0.5 0.5 0 1 0 coarse grid (2) −1.5 Equation for the circle z = −λ + (a + λ)eiθ Equation for the airfoil ζ = −λ + (a + λ)eiθ + Complex potential in the z-plane F = U (z + λ)e−iα + U Complex velocity in the z-plane W = Complex velocity in the ζ-plane ω= dF/dz = dζ/dz U e−iα − U (a + λ)2 iα iΓ e + 2 (z + λ) 2π(z + λ) 1− a2 z2 ζ 2 /2 − a2 into (a + λ)2 iα iΓ dF = U e−iα − U e + 2 dz (z + λ) 2π(z + λ) a2 −λ + (a + λ)eiθ

(a + λ)2 iα iΓ e + ln(z + λ) (z + λ) 2π

The velocity can then be found by introducing the reversed transformation z = ζ/2 + u∗ = Re{ω(z)}, b) Calculate the Joukowski condition for the airfoil. v∗ = −Im{ω(z)}

The ﬂow ﬁeld has a singular point at the trailing edge of the airfoil at ζ = 2a. Resolve the singularity by choosing the circulation Γ so the numerator vanishes at the trailing edge z = a U e−iα − U eiα + Γk = 4π(a + λ)U iΓ =0 2π(a + λ) ⇒

eiα − e−iα = 4π(a + λ)U sin α 2i

B.11

Boundary layers

Exercise 1

Consider the high Reynolds number ﬂow in a conversing channel. Compute the boundary layer over the surface at y = 0. Assume the free stream: U (x) = − where Q is the ﬂux. The boundary layer equations read: u ∂u ∂u Q ∂2u +v =− 3 +ν 2 ∂x ∂y x ∂y ∂u ∂v + = 0 (2) ∂x ∂y Seek a similarity solution where the dimensionless velocity u/U only depend on the dimensionless wall distance y y y y Q η= = = = xν ν δ x ν x

|U | Q

Q , x

(1)

The stream function satisﬁes (2): u= Make the stream function dimensionless: f (η) = Determine the terms in equation (1): u= ∂f ∂η 1 Q ∂ψ = Uδ = U δf = U f = − f ∂y ∂η ∂y δ x ∂ψ , ∂y v=− ψ Uδ ∂ψ ∂x

v=−

∂ψ ∂(U δf ) ∂f ∂η ∂δ ∂U =− = −U δ −U f− δf = ∂x ∂x ∂η ∂x ∂x ∂x √ Q ν η Q ν Q ν Qν x f − + f − 2x f =− ηf x Q x x Q x Q x =− − Q ∂f ∂η Q Q + 2f = − f x ∂η ∂x x x Q f x =− − η Q Q + 2 f = 2 (ηf + f ) x x x Q f ν

∂u ∂ = ∂x ∂x

−

Q f x

∂u ∂ = ∂y ∂y

Q ∂f ∂η Q 1 Q =− f =− 2 x ∂η ∂y x δ x Q ∂f ∂η Q2 =− 3 f ν ∂η ∂y x ν

**∂2u Q =− 2 ∂y 2 x Insert into equation (1): − Divide by Q2 /x3 : −ηf f − f f Boundary conditions: u(0) = 0 u(∞) = 1
**

2

Q Q f (ηf + f ) + x x2

√

Qν Q ηf 2 x x

Q Q2 Q2 f =− 3 −ν 3 f ν x x ν ⇒

+ ηf f = −1 − f −f

2

+1=0 f (0) = 0 f (∞) = 1

⇒

M2 Substitute F (η) = f (η): M F − F 2 + 1 = 0 m F (0) = 0 m+2 1 F (∞) = 1 h V Solution: η 2 Vh F = 3 tanh2 √ + −2 3 2 u uh Γ0 Exercise 2 Γ1 ΩConsider the ﬂow downstream of a 2D streamlined body at high Reynolds number. Study the thin wake to be updateddownstream of the body where variations in y are much more rapid than variations in the downstream : new valuesdirection. Also assume that the wake is so small that we can write u = U + u1 where u1 is negative and : old valuesdescribes the wake. The length scale in x is L and in y it is δ with δ << L. U

i−1 i i+1 j −1 j j +1 ration direction x y n x y i j ﬁne grid (1) coarse grid (2)

Y

U+u 1

X

Figure B.11: Coordinate system for wake problem

a) Find the governing (linear) equations: We start with the boundary layer equations since δ << L, u ∂u 1 ∂p ∂2u ∂u +v =− +ν 2 ∂x ∂y ρ ∂x ∂y ∂u ∂v + =0 ∂x ∂y In the wake we have no pressure gradient, ∂p = 0. ∂x ∂u1 ∂x δ u1 << U L

From the continuity equation we get, ∂v ∂u ∂ =− =− U + u1 ∂y ∂x ∂x =− ⇒ v∼

Inserting u = U + u1 in the boundary layer equations gives, U Neglect the quadratic terms, u1 ∂u1 u2 ∼ 1 ∂x L and v ∂u1 δ 1 u2 ∼ u1 u1 ∼ 1 ∂y L δ L ∂u1 ∂u1 ∂u1 ∂ 2 u1 + u1 +v =0+ν ∂x ∂x ∂y ∂y 2

⇒ b) Show that

∞ −∞

U

∂u1 ∂ 2 u1 =ν ∂x ∂y 2

u1 dy = constant:

∞ −∞

Consider the relation, Q1 =

u1 dy

**which gives the linear contribution from u1 to the momentum ﬂux. d d Q1 = dx dx
**

∞ −∞

u1 dy =

∞ −∞

∂u1 dy = {From the governing equation} = ∂x

∞ −∞

ν ∂ 2 u1 ν ∂u1 dy = U ∂y 2 U ∂y

∞ −∞

=0

This means that Q1 is constant. c) Find a similarity solution for u1 : Seek a solution on the form, u1 (x, y) = F (x)f (η) This gives, Q1 = Q1 = constant requires, F (x) = 1/g(x) and This means, u1 (x, y) = Insert this into the equation of motion, −U g (x) ηf (η) + f (η) g(x)2 =ν 1 f (η) g(x)3 ⇒ ⇒ 1 y 1 f (η) = f g(x) g(x) g(x)

∞ −∞ ∞ −∞

where η =

y g(x) f (η) dη

F (x)f (η) dy = F (x)g(x)

∞ −∞

f (η) dη = Q1 = constant

**U f (η) g (x)g(x) = − =C ν ηf (η) + f (η) g(x) = This gives the equation for f , f (η) + Cηf (η) + Cf (η) = From the symmetry condition
**

∂ ∂y u(x, 0)

2C

νx U

1/2

d f (η) + Cηf (η) + D dη

=0

= 0 we can determine that f (0) = −D = 0. This gives, f (η) + Cηf (η) = 0

**Integrating this we get, f (η) = ae− 2 Cη
**

1 2

**We can determine the constant a from the condition that Q1 is constant,
**

∞ −∞

f (η) dη = Q1

⇒

f (η) =

2 1 C Q1 e− 2 Cη 2π

**This now gives u1 : u1 (x, y) =
**

y2 C Q1 − 1 C g(x)2 e 2 = 2π g(x)

g(x) = d) Relate u1 to the drag FD :

2C

νx U

1/2

Q1 = √ 2 π

U νx

1/2

e− 4νx

U y2

The drag is FD = −ρ

∞ −∞

u1 (U + u1 ) dy

**Remember that u1 << U and neglect the u2 term, 1 FD = −ρ This means that Q1 = − FD ρU ⇒
**

∞

−∞

U u1 dy = −ρU Q1 FD √ 2µ π ν Ux

1/2

u1 (x, y) = −

e− 4νx

U y2

Exercise 3

Give an order of magnitude estimate of the Reynolds number for: i. Flow past the wing of a jumbo jet at 150 m/s (≈ Mach 0.5) ii. A wing proﬁle in salt water with L = 2 cm and U = 5 cm/s iii. A thick layer of golden syrup draining of a spoon. iv. A spermatozoan with tail length of 10−3 cm swimming at 10−1 cm/s in water. Estimate the boundary layer thickness in case (i).

Fluid Water Air Syrup

ν cm2 /s 0.01 0.15 1200

i. Flow past the wing of a jumbo jet at 150 m/s (≈ Mach 0.5) U = 150 m/s, ν = 1.5 10−5 m2 /s, L = 4 m ⇒ Re = ii. A wing proﬁle in salt water with L = 2 cm and U = 5 cm/s U = 0.05 m/s, ν = 10−6 m2 /s, L = 0.02 m ⇒ Re = iii. A thick layer of golden syrup draining of a spoon. U = 0.04 m/s, ν = 0.12 m2/s, L = 0.01 m ⇒ Re = UL ≈ 0.003 ν UL ≈ 103 ν UL ≈ 4 × 107 ν

iv. A spermatozoan with tail length of 10−3 cm swimming at 10−2 cm/s in water. U = 10−4 m/s, ν = 10−6 m2 /s, L = 10−5 m ⇒ Re = UL ≈ 10−3 ν

The boundary layer thickness in (i) is O(1 mm) δ 1 =O √ L Re

m+2 More boundary layers 1B.12 h V Drag on a circular cylinder Vh Use the integral form of the Navier–Stokes equations, Consider a ﬁx volume in a ﬂuid with velocity u. The ¯ ucontinuity equation becomes, uh d ρ dV = − ρuk nk dS. Γ0 dt V S Γ1 The momentum equation, Ω : to be updated d ρui dV = − ρuj nj ui + pni − τij nj dS + ρFi dV : new values dt V S V : old values i − 1Use this to compute the drag force on a cylinder: i U n ωU 0 ≈ 0 i+1 j−1 j j+1 iteration direction x l n d Wake y n n 1 x y i j ﬁne grid (1) coarse grid (2)

L

**n ∂u ¯ = 0. Incompressible ﬂow with constant density. The ∂t
**

l/2

**The cylinder is stationary and so is the ﬂow ⇒ continuity equation gives: uk nk dS = 0
**

S

⇒

−U0 l − Lv|y=−l/2 + Lv|y=l/2 +

l/2

Uw (y) dy = 0

l/2

**Due to the symmetry we get, −U0 l + 2Lv|y=l/2 + The momentum equation in x:
**

S

Uw (y) dy = 0

l/2

ρnj uj ux + nx p − nj τxj dS = 0

l/2

**Far away from the cylinder we can assume p = p0 and τxx = τxy = 0. This gives,
**

2 −ρU0 l + 2ρU0 v|y=l/2 L + 2 ρUw (y) dy +

−l/2

cyl

[nx p − nj τxj ] dS

FD

**Multiply the continuity equation by ρU0 ,
**

l/2 2 −ρU0 l + 2ρU0 v|y=l/2 L = −ρU0

Uw (y) dy

−l/2

**Insert into momentum equation:
**

l/2 l/2

−ρU0

l/2

Uw (y) dy +

−l/2 −l/2

2 ρUw (y) dy + FD = 0 l/2 −l/2

FD = ρ

−l/2

2 2 U0 Uw (y) − Uw (y) dy = ρU0

Uw (y) − U0

Uw (y) U0

2

dy

**Make the variable substitution r = y/d,
**

2 FD = ρU0 d l/2d −l/2d

Uw (r) − U0

Uw (r) U0

2

2

dr

Now if d << l we get,

2 FD = ρU0 d ∞ −∞

Uw (r) − U0

θ

Uw (r) U0

2 dr = ρU0 θ

The momentum thickness is called θ and is a measure of the loss of momentum in the ﬂow. Now compute FD θ the drag coeﬃcient CD = 1 2 = 2 d ρU0 d 2

**Glauert wall jet
**

Compute an ODE for a self-similar wall jet. The boundary layer equations read: u ∂u ∂u ∂2u +v =ν 2 ∂x ∂y ∂y (1)

∂u ∂v + = 0 (2) ∂x ∂y Seek a similarity solution where the dimensionless velocity u/Um(x), where Um (x) is the jet core velocity, only depend on the dimensionless wall distance y η= g(x) The stream function satisﬁes (2): u= Make the stream function dimensionless: f (η) = Determine the terms in equation (1): u= v=− ∂ψ ∂f ∂η 1 = Um g = Um gf = Um f ∂y ∂η ∂y g − yg g2 = −Um gf − Um g f + Um g f η ψ Um (x)g(x) ∂ψ , ∂y v=− ∂ψ ∂x

∂ψ ∂(Um gf ) =− = −Um gf − Um g f − Um gf ∂x ∂x

**∂u ∂ g = (Um f ) = Um f − Um f η ∂x ∂x g ∂u ∂ 1 = (Um f ) = Um f ∂y ∂y g ∂2u ∂ = ∂y 2 ∂y Insert into equation (1): Um f (Um f − Um Um Um f
**

2 2 − Um

Um f

1 g

= Um f

1 g2

g 1 1 ηf ) + (Um g ηf − Um gf − Um g f )Um f = νUm 2 f g g g

⇒ ⇒

g 1 2 g 2 g ηf f + Um ηf f − Um Um f f − Um f f = νUm 2 f g g g g f + Um g 2 Um g g + ν ν

α

ff −

Um g 2 f ν

β

2

=0

⇒

f

+ αf f + βf

2

=0

Boundary conditions:

Similarity requires that α and β are constant. Assume the x-dependence: Um = axm , Then we get g = bxn

u(0) = 0 v(0) = 0 u(∞) = 0

⇒

f (0) = 0 f (0) = 0 f (∞) = 1

1 ab2 β = − amxm−1 b2 x2n = − m, ν ν α=

2n + m = 1

1 1 ab2 amxm−1 b2 x2n + axm bnxn−1 bxn = (m + n) ν ν ν We can choose the length scale g such that α = 1: ab2 1 = ν m+n So we have: f + f f + βf

2

⇒

β=−

m m+n

= 0 (3)

**For what β is the boundary conditions fulﬁlled? Work on (3): f f Integrate: f + f f + (β − 1)g = 0, Multiply with f : f f + ff f f + (f f Integrate: 1 f 2
**

2 2 2

+ f f + βf

2

=0

⇒

2

+ (f f + f 2 ) + (β − 1)f g(η) =

=0 f 2 dη > 0

∞ η

+ (β − 1)gf = 0

⇒

+ gf ) + (β − 2)gf = 0

∞ η

+ f g + (β − 2)

gf dη = 0

**Put in η = 0 and use the boundary conditions: (β − 2) β=− Solve for m and n:
**

∞ 0

gf dη = 0 ⇒

⇒

β=2

⇒

m =2 m+n 1 m=− , 2

3m + 2n = 0 3 4

n=

B.13

Introduction to turbulence

Exercise 1

Compute the far-ﬁeld two-dimensional turbulent wake U (x, y) behind a cylinder. Let U1 (x, y) = U0 − U (x, y), Assume Us Continuity U0 ⇒ ⇒ U1 U0 , u, v ∼ Us , x ∼ L, ⇒ y∼l V ∼ Us l L Us (x) = U1 (x, y = 0)

∂U ∂V + =0 ∂x ∂y

∂ ∂V ∂U1 = − (U0 − U1 ) = ∂y ∂x ∂x

∼V l ∼ Us L

The turbulent boundary layer momentum equation read: U ∂U ∂U 1 ∂P0 ∂2U ∂ +V =− +ν 2 − (uv) ∂x ∂y ρ ∂x ∂y ∂y ⇒

Disregard the pressure gradient and insert U = U0 − U1 : (U0 − U1 ) − Neglect small terms: U0 Self-similar hypothesis: U1 (x, y) = f (η) Us (x) −uv = g(η) 2 Us where η= y l(x) ⇒ ∂η y η = − 2l = − l , ∂x l l ∂η 1 = ∂y l ∂U1 ∂x −V ∂U1 ∂ 2 U1 ∂ = −ν − (uv) ∂y ∂y 2 ∂y ⇒

∂U1 ∂ = (uv) (1) ∂x ∂y

Determine the terms in equation (1): ∂ ∂η l ∂U1 = (Us f ) = Us f + Us f = Us f − Us ηf ∂x ∂x ∂x l ∂ ∂ 2 2 ∂η 21 (uv) = (−Us g) = −Us g = −Us g ∂y ∂y ∂y l Insert into equation (1): l 21 U0 (Us f − Us ηf ) = −Us g l l U0 lUs U0 l g =− f+ ηf 2 Us Us Boundary conditions: U1 (−∞) = 0 U1 (∞) = 0 ⇒ f (−∞) = 0 f (∞) = 0 ⇒

Similarity requires: lUs = const 2 Us Assume the x-dependence: Us = Axm , Then we get Bxn Amxm−1 = const · x2m ⇒ n + m − 1 = 2m l = Bxn ⇒ m=n−1 ⇒ l = Bxn ⇒ l = const Us

Us = Axn−1 ,

**We need one more condition. The momentum loss thickness is constant: θ=
**

∞ −∞

U U0

1−

U U0 1 U0

dy =

∞ −∞

∞ −∞

U0 − U 1 U0

1−

U0 − U 1 U0

dy =

∞ −∞

1−

U1 U0

U1 dy ≈ U0

U1 dy ≈=

∞ −∞

1 CD d = const (Recitation 12) 2

∞ −∞

1 U0

U1 dy = Us l

f dη = const

⇒

**const Us l = Axn−1 Bxn = const ⇒
**

1

2n − 1 = 0 l = Bx 2

1

⇒

n=

1 2

⇒

Us = Ax− 2 ,

For a suﬃciently large Reynolds number, the wake will be turbulent. Rel = A turbulent wake will remain turbulent! To determine the wake-velocity proﬁle, we need to model the turbulent shear stress. E.g. −uv = νT

2 Us g = ν T

AB lUs = = const ν ν

⇒

∂U ∂y

⇒ 1 l ⇒ ⇒ ⇒ ⇒

−

∂U1 ∂y ⇒

= −νT Us f g =− νT f Us l

g=− −

νT f Us l

νT U0 lUs U0 l f =− f+ ηf 2 Us l Us Us U0 l l U0 l 2 Us f − f =0 νT ν T Us U0 B 2 (f + f ) = 0 (2) 2νT

f +

f +

Equation (2) can be solved together with the boundary conditions f (±∞) = 0

B.14

Old Exams i

j

2

**Exam 031023 Inﬂow
**

Solid 1. Relative motion.wall n Consider the relative motion of two ﬂuid particles initially separated by the distance dx 0 . (Here x0 i i Γ0 ˆ and t are the Lagrangian coordinates). Γ1 ˆ a) (5) Show that the separation at time dt is Ω M2 ∂ui M ˆ ˆ dri (dt) = dx0 + 0 dx0 dt. i m ∂xj j m+2 b) (2) Use this relation to write down the expression for the deformation of the sides of a small cube 1 aligned with the coordinate directions. Thus, show that the components of the side aligned with the h x-direction, Rδi1 , is deformed into V Vh ∂ui ˆ u (1) dri = R(δi1 + 0 dt), uh ∂x1 Γ0 with analogous expressions for the other two sides. Γ1 Ω c) (4) Show that the deformation rate of the volume of the cube is given by the divergence of the : to be updated ˆ velocity ﬁeld. Note that x0 = xi at t = 0. i : new values 2. Flow between values and a cylinder. : old a rod −1 Consider a long ihollow cylinder with radius b, a concentric rod with radius a inside the cylinder, and i water occupying the space between the rod and the cylinder, see ﬁgure B.14. The rod is rotating i+1 with the constant angular frequency ω. j−1 (10) Find the velocity ﬁeld of the water driven by the rotating rod and the gravitational acceleration j g. j+1 iteration direction z x y θ 1 n x g y i j ﬁne grid (1) a b r coarse grid (2) Figure B.12: Water and a concentric rotating rod inside a cylinder.

j Inﬂow Solid wall n Γ0 Γ1 Ω M2 M m m+2 1 h V Vh u uh Γ0 Γ1 Ω : to be updated : new values : old values i−1 i i+1 j−1 j j+1 iteration direction x y n x y i j ﬁne grid (1) coarse grid (2)

**3. Bernoulli’s equation a) (3) Derive the following formula: uj ∂ui 1 ∂ = (uj uj ) + ∂xj 2 ∂xi
**

ijk ωj uk .

b) (3) Use this in the momentum equation to derive Bernoulli’s equation for unsteady potential ﬂow. c) (3) Use the result in (a) to derive Bernoulli’s equation for steady inviscid ﬂow. Discuss the validity of the derived relation. 4. Laminar wake ﬂow. Consider the laminar wake ﬂow downstream of a two-dimensional streamlined body at high Reynolds number, see ﬁgure B.13. Assume that the wake is so weak that we can write u = U + u 1 where u1 is negative and much smaller than the free-stream velocity U . a) (3) Motivate the usage of the equation U ∂u1 ∂ 2 u1 =ν . ∂x ∂y 2

**b) (2) Show that the mass ﬂux Q1 , in the direction of the wake is constant in x Q1 =
**

∞ −∞

ρu1 dy. (∗)

c) (7) A similarity solution can be found by scaling the wake velocity with Us (x) = u1 (x, 0) so that f (η) = u1 (x, y)/Us (x), where η = y/δ(x) is the similarity variable. Show, using (∗), that Us = k/δ(x), where k is a given constant. Find the similarity solution for u1 . U

Y U+u

1

X

Figure B.13: Wake downstream of a ﬂat plate. 5. Mean heat equation. The heat equation for the instantaneous turbulent ﬂow is ˜ ˜ ˜ ∂T ∂T ∂2T + uj ˜ =κ . ∂t ∂xj ∂xj ∂xj (8) Derive the heat equation governing the mean ﬂow.

**Answers to exam in Fluid mechanics 5C1214, 2003-10-23
**

1. See Kundu & Cohen p. 57. 2. Flow between a rod and a cylinder. uθ = 1g 4ν ωa2 b2 − a 2 b2 − r 2 r ln(r/a) ln(b/a)

uz = 3. See Kundu & Cohen pp. 110-114. 4. Laminar wake ﬂow.

r2 − a2 − (b2 − a2 )

u1 = 5. See Kundu & Cohen pp. 511-512.

Q1 2ρ

U − U y2 e 4νx πνx

Exam 040113

Γ0 Γ1 Ω 1. (5) Consider the unsteady ﬂow, M2 u = u0 , v = kt, w = 0, M where u0 and k are positive constants. Show that the streamlines are straight lines, and sketch them m at two diﬀerent times.2Also show that any ﬂuid particle follows a parabolic path as time proceeds. m+

2. (5) For surface waves1on water with ﬁnite depth h the phase speed is given by the relation c 2 = g h k tanh(kh). Compare the group velocity to the phase speed in the limit of long and short waves. V of a wave packet and explain how it is moving. Make a rough 2D sketch Vh 3. Consider a thin layer u liquid in contact with a solid ﬂat vertical wall. See ﬁgure 1. The wall is of u temperature gradient with increasing temperature in the downward direction. subject to a constant h Γ0 Due to a varying surface tension, as a result of the temperature gradient, a constant shear stress τ at the free surface Γ1 acting upward. At the same time, the gravitational acceleration g is acting is Ω downward. : to be updated a) (7) Assume constant layer thickness h and calculate the velocity ﬁeld of the liquid layer. : new values b) (3) For what layer thickness h is the ﬂux zero? : old values i−1 z i i+1 j−1 j τ j+1 iteration direction x g y 1 n x y i x h j ﬁne grid (1) coarse grid (2) Figure B.14: Liquid ﬁlm on vertical wall driven by gravity and surface tension.

**4. a) (3) Show that uj ∂ui 1 ∂ = (uj uj ) + ∂xj 2 ∂xi
**

ijk ωj uk

b) (4) Derive the vorticity equation starting with the Navier–Stokes equation for incompressible ﬂow. c) (3) Show that the following relation holds for incompressible ﬂow and discuss its implication for inviscid ﬂow ∂τij = −µ ∂xj 5. Consider the ﬂow over an oscillating plate. a) (3) Motivate that the solution can be written as [u(y, t), 0, 0] and that it thus satisﬁes the equation ∂2u ∂u =ν 2 ∂t ∂y b) (7) Show that u(y, t) = U e−ky cos(ky − ωt), where k = ω . 2ν

ijk

∂ωk ∂xj

6. a) (5) Derive the Reynolds average equation valid for turbulent ﬂow. b) (5) Assume that the Reynolds stress can be modeled as a turbulent viscosity and introduce this into the Reynolds equation and simplify.

2

j−1 A B C 3 u 2 ,1 v1, 3 2 p1,1 1

**Solutions to exam in Fluid mechanics 5C1214, 2004-01-13
**

1. Streamline (ﬁx t): dx = u = u0 ds dy = u = kt ds Then we see that y(x) is a linear function, y= ⇒ ⇒ x = u0 s + a y = kts + b

i j Inﬂow Solid wall n Γ0 Γ1 Ω M2 M m m+2 1 h V Vh u uh Γ0 Γ1 Ω : to be updated : new values : old values i−1 i i+1 j−1 j j+1 iteration direction x y n x y i j ﬁne grid (1) coarse grid (2)

kt x + d(t) u0

∂x ˆ = u = u 0 ⇒ x = u 0 t + x0 ˆ ∂t ∂y 1 ˆ ˆ = u = k t ⇒ x = k t2 + y 0 ˆ 2 ∂t Then we see that y(x) is a parabola, y= 1 k 2 (x − 2x0 x + x2 ) + y0 0 2 u2 0

Particle path (ﬁx x0 ): ¯

2. The wave number is deﬁned as k = 2π/λ, where λ denotes the wave length of the wave. For long waves, go to the limit k → 0. For short waves, go to the limit k → ∞. Phase speed: c= Short waves:

x→∞

ω = k lim

g tanh(kh) = k ⇒ ⇒

gh

tanh(kh) kh g k gh

tanh(x) 1 = x x tanh(x) =1 x dω c = dk 2

c=

Long waves:

x→0

lim

c=

**Group velocity: cg = Short waves:
**

x→∞

1+

2kh sinh(2kh) ⇒ ⇒ cg = c 2

lim

x =0 sinh(x) x =1 sinh(x)

Long waves:

x→0

lim

cg = c

**Longer waves travel faster. 2
**

1.5

1

0.5

0

−0.5 −1 −1.5

−2

0

0.5

1

1.5

2

2.5

3

3. a) Assume steady, parallel ﬂow u = w(x)¯z and solve the z-momentum equation with boundary ¯ e conditions: ∂ 2w ∂w τ ν 2 − g = 0, w(0) = 0, (h) = ∂x ∂x µ Integrate the equation once and introduce the second boundary condition: ∂w g = x + A, ∂x ν ∂w gh τ (h) = +A = ∂x ν µ ⇒ A= 1 ν τ − gh ρ

Integrate the equation once more and introduce the ﬁrst boundary condition: w= So we get: w= 1g 2 1 x + 2ν ν τ − gh x ρ 1g 2 x + Ax + B, 2ν w(0) = 0 ⇒ B=0

**b) The ﬂux per unit width of the wall can be written
**

h

Q=

0

w dx =

1g 3 11 x + 6ν 2ν

τ − gh x2 ρ 3 τ 2 ρg

h

=

0

11τ 2 1g 3 h − h 2ν ρ 3ν

Zero ﬂux gives: h=

**4. a) Begin with the second term on the right hand side:
**

ijk ωj uk

=

ijk jlm

**∂ ∂um ∂ui ∂uk (um )uk = (δkl δim − δkm δil )uk = uk − uk ∂xl ∂xl ∂xk ∂xi
**

ijk ωj uk

⇒

uj

∂ui ∂uj = uj + ∂xj ∂xi

⇒

uj

∂ui 1 ∂ = (uj uj ) + ∂xj 2 ∂xi

ijk ωj uk

b) Navier–Stokes momentum equation for incompressible ﬂow: ∂ui ∂ui 1 ∂p ∂ 2 ui + uj =− +ν ∂t ∂xj ρ ∂xi ∂xj ∂xj Use the expression from 4. a): ∂ui 1 ∂ + (uj uj ) + ∂t 2 ∂xi ∂ωi + ∂t ∂ωi + ∂t ∂ ( ∂xm

ijk ωj uk

=−

1 ∂p ∂ 2 ui +ν ρ ∂xi ∂xj ∂xj ∂ 2 ωi ∂xj ∂xj

**Take the curl of the equation. The curl of a gradient vanish:
**

imn njk ωj uk )

=ν

⇒ ⇒ ⇒

nim njk

∂ ∂ 2 ωi (ωj uk ) = ν ∂xm ∂xj ∂xj

∂ωi ∂ ∂ 2 ωi + (δij δmk − δik δmj ) (ωj uk ) = ν ∂t ∂xm ∂xj ∂xj ∂ωi ∂ ∂ ∂ 2 ωi + (ωi uk ) − (ωj ui ) = ν ∂t ∂xk ∂xj ∂xj ∂xj Use continuity. We end up with the vorticity equation: ∂ωi ∂ωi ∂ui ∂ 2 ωi + uj = ωj +ν ∂t ∂xj ∂xj ∂xj ∂xj c) Right hand side: −µ

ijk

∂ωk = −µ ∂xj

ijk

∂ ∂xj

klm

∂um = −µ ∂xl

kij klm

∂ 2 um = ∂xj ∂xl

2

−µ(δil δjm − δim δjl ) Left hand side:

**∂ 2 um ∂ 2 uj ∂ 2 ui = −µ − =µ ∂xj ∂xl ∂xj ∂xi ∂xj ∂xj ∂ui ∂uj + =µ ∂xj ∂xi
**

ijk 2

ui

∂τij ∂ =µ ∂xj ∂xj

ui

∂τij = −µ ∂xj Irrotational ﬂow is not subject to viscous forces.

Thus:

∂ωk ∂xj

5. a) Consider the Navier–Stokes equation in the x direction: ∂u ∂u ∂u ∂u 1 ∂p ∂2u ∂2u ∂2u +u +v +w =− +ν + 2 + 2 ∂t ∂x ∂y ∂z ρ ∂x ∂x2 ∂y ∂z With the current velocity ﬁeld only terms with y derivatives will remain since there can be no change in the other directions, ∂u ∂2u =ν 2 ∂t ∂y b) Make the ansatz: u = Insert into the equation: iωf (y)eiω t = νf (y)eiωt f (y) − λ2 − Introduce k = ω , 2ν f (y) = Aeyk(1+i) + Be−yk(1+i) , We have u= Be−yk(1+i) eiwt = Be−ky ei(ωt−ky) = Be−ky cos(ωt − ky) = Be−ky cos(ky − ωt) f (y) → 0 as y → ∞ → A = 0 iω f (y) = 0 ν ⇒ with f (y) = eλy iω =± ν ω ν gives 1+i √ 2 f (y)eiω t = f (y) cos(ω t).

iω =0 ν

λ=±

Boundary condition at y = 0 At y = 0 we have u = U cos(ωt) ⇒ B = U So we have, u(y, t) = U e−ky cos(ky − ωt), where k = ω 2ν

6. a) Turbulent ﬂow is inherently time dependent and chaotic. However, in applications one is not usually interested in knowing full the details of this ﬂow, but rather satisﬁed with the inﬂuence of the turbulence on the averaged ﬂow. For this purpose we deﬁne an ensemble average as 1 N →∞ N

N (n)

**Ui = ui = lim where each member of the ensemble ui
**

(n)

ui

n=1

is regarded as an independent realization of the ﬂow.

We are now going to derive an equation governing the mean ﬂow Ui . Divide the total ﬂow into an average and a ﬂuctuating component ui as ui = u i + u i p=p+p

and introduce into the Navier-Stokes equations. We ﬁnd ∂ui ∂ui ∂ui ∂u ∂ui ∂u 1 ∂p 1 ∂p + + uj + uj i + uj + uj i = − − +ν ∂t ∂t ∂xj ∂xj ∂xj ∂xj ρ ∂xi ρ ∂xi We take the average of this equation, using ui = 0 which gives uj ui = ui · uj = 0

2

ui +

2

ui

where the average of the continuity equation also has been added. Now, let U i = ui , as above, and drop the . We ﬁnd the Reynolds average equations ∂Ui + Uj ∂Ui = − 1 ∂P + ∂ (τij − ui uj ) ∂t ∂xj ρ ∂xi ∂xj ∂Ui =0 ∂xi

∂ui + uj ∂ui = − 1 ∂p + ν ∂t ∂xj ρ ∂xi ∂ui =0 ∂xi

2

ui −

∂ u i uj ∂xj

where ui uj is the Reynolds stress. b) Assume the Reynolds stress:

ui uj =

2 Kδij − 2νT eij 3

The ﬁrst term on the right hand side is introduced to give the correct value of the trace of u i uj . Here the deformation rate tensor is calculated based on the mean ﬂow, i.e. eij = 1 2 ∂Ui ∂Uj 2 ∂Ur + − δij ∂xj ∂xi 3 ∂xr

where we have assumed incompressible ﬂow. Introducing this into the Reynolds average equations gives ⇒ ∂Ui ∂ ∂Ui + Uj = − ∂t ∂xj ∂xj 1 ∂P P 2 + k δij + 2 (ν + νT ) eij = − + (ν + νT ) ρ 3 ρ ∂xi

2

Ui

where the last equality assumes that νT is constant, an approximation only true for very simple turbulent ﬂow. It is introduced here only to point out the analogy between the molecular viscosity and the turbulent viscosity.

Γ0 Exam 041018 Γ1 Ω 1. The kinetic energy dT of an inﬁnitesimal volume element, with density ρ, which is rotating with : to be updated angular velocity Ω can be expressed : new values 1 1 : old values dT = v · vρ dV = (Ω × x) · (Ω × x) ρ dV, 2 2 i−1 i where v is the velocity of the volume element and x is the vector from the center of rotation O to i+1 the volume element, see ﬁgure B.15. j−1 j x3 j+1 dV iteration direction x x y C x 1 x2 n x3 x1 x z y i j O ﬁne grid (1) x2 coarse grid (2)x1 Figure B.15: Coordinate system and deﬁnitions a) (5) Show that the total kinetic energy of the body can be expressed as T =

V

dT =

1 Jjl Ωj Ωl , 2

where Jjl is the moment of inertia tensor with respect to the origin of the unprimed system. It is deﬁned as Jjl = (xk xk δjl − xj xl )ρ dV.

V

b) (5) Show that you can relate Jjl to the moment of inertia tensor with respect to the center of mass C of the body Jjl in the following way

C Jjl = (zk zk δjl − zj zl )M + Jjl , C Jjl =

V

(xk xk δjl − xj xl )ρ dV,

where M is the total mass of the body and the primed coordinates represent the coordinate system centered at the center of mass C of the body. Hints: Integrals of the type V xk ρ dV vanishes as a result of the deﬁnition of the center of mass. The two coordinate systems are related by the formula x = z + x .

2. (10) Consider the laminar ﬂow of a viscous ﬂuid down a vertical cylindrical pipe with radius a. The ﬂow is driven by the gravitational acceleration g. Calculate the velocity ﬁeld. 3. Potential ﬂow and conformal mapping a) (2) Show that the Kutta-Joukowski transformation ζ = z + a2 /z maps a circle with radius a to a ﬂat plate with length 4a. b) (2) Write down the complex potential of the ﬂow around a circular cylinder with radius a at angle of attack α and with clockwise circulation Γ. The uniform far-ﬁeld velocity has the magnitude U . c) (4) Show that a necessary condition for ﬁnite velocity at the trailing edge of the ﬂat plate is (Kutta condition) Γ = 4πU a sin α. d) (2) How should one modify the circle in order to obtain a ﬁnite velocity at the leading edge? 4. Blasius boundary layer a) (2) Write down the boundary layer equations. b) (8) Make appropriate assumptions for similarity and reduce the boundary layer equations to an ordinary diﬀerential equation for the boundary layer over a ﬂat plate (i.e there is no pressure gradient in the free stream). Also state the boundary conditions for the similarity function. c) (2) Sketch the streamwise and normal velocity proﬁles. 5. Turbulent ﬂow a) (2) Deﬁne the velocity and length scales appropriate for turbulent ﬂow near a wall, using the wall shear stress τw , the viscosity ν and the density ρ. b) (2) What is the law of the wall and what is the explicit functional dependence of the turbulent mean velocity in the intermediate region between the wall and the outer ﬂow? c) (4) In two-equation turbulence models, partial diﬀerential equations for the turbulent kinetic energy ¯ k and the dissipation rate ε are solved. In those equations, the turbulent viscosity ν T = uT l, where uT and l are the turbulent velocity and length scales, respectively. Use dimensional analysis to determine ¯ uT , l and νT in terms of k and ε.

**Solutions to exam in Fluid mechanics 5C1214, 2004-10-18
**

1. a) The kinetic energy of an inﬁnitesimal volume element dT = dT = 1 1 v · vρ dV = (Ω × x) · (Ω × x) ρ dV ⇒ 2 2

**1 1 1 vi vi ρ dV = εijk Ωj xk εilm Ωl xm ρ dV = (δjl δkm − δjm δkl )xk xm Ωj Ωl ρ dV = 2 2 2 1 (δjl xk xk − xj xl )Ωj Ωl ρ dV ⇒ 2 T =
**

V

dT =

1 2

V

dT (δjl xk xk − xj xl )ρ dV Ωj Ωl = (δjl xk xk − xj xl )ρ dV.

1 Jjl Ωj Ωl , 2

where Jjl =

V

**b) The moment of inertia tensor Jjl =
**

V

(δjl xk xk − xj xl )ρ dV =

V

[δjl (zk + xk )(zk + xk ) − (zj + xj )(zl + xl )]ρ dV =

V

**[δjl (zk zk + 2zk xk + xk xk ) − (zj zl + zj xl + zl xj + xj xl )]ρ dV = (zk zk δjl − zj zl ) 2zk δjl
**

V

ρ dV +

V V

(xk xk δjl − xj xl )ρ dV + xj ρ dV.

xk ρ dV + zj

V

xl ρ dV + zl

V

**The three last terms vanishes due to the deﬁnition of the center of mass. So we have
**

C Jjl = (zk zk δjl − zj zl )M + Jjl ,

where M=

V

ρ dV

C and Jjl =

V

(xk xk δjl − xj xl )ρ dV.

2. Poiseuille ﬂow ( Pipe ﬂow ) Use the Navier–Stokes equations (with no volume force) in cylindrical coordinates ∂ur + (u · ∂t ∂uθ + (u · ∂t )ur − )uθ + u2 1 ∂p θ =− +ν r ρ ∂r

2

ur −

2

ur 2 ∂uθ − 2 2 r r ∂θ 2 ∂ur uθ − 2 r2 ∂θ r

ur uθ 1 ∂p =− +ν r ρ r ∂θ )uz = −

uθ +

2

∂uz + (u · ∂t

1 ∂p +ν ρ ∂z

uz

1 ∂ 1 ∂uθ ∂uz (r ur ) + + = 0. r ∂r r ∂θ ∂z ∂p ∂p We know that = 0 and = 0 and can directly see that ur = uθ = 0 satisfy the two ﬁrst ∂θ ∂r equations. From the continuity equation we get ∂uz =0 ∂z ⇒ uz = uz (r, θ) only.

**Considering a steady ﬂow we get from the Navier–Stokes equations (u · )uz = uz ∂uz ∂z ⇒ uz ∂uz =ν ∂z
**

2

uz − g

But we know that

**∂uz = 0 from the continuity equation. We get ∂z
**

2

uz =

1 ∂ ∂uz g (r )= r ∂r ∂r ν ⇒

⇒

∂ ∂uz g (r )= r ∂r ∂r ν

Integrate r Integrating again we get uz =

∂uz 1g 2 = r + c1 ∂r 2ν

∂uz 1g c1 = r+ ∂r 2ν r

1g 2 r + c1 ln r + c2 4ν

using the boundary conditions uz (r = 0) < ∞ ⇒ c1 = 0 ga2 and we have 4ν

We also have uz = 0 at r = a and this gives c2 = − uz = − 3. Potential ﬂow and conformal mapping a) Equation for the circle

g 2 (a − r2 ) 4ν

z = aeiθ Equation for the plate ζ =z+ a2 eiθ + e−iθ = aeiθ + ae−iθ = 2a = 2a cos θ, z 2

which is a real function taking the values from −2a to 2a, thus a ﬂat plate with length 4a. b) Complex potential for a circular cylinder a2 iα iΓ e + ln z z 2π

F = U ze−iα + U c) Complex velocity for the circular cylinder W = Complex velocity in the ζ-plane ω= dF/dz = dζ/dz U e−iα − U

dF a2 iΓ = U e−iα − U 2 eiα + dz z 2πz

a2 iα iΓ e + 2 z 2πz

1−

a2 z2

The ﬂow ﬁeld has singular points at the leading and trailing edges of the ﬂat plate. Resolve the singularity at the trailing edge by choosing the circulation Γ so the numerator vanishes at z = a U e−iα − U eiα + Γk = 4πU a iΓ =0 2πa

eiα − e−iα = 4πU a sin α. 2i

d) Move the point z = −a inside of the circle by putting the center at z = −λ. We still want a sharp trailing edge so the circle crosses the real axis at z = a. The radius of such a circle is a + λ and the circle is described by z = −λ + (a + λ)eiθ 4. Blasius boundary layer a) Boundary layer equations u ∂u ∂u dU ∂2u +v =U +ν 2 ∂x ∂y dx ∂y ∂u ∂v + =0 ∂x ∂y (B.3) (B.4)

b) Seek a similarity solution where the dimensionless velocity u/U , where U is the constant free-stream velocity, only depends on the dimensionless wall distance η= The stream function satisﬁes (2): u= Make the stream function dimensionless: f (η) = Determine the terms in equation (1): u= v=− ∂ψ ∂f ∂η 1 = Uδ = U δf = U f ∂y ∂η ∂y δ − yδ δ2 = U δ ηf − U δ f ψ(x, y) U δ(x) ⇔ ψ = U δf ∂ψ , ∂y v=− ∂ψ ∂x y δ(x)

∂ψ ∂(U δf ) =− = −U δ f − U δf ∂x ∂x

∂u ∂ δ = (U f ) = −U ηf ∂x ∂x δ ∂u ∂ 1 = (U f ) = U f ∂y ∂y δ ∂2u ∂ = 2 ∂y ∂y U Insert into equation (1): U f (−U −U 2 δ 1 1 ηf ) + (U δ ηf − U δ f )U f = νU 2 f δ δ δ ⇒ ⇒ Uf 1 δ =U 1 f δ2

dU =0 dx

δ δ δ 1 ηf f + U 2 ηf f − U 2 f f = νU 2 f δ δ δ δ Uδ δ f + ff = 0 ν

α

Similarity requires that α is a constant. We can integrate it with respect to x in order to ﬁnd δ Uδ δ =α ν ⇒ 1 2 ανx δ = 2 U ⇒ δ(x) = 2ανx U

The value of the constant α is just a matter of scaling, we can choose it to be 1/2 so δ(x) = The ordinary diﬀerential equation becomes f We have the boundary conditions u(0) = 0 v(0) = 0 u(∞) = U 1 + f f = 0. 2 f (0) = 0 f (0) = 0 f (∞) = 1. νx . U

⇒

c) Figure B.16 shows the dimensionless streamwise and normal velocity proﬁles.

i−

2 1 2

i+ i+1 i+ 3 2 j+1 2 j j−1 2 j −1 A B C 3 u 2 ,1 v1, 3 2 p1,1 1

i

5. Turbulent ﬂow a) Velocity scale near the wall: uτ = Length scale near the wall: l∗ = b) Law of the wall U+ = U = f (y + ), uτ where y + = y l∗ τw ρ ν uτ

Functional dependence in the intermediate region U+ = where κ is the Karmans constant. ¯ c) Dimensions for k and ε: 1 ln y + + B, κ

i j Inﬂow Solid wall n Γ0 Γ1 Ω M2 M m m+2 1 h V Vh u uh Γ0 Γ1 Ω to be updated : new values 7 : old values i−1 6 i i+1 j − 15 j j +1 4 ration direction η x y3 n2 x y i1 j ﬁne grid (1) 0 coarse grid (2) 0

Dimension analysis for uT :

¯ 1 k = 2 ui ui [L2 T −2 ] ∂u ∂u ε = ν i j [L2 T −3 ] ∂xj ∂xi ¯ uT ∝ k m ε n [LT −1 ] ⇒ ⇒ ⇒

LT −1 = L2m T −2m L2n T −3n 2m + 2n = 1 −2m − 3n = −1 uT ∝ Dimension analysis for l: ¯ l ∝ k m εn ⇒ ¯ k [L] ⇒

n=0 m= 1 2

L = L2m T −2m L2n T −3n 2m + 2n = 1 −2m − 3n = 0 ⇒

⇒ ⇒

n = −1 3 m= 2

v

u

0.2

0.4

0.6

0.8

1

Figure B.16: Velocity proﬁles of Blasius boundary layer.

l∝ So we have

¯3 k2 ε ⇒

¯3 2 ¯k νT = uT l ∝ k ε ¯ k2 νT ∝ ε νT = C µ ¯ k2 . ε

or

Appendix C

**Study questions 5C1212
**

1. Deﬁne and use the Rankine-Hugoniot jump condition to compute the shock speed for the following problem ut + uux = 0 − ∞ < x < ∞, t>0

u(x, 0) =

1 x≤0 0 otherwise

2. Deﬁne the entropy condition for a scalar conservation law. ut + f (u)x = 0 − ∞ < x < ∞, t>0

with a convex ﬂux function f (u). The shock is moving with speed s and the state to the left is given by uL and the state to the right by uR . Why do we need an entropy condition ? 3. Write a diﬀerence approximation (in 1D) on conservative form and deﬁne the notation. 4. Solve ut + ( u2 )x = 0, 2 u(x, 0) = 1 x<0 0 x≥0 (C.1)

by the upwind scheme on the following form, un+1 = un − j j b) If not, suggest another way of solving C.1. 5. Deﬁne a total variation decreasing (TVD) method. Why is this a desirable property ? 6. Investigate the one-sided diﬀerence scheme un+1 = un − a j j for the advection equation ut + aux = 0 Consider the cases a > 0 and a < 0. a) Prove that the scheme is consistent and ﬁnd the order of accuracy. Assume k/h constant. b) Determine the stability requirement for a > 0 and show that it is unstable for a < 0. 7. Apply Lax-Friedrichs scheme to the linear wave equation ut + aux = 0. a) Write down the modiﬁed equation. b) What type of equations is this? c) What kind of behavior can we expect from the solution? 173 ∆t n (u − un ) j−1 ∆x j ∆t n n u (u − un ) j−1 ∆x j j

a) Will the numerical solution converge to the analytical solution?

8. A the three-point centered scheme applied to ut + aux = 0, yields the approximation un+1 = un + j j a > 0.

a∆t (uj+1 − uj−1 ) 2∆x

Show that this approximation is not stable even though the CFL condition is fulﬁlled. 9. What does Lax’s equivalence theorem state? 10. What is the condition on the n × n real matrix A(u) for the system ut + Aux = 0 to be hyperbolic ? 11. The barotropic gas dynamic equations ρt + ρux = 0 1 ut + uux + px = 0 ρ where p = p(ρ) = Cργ i and C a constant, can be linearized by considering small perturbations (ρ , u ) around a motionless gas. a) Let ρ = ρ0 + ρ and u = u0 + u where u0 = 0. Linearize the system (C.2) and show that this yields the following linear system (the primes has been dropped) ρ t + ρ 0 ux = 0 a2 ut + ρ x = 0 ρ0 where a is the speed of sound. a and ρ0 are constants. b) Is the system given by (C.3) a hyperbolic system? Motivate your answer. c) Determine the characteristic variables in terms of ρ and u. d) Determine the partial diﬀerential equations the characteristic variables fulﬁll - characteristic formulation. e) Given initial conditions at t = 0 and let −∞ < x < ∞ (no boundaries) ρ(0, x) = sin(x) u(0, x) = 0 determine the analytical solution to (C.3) for t > 0. Hint: Start from the characteristic formulation. 12. The linearized form of (C.3) is given by ρ u +

t

(C.2)

(C.3)

0 a2 /ρ0

A

ρ0 0

ρ u

= 0,

x

(C.4)

where a is the speed of sound. a and ρ0 are constants. a) Draw the domain of dependence of the solution to the system (C.4) in a point P in the x-t plane. b) The system is solved numerically on a grid given by xj = j∆x, j = 0, 1, 2... and tn = n∆t, n = 0, 1, 2, ..... using an explicit three-point scheme, see the ﬁgure below. Draw the domain of dependence of the numerical solution at P (in the same ﬁgure as a)) of the three-point scheme in the case when i) the CFL condition is fulﬁlled ii) the CFL condition is NOT fulﬁlled. Assume that P is a grid point.

1t n+1 x y i t ﬁne grid (1) n coarse grid (2) 13. To solve Euler equations in 1D

j−1

j

j+1

ρt + ρux + uρx = 0 1 ut + uux + px = 0 ρ pt + ρc2 ux + upx = 0 How many boundary conditions must be added at (motivate your answer) inﬂow boundary when the ﬂow is a) Supersonic b) Subsonic outﬂow boundary when the ﬂow is c) Supersonic d) Subsonic 14. Describe the ideas behind a ﬂux splitting scheme for solving a non-linear hyperbolic system of equations, Ut + F (U)x = 0 15. a) Show that a vector ﬁeld wi can be decomposed into wi = u i + ∂p ∂xi

where u is is divergence free and parallel to the boundary. b) Apply this to the Navier-Stokes equations, show that the pressure term disappears and recover an equation for the pressure from the gradient part. 16. From the diﬀerential form of the Navier-Stokes equations obtain a) the Navier-Stokes equations in integral form used in ﬁnite-volume discretizations, b) a variational form of the Navier-Stokes equations used in ﬁnite-element discretizations. 17. a) Write down the appropriate function spaces for the pressure and velocity used to deﬁne weak solutions to the Navier-Stokes equations. b) Explain the concept of essential and natural boundary conditions. 18. a) How is a ﬁnite-element approximation deﬁned? b) Explain how to convert a FEM discretization to an algebraic problem, e.g. that the Navier-Stokes equations yield νK + C(u) G GT 0 u p = f 0

19. a) By choosing zh = uh in the ﬁnite element approximation of the Stokes problem, show that any Galerkin velocity solution is stable. b) Describe and interpret the LBB condition.

kk ll

g hgh opopop opop opop opop mnm op

jij i

20. Choice of elements. Discrete elements pairs a) constant pressure-bilinear velocities, b) the Taylor-Hood pair, c) a stable choice with piecewise linear velocities. 21. a) Derive the ﬁnite volume (FV) discretization on arbitrary grids of the continuity equation (∂u i /∂xi = 0), b) derive the FV discretization for Laplace equation on a Cartesian grid, c) show that both are equivalent to a central diﬀerence approximation for Cartesian grids. 22. Derive the ﬁnite element (FEM) discretization for Laplace equation on a Cartesian grid and show that it is equivalent to a central diﬀerence approximation. 23. Iterative techniques for linear systems. a) Deﬁne Gauss-Seidel iterations for the Laplace equations, b) Deﬁne the 2-level multigrid method for the Laplace equation, 24. State the diﬃculties associated with the the ﬁnite-volume discretizations of the Navier-Stokes equations on a colocated grid? and show the form of the spurious solution which exist. 25. a) Deﬁne an appropriate staggered grid that can be used for the discretization of the Navier-Stokes equations, b) write down the FV discretization of the Navier-Stokes equations on a staggered cartesian grid, c) discuss how to treat noslip and inﬂow/outﬂow boundary conditions. 26. Time dependent ﬂows. a) Deﬁne a simple projection method for the time dependent Navier-Stokes equations d dt u 0 + N(u) G D 0 u p = f g

b) show in detail the equation for the pressure to be solved at each time step and discuss the boundary conditions for the pressure. 27. Time step restriction for Navier-Stokes solutions. a) Motivate the use of an appropriate form of the advection-diﬀusion equation as a model equation for stability analysis, b) derive the time step restrictions for the 1D version of that equation, c) state the 2D equivalent of that restriction.

Bibliography

[1] D.J. Acheson, 1990, Elementary Fluid Dynamics, Oxford University Press. [2] J.D. Anderson, Jr., 1995, Computational Fluid Dynamics, The basics with applications, Mc Graw-Hill. [3] C.A.J. Fletcher, 1991, Computational Techniques for Fluid Dynamics Volume 1, Second Edition, Springer-Verlag. [4] C.A.J. Fletcher, 1991, Computational Techniques for Fluid Dynamics, Volume 2, Second Edition, Springer-Verlag. [5] M. Griebel, T. Dornseifer and T. Neunhoeﬀer, 1998, Numerical Simulation in Fluid Dynamics, A practical Introduction, SIAM Monographs on Mathematical Modeling and Computation. [6] P.K. Kundu and I.M. Cohen, 2002, Fluid Mechanics, Second Edition, Academic Press. [7] R.L. Panton, 1995, Incompressible Flow, Second Edition, Wiley-Interscience. [8] R. Peyret, T.D. Taylor, 1983, Computational Methods for Fluid Flow, Springer-Verlag. [9] J.C. Tannehill, D.A. Anderson and R.H. Pletcher, 1997, Computational Fluid Mechanics and Heat Transfer, Second Edition, Taylor and Francis. [10] P. Wesseling, 2001, Principles of Computational Fluid Dynamics, Springer-Verlag.

177

Computational fluid dynamics, properties and software paradigms.

Computational fluid dynamics, properties and software paradigms.

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd