MorseDressler FluidMechanics v1 Comp

1
Fluid Mechanics
Roman Morse
Kristofer Dressler
December 17, 2019
c 2019 - Morse and Dressler - All Rights Reserved

c Copyright by
Roman W. Morse and
Kristofer M. Dressler
2019
All Rights Reserved

Acknowledgements
At some point we’ll write something really nice about

• Prof. Chris Rutland
• Prof. Mario Trujillo
• Prof. Greg Nellis
• Prof. Arganthael Berson

but for now... Thanks.

Contents
0.1 Some mechanics of this class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

0.1.1 Grade calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.1.2 Participation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.1.3 Homework Grading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.1.4 Goals for this class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.2 What is a fluid? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.3 Systems and Control Volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 A Mathematics Review 1
1.1 Scalars, Vectors & Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Scalars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.3 Unit & Normal Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.4 Velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Vector Calculus pt.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 Gradient of scalar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.2 Gradient of a vector field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Vector & Matrix Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4.1 Dot Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4.2 Cross Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4.3 Matrix Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.5 Vector Calculus Part 2: Integrals, Fluxes and Integrated Fluxes . . . . . . . . . . . . . . . . . 16
1.5.1 Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.5.2 Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5.3 Divergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.5.4 Curl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.5.5 Surfaces, Volumes & The Divergence Theorem . . . . . . . . . . . . . . . . . . . . . . 23
2 Fluid Statics 27
2.1 Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2 Statics in general . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.3 A Fluid Element at Rest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.3.1 Manometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.3.2 Pressure Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.4 Buoyancy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5 Frank’s Tips for Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3 Fluids in Motion 46
3.1 Flow Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.2 Lagrangian and Eulerian Flow Field Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.3 Flow Regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.3.1 Steady vs Transient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.3.2 Compressible vs Incompressible . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.3.3 1- 2- and 3-dimensional Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.3.4 Internal vs External Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.3.5 Viscous vs Inviscid Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.3.6 Laminar vs Turbulent Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.3.7 Single phase vs multiphase Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4 Flow Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4.1 Timelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4.2 Streaklines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4.3 Pathlines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4.4 Streamlines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.5 Reynolds Transport Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.6 Material Derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.6.1 Derivatives in Lagrangian and Eulerian Reference Frames . . . . . . . . . . . . . . . . 58
3.6.2 Some Physical Intuition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.6.3 Acceleration, du/dt, Lagrangian and Eulerian Senses . . . . . . . . . . . . . . . . . . . 62
4 Conservation Laws 66
4.1 Conservation of Mass - Integral Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.1.1 Example: Shrinking Balloon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.1.2 Example: Developing Flow over a Flat Plate . . . . . . . . . . . . . . . . . . . . . . . 68
4.1.3 Example: Heating Air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.2 Conservation of Momentum - Integral Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.2.1 Frictionless Di↵erential Cylindrical Slug of Liquid Steadily Accelerating Along a Stream-
line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.2.2 Steady Flow with One Inlet and One Outlet . . . . . . . . . . . . . . . . . . . . . . . . 76
4.2.3 Summary of control volume analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2.4 Accelerating Control Volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5 Di↵erential Fluid Flow 90

5.1 Conservation of Mass - Di↵erential Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.1.1 Incompressible fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.2 Forces on Di↵erential Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.2.1 Inviscid Di↵erential Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.2.2 Viscous Di↵erential Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.3 Canonical Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.3.1 Terms that Indicate Flow Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.3.2 Couette Flow - Steady, Incompressible, Shear -driven, Isothermal, Two-Dimensional
Flow between Two Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.3.3 Steady, Incompressible, Pressure-driven, Isothermal, Two-Dimensional Flow between
Two Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
5.3.4 Navier-Stokes in Cylindrical Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.3.5 Poiseuille Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.3.6 Strategy for di↵erential fluid analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6 External flows 121

6.1 von Karman Drag Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6.2 Prandtl Boundary Layer Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.3 Blasius Boundary Layer Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.4 Turbulent Boundary Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
6.5 Lift and Drag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

7 Dimensional Analysis and Similitude 144
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
7.2 Experimental Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
7.2.1 Comparing Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
7.3 Dimensional Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
7.4 Buckingham ⇧ Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
7.4.1 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
7.4.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
7.5 Some Classic Pi Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
7.6 Entrance Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
7.7 Drag Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
8 Pipes & Pumps 171

8.1 Pipe Flow Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
8.1.1 Thermodynamics Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
8.1.2 Pipe Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
8.2 Head Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
9 Streamlines & Potential Flow 181

9.1 Stream Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
9.2 Potential Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
9.3 Properties of and . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
9.3.1 Uniform Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
9.3.2 Sources & Sinks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
9.3.3 Vortex flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
9.4 Super position of potential functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
9.4.1 Dipole & Doublets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
9.4.2 Doublet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
9.4.3 Half-body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
9.4.4 Rankine Oval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194

List of Figures
1 So.... Fluid mechanics is uninteresting but really hard? . . . . . . . . . . . . . . . . . . . . . .
1.1 Flow visualization of velocity field for a turbulent jet . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Normal vectors on a cylinder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 A cubical volumetric element experiences stress on three planes. To fully describe the stress,
three faces are identified by their normal vectors, and each face has stress in two directions. . 6
1.4 Three pairs of vectors: mostly aligned, mostly perpendicular and mostly opposite. . . . . . . 10
1.5 Schematic of a huge Icelandic guy pulling a rail car with a chain. Vector representation of the
same. Unit vector representing the direction of the track. Projection of chain tension onto
track. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6 Cross product does not commute. a ⇥ b has the same magnitude as b ⇥ a, but according to
the right hand rule, the resulting vectors point in opposite directions. . . . . . . . . . . . . . 14
1.7 Smelt, typical of Lake Superior. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.8 The Sun imaged by NASA (left) and a visualization of energy flux vectors (right) . . . . . . . 20
1.9 Typical spherical coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.1 Shipwreck in Lake Michigan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.2 A cubical fluid element experiencing a pressure acting on both x-surfaces. The pressures di↵er
by @P/@x dx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.3 Pressure on point within the water is felt equally from all directions . . . . . . . . . . . . . . 34
2.4 Fluid at the same depth, but di↵erent columns above . . . . . . . . . . . . . . . . . . . . . . . 35
2.5 James Cameron traveling in his submarine to the Titanic and Mariana Trench . . . . . . . . 35
2.6 Collapsable water trough. Two plates, hinged at the bottom and held together by a cable. . . 37
2.7 USS Gerald R. Ford - U.S. Navy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.8 A cylinder with radius, r, and length, L. The left cylinder is just floating in space. But the
right cylinder is floating in a tank of water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.9 A physicist, a horse, and the sphere the physicist uses to represent the mass of the horse.
Images from Aki, Wikipedia and Wolfram respectively. Used without permission but in good
fun. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.10 The floating USS Gerald Ford represented as a collapsable horse trough. Important dimensions
of the ship are length, L, draft, D, beam, B and dead rise, ✓. . . . . . . . . . . . . . . . . . . 42
3.1 Ocean waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

3.2 A steady-state nozzle operating at conditions ṁin = ṁout . . . . . . . . . . . . . . . . . . . . 47
3.3 Lagrange description - following particles as they travel downstream . . . . . . . . . . . . . . 47
3.4 Euler description - watching a control volume . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.5 Euler the fisherman and the Lagrangian fish . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.6 An evaporator with liquid refrigerant entering at the bottom left and vapor exiting the top
right. The boiling process results in several di↵erent two-phase flow regimes. . . . . . . . . . 52
3.7 A converging nozzle with some fluid passing through at time, t. The control volume is shown
with dashed lines and the system is the cross hatched area within the control volume. . . . . 54
3.8 A converging nozzle with some fluid passing through at time t + t. Some of the system has
left the control volume. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

3.9 An inconveniently shaped control volume with an awkwardly distorted system and mass fluxes
in and out at random places at all di↵erent angles other than orthogonal to the control surface. 56
3.10 A SCUBA diver swimming through a cloud of octopus ink in an attempt to photograph a sea
urchin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.11 Steady through a nozzle with velocity vectors at inlet and outlet . . . . . . . . . . . . . . . . 64
4.1 SpaceX launches the Falcon Heavy rocket in April 2019 . . . . . . . . . . . . . . . . . . . . . 66

4.2 The New Zealand based Te Aihe AC75 class sailboat. It converts wind motion into forward
thrust to move it through the water. As it moves through the water, foils convert the flow of
water into a vertical force to lift it out of the water. The rudder converts the flow of water
into a lateral force to steer. In e↵ect, horizontal wind is lifting a 7,600 kg boat out of the
water at over 50 mph (20 m/s). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.3 A cylindrical slug of liquid accelerating along streamline . . . . . . . . . . . . . . . . . . . . . 74
4.4 Falcon Heavy rocket launch - SpaceX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.5 Falcon Heavy rocket with control volume displayed in green . . . . . . . . . . . . . . . . . . . 81
4.6 A steady 30 m/s jet being deflected by a curved surface mounted to a cart that is moving at
a steady10 m/s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.7 Good control volume labeled with green dashed line . . . . . . . . . . . . . . . . . . . . . . . 84
4.8 A pipe bent into a 180 C flow reversal device. . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.9 A nozzle bent into a 180 C flow reversal device. . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.10 A nozzle accelerating and deflecting a stream of water downward. . . . . . . . . . . . . . . . . 89
5.1 Clouds flowing around a mountain generates Von Karman vortex street - NASA . . . . . . . 90
5.2 Fluid element with volume equal to V – = dx dy dz. . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.3 Velocity field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.4 Wave propagating in the x-direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.5 Force due to pressure on the x-direction faces . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.6 Forces in the x-direction due to shear at the surfaces. . . . . . . . . . . . . . . . . . . . . . . 100
5.7 Top-down view of the cubical element from Figure 5.6. Forces in the x-direction due to shear
on the z-surfaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.8 Flow between two plates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.9 Horizontal flow through a circular pipe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.10 Flow between two infinite plates with applied pressure gradient. . . . . . . . . . . . . . . . . . 114
5.11 Thin (h) film of water flowing down an inclined plate. . . . . . . . . . . . . . . . . . . . . . . 116
6.1 CFD simulation with steamlines over airplane wing - ANSYS . . . . . . . . . . . . . . . . . . 121
6.2 Free stream flow of U1 approaching a sharp flat plate. The region of influence of the plate,
the boundary layer, grows as the flow moves down the plate. . . . . . . . . . . . . . . . . . . . 122
6.3 Boundary layer over flat plate with von Karman’s control volume. CV is bounded on the left
and right by vertical surfaces at x = 0 and x = x. Bottom surface is the flat plate itself. Top
surface follows a streamline just outside the boundary layer. . . . . . . . . . . . . . . . . . . . 123
6.4 Displacement thickness and momentum thickness . . . . . . . . . . . . . . . . . . . . . . . . . 131
6.5 Laminar and turbulent boundary layer thicknesses, 99 , in air at 20 C and 100 kPa along a
flat plate with no pressure gradient. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
6.6 Laminar and turbulent viscous drag coefficients in air at 20 C and 100 kPa along a flat plate
with no pressure gradient. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6.7 Cartoon showing boundary layer concepts as the transition from laminar to turbulent occurs.
Not to scale. Image source uncertain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6.8 Turbulent boundary layers begin as laminar boundary layers and a transition zone before they
become fully turbulent. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6.9 Fluid draining through a rectangular channel . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
6.10 Fluid draining through a circular pipe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
6.11 Cross flow past a sphere (or cylinder) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
6.12 Cross flow past a strange looking shape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

6.13 Pressure distribution on a cambered airfoil. Note that the majority of the lift comes from the
leading edge of the airfoil. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6.14 Typical explanation for powered flight. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.15 Side view of a Cessna 172N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
6.16 Some geometric definitions of a symmetric airfoil. Image from White. . . . . . . . . . . . . . 139
6.17 Smoke lines around airfoils for low and high angles of attack . . . . . . . . . . . . . . . . . . . 140
6.18 Lift and drag on a NACA 0009 symmetric airfoil. As lift increases, so does drag. . . . . . . . 140
6.19 Cessna above Lake Michigan, making the 100 mile flight from Milwaukee, WI to Grand Haven,
MI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.20 Cessna now gliding (hopefully to safety) after the engine has failed. . . . . . . . . . . . . . . . 142
6.21 Some more airfoil geometry, including camber. This airfoil is no longer symmetric, and would
behave di↵erently if flying upside down. Image from, you got it, Wikimedia Commons. . . . . 142
6.22 The Rutan Voyager and the P-51 Mustang. The two aircraft had very di↵erent missions and
hence very di↵erent wing designs. Photo credits NASA (Voyager) and Ron Fernuik (P-51).
Printed under fair use. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
7.1 Proper scaling between model and full scale must be done to accurately correlate model results
and full-scale e↵ects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
7.2 A cantilevered beam deflecting by some amount, , under some load, F . . . . . . . . . . . . . 145
7.3 Some velocity profiles in a turbulent boundary layer from the MTL wind tunnel at the Royal
Institute or Technology in Stockholm, Sweden. The Reynolds numbers are between 105 and 108 148
7.4 Same velocity profile data as plotted in Figure 7.3, plotted in non-dimensional wall units
showing that y + = f (u+ ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
7.5 Some made up data relating the the drag coefficient, CD , to the Reynolds number, Re. . . . . 154
7.6 A specific Reynolds number corresponding to a specific drag coefficient. Any wing operating
at this specific Re will experience this specific CD . . . . . . . . . . . . . . . . . . . . . . . . . 154
7.7 Flow development for laminar flow in the x-direction. . . . . . . . . . . . . . . . . . . . . . . . 160
7.8 Variation of drag coefficient with Reynolds number for flow over a flat plate. . . . . . . . . . 164
7.9 Flow over and around a flat plate oriented perpendicular to the free stream. . . . . . . . . . . 164
7.10 Sources of drag for common geometries. Taken from Wikipedia. . . . . . . . . . . . . . . . . . 165
7.11 Drag coefficients of several 2-dimensional objects. Table taken from Frank White, Fluid Me-
chanics, McGraw-Hill, 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
7.12 Drag coefficients of several 3-dimensional bodies. Table taken from Frank White, Fluid Me-
chanics, McGraw-Hill, 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
8.1 CFD simulation through a pump - Siemens . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

8.2 Roughness shown for laminar, transitioning and turbulent flows . . . . . . . . . . . . . . . . . 175
8.3 Roughness values for common pipe and duct materials . . . . . . . . . . . . . . . . . . . . . . 175
8.4 Moody Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
9.1 Pressure field and streamlines of external flow over air foil . . . . . . . . . . . . . . . . . . . . 181
9.2 Two-dimensional channel flow with a floating vorticity gauge made of popsicle sticks. The
vorticity gauge is translated by the flow and rotated by the velocity gradient. . . . . . . . . . 182
9.3 Infinitesimal change in the direction of S, is denoted as the vector ds. . . . . . . . . . . . . . 184
9.4 Two-dimensional inviscid, irrotational channel flow with arbitrary values of stream functions
assigned to streamlines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
9.5 Two-dimensional inviscid, irrotational converging-diverging elbow. Vertical velocity is low at
the inlet, where the streamlines are far apart. As streamlines converge, velocity increases.
Values of and are arbitrary, but the relationship between them is correct. . . . . . . . . 186
9.6 Streamlines for source and sink flow fields. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
9.7 Pure vortex flow in the clockwise direction with streamlines drawn in blue . . . . . . . . . . . 190
9.8 Dipole flow field due to source at ( , 0) and a sink at ( , 0) . . . . . . . . . . . . . . . . . . . 191
9.9 Full 2D dipole flow field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
9.10 Streamlines of a doublet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

9.11 The flow around a half-body: (a) superposition of a source and a uniform flow; (b) replacement
of streamline = ⇡bU with solid boundary to form half-body. Figure from Munson, et. al.
Printed under fair use. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
9.12 The flow around a Rankine Oval: (a) superposition of source - sink pair and a uniform flow;
(b) replacement of streamline = 0 with solid boundary to form a Rankine oval. Figure from
Munson, et. al. Printed under fair use. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
9.13 E↵ect of oval parameters on shape. Figure from Munson Printed under fair use. . . . . . . . . 195
9.14 E↵ect of rotation on streamlines around a cylinder. Figure from White. Printed under fair use.195

Welcome
Welcome to ME 363 - Fluid Mechanics. We hope that you will find this class and the topic interesting,
inspiring, challenging and rewarding.
You may have heard grumbling in the hallways about how hard fluid mechanics is. It’s an engineering
class, so it really shouldn’t be easy. But there may be some misconceptions...
Figure 1: So.... Fluid mechanics is uninteresting but really hard?
Fluids is not only fascinating regarding the natural world, but it could also be a very lucrative business
or professional pursuit. For instance,
• Wind Power - The United States is the Saudi Arabia of wind. We have the greatest resource available
for harvesting power from wind. Boston is becoming the Silicon Valley of wind.
• Drag Reduction - Goods are shipped, literally, around the world. The single biggest cost of this shipping
is overcoming the drag on the hulls of container ships. Reduce drag by 10%, reduce shipping costs by
10%.

• Gas turbines - You learned about these in thermo, but you didn’t really learn how they work. It’s fluid
mechanics, flow over a rotating foil. Jet engines, power generation, naval propulsion.
• KiteGen, and other kite based power generation. Most of the power of a wind turbine is generated at
the tip of the blade, so why bother with the rest. Put a kite on the end of a large string and have that
spin the generator. Google, Gates, etc.
• Modern electronics are not cooled with fans and heat sinks, they are cooled with little pulsating heat
pipes. Fluid flows and boils between the hot processor and the cool reservoir.
• Hard drives are far too small for ball bearings (or roller bearings). The platters ride on a thin cushion
of Helium. We may get to the famous Reynolds lubrication problem later this semester.
• Hearts, heart valves, aneurisms. Next time you talk to a cardiologist, ask them how much they know
about fluids. Here at UW ME, we use MRI data fo real people’s bodies, build silicon models of the
body part, then pump fake “blood” through it to analyze the flow, prevent clots, etc.
• Weather models rely heavily on crazy complicated geophysical fluid dynamics (GFD) models.
• Just about everything Elon Musk does
0.1 Some mechanics of this class
0.1.1 Grade calculation
Your grade in this class will be based on:
• 40% Friday Quizzes (every other Friday)
• 35% Homework - due on Wednesdays and Fridays
• 15% Final exam
• 5% Lecture Preview Quizzes
• 5% Participation
0.1.2 Participation
Your participation grade is somewhat subjective, but very, very important. The only way to really participate
is to be thinking about the material. The only way to really succeed in the class is to think about the material.
There are four ways earn your full credit in participation:

• Ask questions in class. If we don’t explain something clearly, ask. If you want to know how we did
something in class, ask.
• Answer questions we ask in class.
• Correct mistakes we make on the board. They happen, help us out.
• Come to office hours with careful, thoughtful questions about the homework or lecture material.
0.1.3 Homework Grading
We won’t grade every problem on every HW, we simply don’t have time. We will post solutions the HW,
however. In general, your homework will be graded on:
• Accuracy
• Organization
• Neatness
• E↵ort
0.1.4 Goals for this class
We hope that by the end of the semester, you have
• An intuitive understanding of
– internal flows
– external flows
– pumps and losses
– drag forces
– boundary layers
• And a basic understanding of the equations that govern
– single phase
– incompressible flow
• And an elementary understanding of
– di↵usion

0.2 What is a fluid?
A fluid is a substance that continuously deforms under a shear force.
Constant shear force -¿ Continuous deformation (as opposed to a sponge) Takes the shape of its container
(always? Under gravity?)
What are some examples of fluids?
• Water
• Air
• Oil
• Steam
• Hydrogen
• Blood
• Saliva
• Most of your body
(Newtonian and non-newtonian will come later)
0.3 Systems and Control Volumes
In Thermodynamics, we got a little sloppy regarding our definitions. We had a few di↵erent types of systems
we worked with:
• Fixed volume in space - nothing in, nothing out Thermo “closed system?
• Fixed chunk of mass - maybe expands, maybe compresses“Piston cylinder device?
• Fixed volume in space with stu↵ entering and leaving “steady state open system?
• Fixed volume in space with stu↵ entering or leaving “filling and draining problems”
More rigorous definition used in fluids System: Fixed identifiable quantity of mass (statics, dynamics,
toss a hammer)
Control volume: Arbitrary volume in space, bounded by a control surface
• Control surface may be imagined (a cube in the atmosphere)

• Control surface could be physical: e.g. the walls1 of a pipe
• Control surface may be in motion
• Control surface may be at rest
Fluid mechanics is a field within Continuum Mechanics
What is continuum mechanics? Continuum:
Smooth and continuous medium
Smooth: homogeneous (homogenized milk)
Continuous: Properties stay the same no matter how small the element (i.e. we can apply principles of
calculus)
We will ignore the fact that fluids are made of molecules (Hypersonic flight, rarified gases, orbit re-entry)
Mechanics: Obeys Newton F = ma Important to write this as a vector equation!!! How do we identify
vectors and scalars when writing by hand?
We will develop a very special case of F = ma where the forces come from two places:
• Pressure (just like in thermo, it’s the pressure you feel on your ears swimming to the bottom of the
diving well)
• Shear (just like in mechanics of materials, its related to a force, but causes deformation)
Viscosity
The two forces then, must result in an acceleration or Pressure and Shear must balance with acceleration
Thermo: Energy balance Fluids: Momentum balance
Write out momentum equation?
Units: SI units
1 “Wall” is a term that comes up in fluids a lot. While the room you’re in may have four walls, a wall in fluid mechanics is
any rigid surface that doesn’t move with the fluid.

1
Chapter 1
A Mathematics Review
Figure 1.1: Flow visualization of velocity field for a turbulent jet
Fluid mechanics is a mechanical engineering course. However, it requires a bit of mathematical rigor to
investigate phenomena related to fluids. In this section we will review the necessary mathematical tools
required for an introductory fluids course.
1.1 Scalars, Vectors & Tensors
1.1.1 Scalars
Scalars are quantities who can be described by a single number. The number can be an element of di↵erent
numerical fields but the most common scalar we will see in this course is that of the real numbers. If scalar
quantity c is an element of the real numbers we can write this as, c 2 R. Some examples of scalar quantities
that mechanical engineers often use are
• Temperature

2
• Pressure
• Speed
• Concentration
• Distance
All of these quantities have specific values at each location in space, the magnitude of which can be repre-
sented as a single number.
1.1.2 Vectors
Quantities that have both magnitude and a direction are called vectors. Some examples of vectors that a
mechanical engineer might use are:
• Velocity
• Acceleration
• Force
• Position
Vector quantities are labeled as bold. Since it’s hard to write a bold faced letter, a, by hand, people
often use an overarrow, !

a or an underline, a, to make it clear to the reader that it is a vector. Preference
on how you might designate your vector quantities is unimportant so long as it is consistent. It is important
to remember that vectors have a magnitude and a direction. Vectors do not have a location. You learned
this in Statics and Mechanics of Materials when you learned that a torque or moment vector can be moved
where ever you want it to be (as long as it is on the same body).
In three-dimensional space, vector quantities are represented as:
a = ai êi = a1 ê1 + a2 e2 + a3 e3 (1.1)
where êi are unit vectors, who together form the basis of a coordinate system.
1.1.3 Unit & Normal Vectors
Unit vectors are vectors with magnitude equal to 1. Any vector can be represented as its magnitude
multiplied by a unit vector. Take r, a position vector, for example. We can decompose r into the following:

3
r = rr̂ (1.2)
Furthermore, we can obtain the unit vector r̂ by dividing the vector r by its magnitude:
r
r̂ = (1.3)
|r|
Mathematicians often use the symbol êi to represent a generic unit vector.1 Engineers generally work in
three-space.2 Hence, we can define a set of unit vectors as ê1 , ê2 and ê3 . We can make it simpler yet! In
statics, we did a lot of work in rectangular Cartesian3 coordinates, so we could use êx , êy and êz . These are
unit vectors in the x-, y- and z-directions, respectively. But this is still slow, and uses a little more ink than
necessary, so we defined some nice new ones,
êx = î = x̂
êy = ĵ = ŷ
êz = k̂ = ẑ
You became very familiar with these and eventually took their meaning for granted. The vector i is the
unit vector in the x direction, and so on. In this course, we will work mostly in rectangular, Cartesian
3D-space.
It’s important that we recognize that. Engineers, especially mechanical engineers, often work with pipes.
Pipes are cylinders, so pipe problems are better solved in cylindrical coordinates with r, ✓ and z. The unit
vectors for this system are,
ê1 = êr = r̂
ê2 = ê✓ = ✓ˆ
ê3 = êz = ✓ˆ
Many of you have a weather app on your phone, which you check nonchalantly as you make plans for
the weekend. The models used to generate these forecasts are fabulously complex thermodynamic and fluid
mechanic systems. Since the earth is a sphere, the fluids are better modeled in spherical coordinates. The
1 The letter “e” often means that the vector is part of an orthogonal basis and the subscript “i” is an index keeping track of
the dimensions. The hat, ˆ, means its length is 1.

2 Most of the problems we work on are in a nice, simple, three-dimensional world.
3 Named for Renee Des Cartes.

4
coordinate system is comprised of,
ê1 = êr = r̂
ê2 = ê✓ = ✓ˆ
ê3 = ê = ˆ
Even a sphere is relatively easy compared to the blade on a wind turbine. The 90 meter turbine blades
have chords, cambers, thicknesses and twists that change along the entire length of the blade. We are forced,
for these problems, to use the Frenet4 frame consisting of tangential, normal and bi-normal unit vectors. In
this framework, the three orthogonal unit vectors are
ê1 = T̂
ê2 = N̂
ê3 = B̂
It is up to the problem solver to determine which coordinate system (or systems) work best for a given
situation. For this class want to focus on learning the fluids, not mastering di↵erent coordinate systems, so
we’ll stick with Cartesian and polar coordinates for the most part.
Normal vectors are unit vectors that are perpendicular (normal is another word for ?) to a line or
surface. By convention, the unit normal n̂ to a closed surface or a volume is an outward facing normal. Let’s
look at a cylindrical system. Being a clever mathematician, you choose to work in cylindrical coordinates
(r, ✓, z) with the ẑ axis passing through the center of both the top and bottom face of the cylinder. With
this construction there are 3 separate surfaces: top, bottom and side of the cylinder. The unit vectors for
the 3 surfaces are the following:
n̂top = [0 0 1] = (1)ẑ = ẑ
n̂bottom = [0 0 1] = ( 1)ẑ = ẑ
n̂side = [1 0 0] = (1)r̂ = r̂
Keep in mind, the unit vectors aren’t “in the center of the top”, “in the center of the bottom” and “on
the surface pointing outward.” Vectors don’t have a position, and unit vectors don’t even have a meaningful
magnitude. They simply define a direction. Unit vectors are sometimes called direction vectors.
4 It’s French, and hence pronounced, “fruh NAY.” Visit Dr. Arganthaël Berson for a correct pronunciation.

5
Figure 1.2: Normal vectors on a cylinder
1.1.4 Velocity
In general, we will use the letter u to represent velocity.5 . In rectangular Cartesian coordinates, we’ll use u,
v and w for the velocities in the x- y- and z directions, respectively.
We can break the general velocity vector, u, down into its three components,
u = ui + vj + wk.
1.2 Tensors
We’ve discussed two di↵erent types of quantities so far:
• Scalar - magnitude but no direction. simply a number.
• Vector - magnitude with one direction

5 It would make more sense to use v, but since we live in 3-space, we need three letters. The letters u, v and w fit together
and make a lot more sense than vw and x

6
Would we ever need to represent a magnitude and two directions? Yes. For instance, the stress on a
cube, as shown in Figure 1.3. There are three directions in which a force could be acting, and another three
directions to define the surface on which it is acting. To cover all these combinations, we need a quantity
with 3 ⇥ 3 values. We call this a tensor.
Figure 1.3: A cubical volumetric element experiences stress on three planes. To fully describe the stress,
three faces are identified by their normal vectors, and each face has stress in two directions.
The stress tensor from our cube above looks like this,
2 3
6 11 12 13 7
6 7
=6
6 21 22
7
33 7 . (1.4)
4 5
31 32 33
Notice that to show the stress is second order tensor, it gets underlined twice. Even in print. Let’s
examine the physical interpretation by looking at the subscript indexes:
• First index - specifies the orientation of the surface
• Second index - specifies the direction of the stress
Let’s revisit our list from above and revise the definitions slightly,
• Scalar - magnitude but zero direction, zeroth order tensor
• Vector - magnitudes with one direction, first order tensor
• Tensor - magnitudes with two directions, second order tensor
Get comfortable using these quantities.

7
1.3 Vector Calculus pt.1
1.3.1 Gradient of scalar
For a scalar function f , the gradient6 is as follows:
@f @f @f
rf = î + ĵ + k̂ (1.5)
@x @y @z
First of all, what is a scalar function? Imagine that you share an office with a few other people. One of your
office mates regularly eats a can sardines in mustard sauce as a snack, then leaves the empty can sitting
somewhere in the office. You can think of smell, S, as a scalar quantity. Smell doesn’t have a direction, it
just has a magnitude.
When we consider rS, what are we really after? Imagine sniffing around the room to find the can. You
tip your head up, sni↵ repeatedly, and move your head around the room. You’re really performing rS in
7
your nose! You could easily set up a Cartesian coordinate system in your office. Your desk is the origin
North is positive y, East is positive x and up is positive z. As you are moving your head around sniffing,
you are trying to find how the intensity of the smell increases in di↵erent directions. Ultimately, you want
to put together a vector that points toward the stinky can.
The smell of sardines in your office, a scalar, S, changes in the x, y and z directions. Hence, the vector
that results must have three components. Following the vector, rS, will lead you to the o↵ending stinky
sardine can.
1.3.2 Gradient of a vector field
Now let’s make the example a little more complicated. Rather than a scalar field, let’s say we’re defining
velocity of the air in the same room, u, as a function of space. You could define
u = f (x, y, z).
So what does the gradient of this velocity field really mean? Instead of sniffing around the room, you’re going
to hold your hand up and feel the air moving. At any given location, the air has three velocity components,
ux , uy and uz . Of course, in the interest of time, we would usually call these three velocity components u,
6 Mathematicians would pronounce this grad f
7 If you’re not the center of the universe, you can at least be the center of your office.

8
v and w, respectively.
Now as you “feel” around the room with your hand, you may start by moving your hand East and West,
trying to get a feel for @u/@x. There’s a little complication here, though. As you feel East West, there are
three components @u/@x, @v/@x and @w/@x. You have to do this for North-South and Up-Down, so the
result of your “feeling around the room” will be
2 3
6@/@x7
6 7
ru = 6 7
6@/@y 7 u v w
4 5
@/@z
2 3
6 @u/@x @v/@x @w/@x 7
6 7
ru = 6
6 @u/@y @v/@y @w/@y 7
7,
4 5
@u/@z @v/@z @w/@z
a second order tensor.
In summary, the gradient of a scalar field is a vector, the gradient of a vector field is a tensor. This goes
on and on, but in our 3-D world, we don’t need to worry about it.
1.4 Vector & Matrix Operations
1.4.1 Dot Product
The dot product is also called the inner product or the scalar product. In general, it answers the question,
“How much of this stu↵8 is acting in this other direction?” or “How parallel are these two directions? ” The
trigonometric function most closely related to the dot product is cosine.
The definition9 of the dot product of two vectors is
a · b = kakkbk cos ✓ = a1 b1 + a2 b2 + a3 b3 (1.6)
You will notice that the result of this operation is a scalar because a1 b1 , a2 b2 and a3 b3 are all scalars.
In the process of determining the magnitude of one vector in the direction of another, we have lost all our
directional information. To gain our direction back, we would have to then multiply the scalar result by the
unit vector in which we are interested.

8 “Stu↵” could be mass flow, momentum, force, distance, etc.
9 You
q the norm of a vector, kak. The double lines around
may notice that the formal definition of a dot product involves
the vector indicate that it is actually the 2-norm, defined by kak = a21 + a22 + a23 in 3-space. In our simple 3-dimensional
approach to things, the Euclidean 2-norm, kak, and the magnitude, |a|, are interchangeable.

9
Example: Consider the four vectors

a = h 12 5
13 , 13 i
b = h 45 , 35 i
3 4
c=h 5 , 5 i
4 3
d=h 5 , 5i
1. Find the magnitudes of all four vectors.
2. Sketch three vector pairs: a & b, a & c and a & a
3. Observing the sketches, estimate the dot product of all three vector pairs.
4. Find a · b, a · c and a · d.
5. Comment on the results.
Solution:
1. Since 3-4-5 and 5-12-13 are both Pythagorean triples, so we know that the magnitude of all four vectors
will be exactly 1. In other words, they’re all unit vectors.

10
2. Since we need to dot the three pairs of vectors, we’ll draw the three pairs, rather than just the four
vectors.
Figure 1.4: Three pairs of vectors: mostly aligned, mostly perpendicular and mostly opposite.
3. When we go through the motions to dot these, we get
a · b = 0.9692
a·c= 0.2462
a·d= 0.9692
4. For the two vectors that are mostly aligned, we keep most of the magnitudes. For the two vectors that
are mostly orthogonal, we lose most of the magnitude. For the two vectors that are mostly opposite,
we keep most of the magnitudes, but it is negative!

11
Example: Let’s say Hafthor Julius Bjornsson10 is puling a 200,000 kg train car with a chain held in his
teeth. The tension in the chain is
C= 500i + 866j [kN ]
and the tracks run straight North-South. The track and wheel bearings are frictionless.11 What is the
acceleration vector of the train car?
Figure 1.5: Schematic of a huge Icelandic guy pulling a rail car with a chain. Vector representation of the
same. Unit vector representing the direction of the track. Projection of chain tension onto track.
To simply use Newtons second law,
F = ma,
would give us bad results. Since we know the train car can only move in the same direction of the tracks,
we are interested in the direction vector, T̂ , or the unit vector of the tracks, j. The only force that we are
interested in is the force of the chain acting in the direction of the track. We use the dot product to do this.
We need to calculate
C · T̂ = ma
Wait, wait, wait.... We can’t do that. Notice that the LHS of the above equation is a scalar and the RHS
of the equation is a vector. We can never, never, never equate a scalar and a vector, a vector and a tensor,
10 Not only a big deal in the World’s Strongest Man competition, but also played The Mountain in Game of Thrones.
11 Of course they’re frictionless.

12
or anything like that.
What we really need in this case is to get the projection of the chain vector, C, on to the track vector,
T. The projection is12 !

C·T
prT C = 2 T (1.7)
kTk
If we want to use Newtson’s second law, F = ma, we need just the component of the chain force, C, that
is acting in the direction along the tracks. The dot product operation gives us the magnitude, the track
vector gives us the direction, and this information, magnitude and direction, is enough for us to define the
acceleration vector.
The projection of the Chain, C, onto the Track, T is simply the vertical, or North-South component,
prT C = 866ĵ [kN].
Solving for acceleration, a, we get13
F 866ĵ [kN] 2
a= = = 0.004 km/s
m 200, 000 [kg]
Example: Combining vector operations
Calculate r · g with g = (g1 , g2 , g3 ) and = z2.
⇣ ⌘
@ @ @
r ·g = @x , @y , @z · (g1 , g2 , g3 )
@ @ @
= @x g1 + @y g2 + @z g3
= 2zg3
1.4.2 Cross Product
While the dot product of two vectors results in a scalar, the cross product results in a vector14 . Spatially
speaking, the cross product of two vectors results in third vector that is a) perpendicular to the first two
vectors, b) in the direction defined by the right hand rule and c) with a magnitude equal to the area of the
parallelogram defined by the two vectors.
Mathematically, the cross product is

12 A bit of algebra is required to get from the definition of vectors and dot product to the expression below. It is useful,
though, and will be assigned as HW.

13 For starters, we get a vector on each side, which is nice.
14 The cross product is also called the vector product

13
Roman’s definition
î ĵ k̂
a ⇥ b = |a||b| sin ✓n̂ = a1 a2 a3 = (a2 b3 a3 b2 )î (a1 b3 a3 b1 )ĵ + (a1 b2 a2 b1 )k̂ (1.8)
b1 b2 b3
Dressler’s definition
î ĵ k̂
a2 a3 a1 a3 a1 a2
a⇥b= a1 a2 a3 = î ĵ + k̂. (1.9)
b2 b3 b1 b3 b1 b2
b1 b2 b3
Another way to think about the cross product is that it results in a vector, c, perpendicular to the first two,
a⇥b=c
15
So the direction of the result, ĉ , is the cross product of the two unit vectors
ĉ = â ⇥ b̂,
which you can get with just the right hand rule. The magnitude of c is the area of the parallelogram16 ,
kck = kakkbk sin(✓)
As seen in Figure 1.6, the cross product cannot commute and follow the right hand rule. Hence,
u⇥v = v⇥u (1.10)
We can work this out using the formal definition as well. The first case, i ⇥ j, is given by Equation 1.11,
15 ĉ
in this definition is equal to the normal n̂ in Roman’s definition
16 Think about this: the area of a rectangle is A = lw. You could also write it as A = lw(sin(✓)), but since ✓ = ⇡/2, sin(✓) = 1,
so A = lw(1) = lw. As ✓ increases or decreases away from ⇡/2, sin(✓) decreases, hence decreasing the area of the parallelogram.
For the grease monkeys reading this, consider using a wrench. You want to maximize your torque, T = r ⇥ F, where r is the
wrench and F is your hand. To ensure maximum torque, you make sure that ✓ = ⇡/2, i.e. your hand is perpendicular to the
wrench.

14
Figure 1.6: Cross product does not commute. a ⇥ b has the same magnitude as b ⇥ a, but according to the
right hand rule, the resulting vectors point in opposite directions.
i j k
0 0 1 0 1 0
i⇥j= 1 0 0 = i j+ k, (1.11)
1 0 0 0 0 1
0 1 0
whereas the second case, j ⇥ i, is given by Equation 1.12,
i j k
1 0 0 0 0 1
j⇥i= 0 1 0 = i j+ k, (1.12)
0 0 1 0 1 0
1 0 0
1.4.3 Matrix Multiplication
Let = ij be a 3x3 matrix where i & j indicate the matrix row and column respectively. Let n̂ be a vector.
2 3 2 3 2 3
6 11 12 n
13 7 6 1 7 6 11 n1 + 12 n2 + n
13 3 7
6 7 6 7 6 7
· n̂ = 6
6 21 22
7 · 6n 7 = 6
23 7 6 2 7 6
7
21 n1 + 22 n2 + 23 n3 7
4 5 4 5 4 5
31 32 33 n3 31 n1 + 32 n2 + 33 n3

15
2 3
 6 11 12 13 7 
6 7
n̂· = n1 n2 n3 6
· 6 21 22
7
23 7 = 11 n1 + 21 n2 + 31 n3 12 n1 + 22 n2 + 32 n3 13 n1 + 23 n2 + 33 n3
4 5
31 32 33
Therefore · n̂ = n̂ · if and only if is symmetric17 . Symmetric Matrices are matrices who are equal
to their own transpose. If ij = ji then
n̂ · = · · n̂
17 An applied mathematician and a pure mathematician may quarrel over this definition, but for the sake of this class column
vectors and row vectors can be interchanged with one another. That is
2 3
a1 ⇥ ⇤
a = 4a2 5 = a1 a2 a3
a3

16
1.5 Vector Calculus Part 2: Integrals, Fluxes and Integrated Fluxes
1.5.1 Integrals
An important question to start o↵ with is: What is an integral?

Z
The integral sign, , looks like an S for a reason. An integral is not just an anti-derivative. The
anti-derivative is a tool we use to evaluate the integral, via the Fundamental Theorem of Calculus.
We will look at one example of using an integral to determine a quantity of interest. It’s something you
probably did in Calc 1, but maybe didn’t fully understand.18
Example: Find the volume of the solid generated by revolving the area bounded by y = 9 x2 and the
horizontal axis about the y-axis.
First, let’s make a sketch of this solid. Let’s sketch what this would look like.
DRAWING - Paraboloid
So how do we find the volume. We don’t have a formula for the volume of a paraboloid, but we do have
a formula for the volume of a box. Let’s draw a box, but wrapped around into a shell.
DRAWING - Thin box and shell
We know the volume of a box is
Vbox = LHW,
where V is volume, L is length, H is the height. If our goal is to represent the volume of our paraboloid
with a bunch of these boxes, we should make it fit nicely.
DRAWING - Shell in paraboloid
The length, in this case, becomes 2⇡r, the height is f (x) and the width (thickness) is actually pretty
thin. Since we’ve wrapped it around, we could call it dr. This isn’t any particular shell, just an arbitrary
shell that’s inside of our solid. It’s very important that we look at the units as we consider of the volume
18 Or even worse, you had a formula on your note sheet and you plugged in the numbers.

17
of this one arbitrary shell.
Vshell = LHW = |{z}

2⇡r f (x) |{z}
dr
|{z}
[m] [m] [m]
Before we even consider integrating anything, we need to know what we are integrating.19 We need to to
know the size, shape and units involved for the individual element before we can possibly dream of integrating
it. We are trying to get a volume, we have an elemental volume, and we have the units (2⇡r [m], f (x) [m]
and dr [m]) to get the units we want, m3 . Now, and only now, can we integrate,
Z 3
Vsolid = 2⇡r(9 r2 ) dr
0 | {z }
|{z} some volumes of shells
add up
Z 3
= 2⇡ 9r r3 dr
0
 3
9r2 r4 81
= 2⇡ = 2⇡ .
2 4 0 4
1.5.2 Flux
Flux is the rate at which some quantity flows through a surface. Some common flux quantities and their
units are:

kg
• Mass flux
s - m2
 
J W
• Energy flux =
s-m2 m2
  
kg - m/s kg - m 1 N
• Momentum flux = = 2 = [Pa]
s-m2 s2 m2 m
The flux quantities used in this course will be vector quantities, such as F, and will therefore possess
both magnitude and direction20 . To calculate the total amount, , of some flux quantity F passing through
an object, we take the projection of flux quantity onto the surface unit normal vector and integrate over the
entire surface area. In maths this is,
Z Z
= F · n̂dA (1.13)
A
where is the total amount of F that passes through the surface.

19 Anything else would be like trying to add up your bill at a restaurant without knowing what anybody ordered.
20 This should make sense. If something is said to ’flow’ obvious questions would be,Where to and How much?

18
Let’s consider an easy example first. A popular activity in Northern Wisconsin is smelting 21 . Figure 1.7
shows a typical smelt in Lake Superior. When the tributaries of Lake Superior warm up to about 42 F, the
smelt run 22 , in large schools, upstream to spawn.
Figure 1.7: Smelt, typical of Lake Superior.
To picture what this looks like, we could define a smelt flux,23

smelt
S = 1.414î 1.414ĵ .
s - m2
What is the magnitude, with units, of the smelt net flux vector, S? To actually picture what this looks
like, we will sketch the swimming smelt.
DRAWING - smelt swimming SE
24
In order to catch the smelt, people use nets, like that shown in our sketch
DRAWING - smelt net
To estimate the rate at which we catch smelt, we simply multiply the smelt flux by the area of our net,
right? You should be netting a couple smelt per second.
Now let’s say we place the net in the water and wait as dinner swims haphazardly in. We place the net
21 Catching small fish. Not to be confused with extracting metal from its ore
22 They don’t really run, they swim.
23 For instance, the entrance to Reefer Creek, about half way between Superior (the city) and the Apostle Islands.
24 Only in our dreams would the Northern Wisconsinites use metric nets.

19
in the water such that the surface normal of the net opening is
7 24
n̂net = i + j,
25 25
as shown in the next sketch.
DRAWING - smelt net in water
Do you see a problem here? Sure you do, because in deep down in your primordial hunter-gatherer mind,
you already know about fluxes and normal vectors. The rate at which smelt enter your net, d⌃/dt, is given
by
 
d⌃ smelt smelt
=S Anet [m2 ] · n̂net [-]
dt s s - m2
The relationship between the flux and the normal really hurts us here, ]
⌧
7 24
h1.414, 1.414i · , = .962
25 25
This is going from bad to worse. When I multiply the dot product by the area of the net, I wind up with
-0.962 smelt per second. So fish are actually leaving my net?
No. Remember that the normal vector points outward. A negative dot product with a normal vector
actually results in an inward flux. This will be very important later.

20
Let’s look at another example, with a little more mathematical rigor.25 The star at the center of our
solar system provides nearly all of the thermal energy to the 8 planets26 orbiting it. The energy leaves the
surface of the Sun and travels outward in the form of radiation. The radiative energy that leaves the surface
can be written in vector form as,
ESun = 6.3 ⇤ 107 r̂ [J/(m2 s)] = Er̂
Since all the energy comes from the fusion process at the center of the Sun, it is expelled from the Sun
in all directions in equal magnitude27 and therefore has the direction of the outward radial vector. This is
illustrated in Figure 1.8.
Figure 1.8: The Sun imaged by NASA (left) and a visualization of energy flux vectors (right)
Let’s take a step back and look at how spherical coordinates are set up.
The bounds for spherical coordinates are,
0rR
0  ✓  2⇡
0 ⇡
Figure 1.9: Typical spherical coordinates

25 Guess which instructor wrote which example.
26 Sorry, Pluto. You’re out. No! You’re back in! Wait, no. You’re out.
27 Ignoring solar flares and other irregularities

21
To find what the surface element, dS, it is easiest to start with the volume element for spherical coordi-
nates dV . The volume obtained by the triple integral in spherical coordinates is,
ZZZ Z Z Z
V = dV = r2 sin( )drd d✓
So for spherical coordinates dV = r2 sin( )drd d✓. If we divide the volume element by the ”thickness”
of the element, dr then we can get the surface element, dS.
dS = dV /dr = r2 sin( )d d✓
So now if we want to integrate to obtain surface area on the sphere, we let r = R and then we do it as,
Z Z Z 2⇡ Z ⇡
SA = dS = R2 sin( )d d✓ = 4⇡R2
0 0
Let’s now calculate the total energy output of the sun, Sun , by integrating the heat flux over the entire
surface area of the the star. Let R be the radius of the Sun, which is R = 7 ⇤ 108 m. The integral becomes,
Z Z Z 2⇡ Z ⇡
Sun = ESun · n̂dA = ESun · n̂R2 sin( )d d✓
A 0 0
Z 2⇡ Z ⇡ Z 2⇡ Z ⇡
2 2
= ER sin( )d d✓ = ER sin( )d d✓
0 0 0 0
Z ⇡ ⇡
= E2⇡R2 sin( )d = E2⇡R2 [ cos( ) ]
0 0
Sun = E4⇡R2 = 3.8 ⇤ 1026 J/s
For our case, the direction of the energy flux outward from the center of the Sun is the same direction as
the outward normal vector r̂. So the dot product becomes simply E · r̂ = E, and the total output becomes
the solar flux from the surface multiplied by the surface area of the sun, E4⇡R2 .

22
1.5.3 Divergence
Consider the velocity field we discussed with the gradient. We’ll call it
u = u(x)
First of all, what does this notation mean?
The gradient operator works on a scalar function or a vector function, but the divergence and curl
operators require a vector function. The divergence of a vector is the dot product of gradient operator and
the vector itself. The dot product between two vectors results in a scalar quantity. If a = a(x) and b = b(x)
(both a and b are functions of space) then,
✓ ◆
@ @ @ @a1 @a2 @a3
r·a= , , · (a1 , a2 , a3 ) = + + (1.14)
@x @y @z @x @y @z
The sum of these three terms will be some scalar, c 2 R.

@a @a
If a were not a function of space, the value of a wouldn’t change as we move around. Hence, @x , @y and
@a
@z would all be zero. The result of the r · a would be the scalar quantity 0.
The divergence of vector field a represents the flux of the vector field from an infinitesimal volume
(dxdydz) around a given point (x, y, z).
Imagine were heating some air. If we consider the velocity of air that is being heated, in the regions where
the air is warming it is also expanding in all directions. This means that the velocity vectors are expanding
outward from the heated regions. Therefore the divergence result in a positive scalar value. In other words,
since the divergence is positive, there is some net flow outward from whatever spot we are studying.
If we change the direction of energy transfer and cool the air we could get the opposite result. The density
of the air will increase and therefore the velocity vectors will be pointing inward to the cooling source giving
a negative value for divergence.28
1.5.4 Curl
The curl of a vector is the result of taking the gradient operator crossed with another vector. The curl
represents the spin or rotation of a vector field, and at every point in the field there is a curl vector that
characterizes the rotation. The curl is computed the same way as a normal cross product, except the
28 Sometimes the vector field is a function of space, i.e. it varies as I move around the room, but yet the divergence, r · a,
is still zero. In this case, we call the field divergence-free or solenoidal. The term solenoidal would be better o↵ as pipe like,
because it is bounded by a pipe. Nothing gets out, nothing gets in. It’s not allowed to diverge.

23
di↵erential operators that make up the gradient will operate on the vector components:
2 3 2 3
6 î ĵ k̂ 7 6 @b
@y
3 @b2
@z 7
6 7 6 7
r⇥b=6 @
6 @x
@
@y
@ 7=6
@z 7 6 (
@b3 @b1 7
@z )7
(1.15)
4 5 4 @x 5
@b2 @b1
b1 b2 b3 @x @y
The curl of velocity is called the vorticity (!) of a flow29 . Vorticity vectors represent the rotation of
di↵erential fluid elements within the flow field. The vorticity is calculated as,
2 3
6 î ĵ k̂ 7
6 7
! =r⇥u=6 @
6 @x
@
@y
@ 7.
@z 7
(1.16)
4 5
u v w
If we work this out, we find that the curl, a vector, is
✓ ◆ ✓ ◆ ✓ ◆
@w @v @u @w @v @u
! =r⇥b= î + ĵ + k̂
@y @z @z @x @x @y
Flows are often characterized as either having vorticity or not. Flows with zero vorticity at all points in
the flow field are called irrotational flows. And for flows with non-zero vorticity we give them the proper
title of rotational flows.
Imagine a group of four tubes floating down a river. The river is straight, steady and of uniform depth...
1.5.5 Surfaces, Volumes & The Divergence Theorem
The integral of a function, f , over an area can be written as,
ZZ ZZ Z
f dx dy = f dA
|{z} = f dA
A | {z } A
a square area a square area | @V {z }
a potato
If the region of interest in represented by @V this means a surface of the volume element V .
The integral of a function, f , over a volume can be written as,
ZZZ ZZZ Z
f dx dy dz = f dV = f dV
V V V
29 Yetanother word that starts with the letter “v”. This part of the alphabet is getting awfully crowded, so we have chosen
the Greek letter omega, !, which looks like a w but is actually analogous to our letter o, which looks like a zero.... Sorry about
that.

24
Remember you can change the order of integration just as you can change the order of di↵erentiation when
working with multiple operators.
The divergence theorem is the following:
Z Z
r · F dV = F · n̂ dS. (1.17)
V @V
This equation states that the divergence of F from the volume V , is exactly equal to the outward flux of F
normal to and through the entire surface @V . In 2-D this works as well. If we take a close curve around
an area A, the equation becomes

Z Z
r · F dA = F · n̂ d . (1.18)
A
This means that the divergence of F from area A is equal to the flux of F through and normal to the closed
curve .

25
Useful Vector Identities

u+v =v+u
u·v =v·u
(u + v) · w =u·w+v·w
u⇥v = v⇥u
(u + v) ⇥ w =u⇥w+v⇥w
Cylindrical Coordinates
@f 1 @f ˆ @f
rf = x̂ + ✓+ ẑ
@x r @✓ @z
1 @rur 1 @u✓ @uz
r·u= + +
r @r r @✓ @z
2 3
6 r̂ r✓ˆ ẑ 7
6 7
r⇥u=6 @
6 @r
@
@✓
@ 7
@z 7
4 5
ur ru✓ uz

26
Parabaloid problem from lecture This problem uses what is called the ‘Shell Integration’ technique.
We are give a function, f (x) = 9 x2 . The formula for revolving around the y-axis is the following,
Z 2⇡ Z b
V = xf (x)dxd✓
0 a
When thinking of revolution around an axis it is easiest to think in terms of radius instead x. So let us
plug in our function and swap out x for r. For the limits of integration we will go from the smallest radius,
0, to the largest radius, 3.
Z 2⇡ Z 3
V = rf (r)drd✓
0 0
Now perform the integral.
Z 2⇡ Z 3
V = r(9 r2 )drd✓
0 0
Z 3
= 2⇡ (9r r3 )dr
0
 3
9r2 r4
= 2⇡
2 4 0
81
V = 2⇡
4

27
Chapter 2
Fluid Statics
Figure 2.1: Shipwreck in Lake Michigan
2.1 Pressure
Pressure is defined as a force applied over some area perpendicular to the surface.
F
P = . (2.1)
A
N
The units are m2 , also defined as a Pascal (Pa). You may also remember the Ideal Gas Law from
chemistry class which looks something like
P = ⇢RT. (2.2)

28
Atmospheric pressure is the pressure felt within the atmosphere of the planet. This pressure is due to
the weight of the molecules within the atmosphere. The absolute pressure is the real or total pressure.
At sea level the absolute pressure of the atmosphere is
Patm = 1 atm = 101.325 kPa ⇠ 14.7 Psi
Gauge pressure

29
2.2 Statics in general
Fluid statics (often called hyrdostatics) is the branch of fluid mechanics that concerns fluids at rest. Fluid
motion relies on Newton’s Laws and therefore for fluid to be static the sum of forces must be exactly zero
in all directions.
As a Mechanical Engineer, you are very often interested in Newton’s Second Law1 ,
F = ma.
You took two two semesters to learn the basics, however. The first was called Statics. Why Statics?
Because you studied only static situations, that is, there were no time derivatives. Since acceleration is a
time derivative, you had to eliminate it. Hence, the “static” equation,
F = 0,
which you then broke down a little further into
X X X
Fx = 0, Fy = 0 and Fz = 0.
Once you developed some tools to work with these simplified situations, you were able to move on to
Dynamics, and once again address time derivatives.2 We’ll do the same in this course.
2.3 A Fluid Element at Rest
There are two general categories of forces that can act on a fluid, body forces and surface forces. A body
force is one which acts on every particle within an element of the fluid, no matter how large or small the
element. Gravity is one example of a body force.
Magnetic and centrifugal3 are two other examples of body forces. To keep things simple, we’ll skip
magnetohydrodynamics and keep things small enough that Coriolis accelerations are negligible, leaving
gravity as the only body force of concern.
Surface forces are the other relevant ones. Again, there are two surface forces: shear and pressure. We’ve
already discussed that a fluid, by definition, continuously deforms under shear. If there is deformation, it’s
1 Ifyou can’t remember which of Newton’s laws is which, don’t worry. You’re not alone.
2 There is at least one person in Mechanical Engineering who believes ME 361 - Thermodynamics should be renamed
Comparative Thermostatics, since there aren’t ever any time derivatives in the course.
3 Of course, centrifugal forces don’t exist in an inertial reference frame. The reader is challenged to come up with a reference
frame that isn’t rotating.

30
not a static situation. So our study of Fluid Statics is limited to pressure as the only surface force.
We are now going to look at the very limited situation in which the only two forces are
• Gravity
• Pressure
We will further limit our discussion to coordinate systems in which gravity acts vertically down and is
constant. Neither of these are completely true, but close enough for our purposes. We also need to make an
important definition about pressure. Just like in your Mechanics of Materials class, we will define a positive
pressure as a compressive normal stress.

31
Figure 2.2 shows a cubical fluid element experiencing pressure forces on both x-surfaces. The key thin to
4
notice is the pressure is di↵erent on each side. The pressure varies by @P/@xdx from one side of the other.
Figure 2.2: A cubical fluid element experiencing a pressure acting on both x-surfaces. The pressures di↵er
by @P/@x dx
Newton’s law doesn’t reference pressures. We need forces. To determine the force, we need to apply the
pressure, P , to the area of that face, dx dy. From Statics class, we will determine the net surface force on
the element.
X ✓ ◆
@P
Fx,s = P dy dz P+ dx dy dz (2.3)
| {z } @x
positive force on left face | {z }
negative force on right face
@P
= dx dy dz (2.4)
@x
It’s a little counterintuitive, but the force goes against the pressure gradient. Think about this, and make
sure that it actually makes sense.
We could rewrite Equation 2 as a vector equation,
✓ ◆
@P
Fx,s = dx dy dz î.
@x
4 To be very thorough, we would have to fully expend the Taylor series. We will work with the assumption that the linear
approximation of the pressure is close enough.

32
Applying the same process to the y- and z-directions, we would get the full surface force,
✓ ◆
@P @P @P
Fs = î + î + î dx dy dz (2.5)
@x @x @x
| {z }
we have another name for this
Fs = rP dx dy dz (2.6)
Fs = rP dV (2.7)
We can’t have a net force, though, since this is the fluid statics section. What else is there? The body
force, of course. Gravity! The gravitational force is the mass times the gravity,
Fg = g dm
= g ⇢ dV.
Now, to really achieve fluid statics, we need to make sure there is no net force, so
rP dV + g ⇢ dV = 0.
We can divide through by the dV to get the hydrostatic vector equation,
rP + ⇢g = 0 (2.8)
There is great value in picking convenient coordinate systems. We generally pick ẑ or k̂ as straight up.
But what exactly is up? The direction that goes against gravity. With this definition, we can break Equation
2.8 into its three Cartesian coordinates,
@P @P @P
= 0, = 0, = ⇢g.
@x @y @z
The only real interesting one is the third one. Since it only deals with gradients in the vertical direction,
they’re not even partial derivatives anymore. The result is the scalar hydrostatic equation
dP
= ⇢g (2.9)
dz

33
2.3.1 Manometers
For most of the semester, we’ll consider density to be a constant. This allows us to rearrange Equation 2.3.2
and integrate it from some reference pressure at some reference elevation to the pressure of interest at the
elevation of interest.
dP
= ⇢g
dz
dP = ⇢g dz
Z P Z z
dP = ⇢g dz
P0 z0
P P0 = ⇢g(z z0 )
P P0 = ⇢g(z0 z)
The description of interest is a little vague. When you swim to the bottom of a swimming pool, your
ears hurt. That’s interesting. So we could take the surface of the pool as z0 and the bottom as z. To make
things a little easier, we can define a new variable, h, as the height of the column of water above you,
h = z0 z.
Finally, we come to the equation you may have “learned”5 in High School physics,
P = ⇢gh , (2.10)
where positive h is measured downward from the surface.
2.3.2 Pressure Distribution
For the most simple example of this just take some rectangular water reservoir. The water in the reservoir
is static. The pressure at the surface of the water is atmospheric pressure. The triangle and 3 lines above
and below the surface in Figure 2.3 are a commonly used symbol to designate the region as a free surface.
A free surface is liquid surface that has no shear stress applied parallel to the surface. But to make things
more convenient we can talk in terms of gauge pressure. So at the surface of the interface between the water
and air the gauge pressure is zero. At any position under the water surface there is also the force of gravity
due to the water above it pressing down. As depth increases the pressure increases in a linear fashion. Each
5 Read: Mindlessly memorized and puked back up on the test.

34
Figure 2.3: Pressure on point within the water is felt equally from all directions
horizontal slice is at the exact same pressure and therefore every particle at the same depth is at the same
pressure. The pressure felt at depth h is
P = ⇢gh (2.11)
where ⇢ is the density of water, g is the acceleration due to gravity (g = 9.81m/s2 6 ).
6 The acceleration due to gravity actually does vary slightly across the earth from 9.76-9.83 m/s2 .

35
Example: Fluid column pressure
Figure 2.4: Fluid at the same depth, but di↵erent columns above
Compare the pressures at locations A,B,C,D and E in Figure 2.4?
Solution:
They are all equal.
Example: James Cameron dives to the Titanic and beyond
Figure 2.5: James Cameron traveling in his submarine to the Titanic and Mariana Trench
When James Cameron7 decided to make the hit film Titanic, he was determined to see the sunken ship
at the bottom of the ocean with his own eyes. So James and a small crew climbed inside a submarine and
7 Famous and talented producer/director/writer

36
made way for the ocean floor. While at the site of the titanic, James becomes fascinated with the deep ocean
and decides he wants to go even deeper in the ocean. In fact, he chooses to travel to the deepest part of the
ocean, called the Mariana Trench. An illustration of James voyage can be seen in Figure 2.5. James wants
to record the exact depth of each location he visits. He also happens to be passionate about fluid mechanics
and knows that if he measures pressure with a sensor on the outside of his submarine he can figure out deep
the submarine is. The pressure sensor provides gauge pressure Pmeas measurements in units of MPa. The
density of the water in the ocean is ⇢ = 1027 [kg/m3 ]. Calculate the depths reached by James Cameron8 .
• Site of Titanic - Pmeas = 38.2 [MPa]
• Mariana Trench - Pmeas = 105.9 [MPa]
Solution:
With the gauge pressure reading from the sensor we apply Equation . Isolate the depth z of the submarine
on one side of the equal sign convert necessary units, and plug in known quantities.
P = Pmeas 0 = Pg = ⇢gz
Pmeas
z=
⇢g
• Site of Titanic - z = 3.8 [km]
• Mariana Trench - z = 10.5 [km]
8 Famous and talented producer/director/writer/submarineoperator. His deep sea experience led to the bioluminescence
featured in Avatar.

37
Example: Collapsable Water Trough
Figure 2.6: Collapsable water trough. Two plates, hinged at the bottom and held together by a cable.

38
More work

39
2.4 Buoyancy
Modern aircraft carriers, like the one seen in Figure 2.7, are true feats of engineering. The USS Gerald R.
Ford is the newest addition to the U.S. Navy’s fleet and is expected to hit the seas by 2019. The ship will
travel at 30 knots (35 mph) and is capable of carrying over 75 aircraft each with masses around 10,000 kg.
The ship will also carry around 2,600 sailors, 2 nuclear reactors, 3 workout gyms, grocery stores and perhaps
even a Starbucks 9 , totaling almost 100,000 tons in weight. So what kind of sorcery keeps these things afloat?
Figure 2.7: USS Gerald R. Ford - U.S. Navy
Of course it is not magic but rather science keeping the ships above water. Buoyancy is the upward
force exerted by a fluid in the opposite direction of the gravitational force. Objects with density higher than
that of the fluid tend to sink, and those with density less than that of the fluid tend so stay afloat. In this
case the density of the fluid, ⇢w is that of sea water.
Perhaps before tackling an aircraft carrier, let’s take a look at something a little easier. Let’s consider a
cylinder, such as that shown in Figure 2.8.
We know from our study of hydrostatics that there will be some pressure in the water. The pressure on
the top and bottom surfaces are
Ptop = ⇢ g htop (2.12)
Pbottom = ⇢ g hbottom . (2.13)

9 There is a Starbucks on both the USS Carl Vinson and USS Theodore Roosevelt

40
Figure 2.8: A cylinder with radius, r, and length, L. The left cylinder is just floating in space. But the right
cylinder is floating in a tank of water.
The forces on the top and bottom surfaces are then
Ftop = ⇢ g htop ⇡r2 (2.14)
Fbottom = ⇢ g hbottom ⇡r2 . (2.15)
From statics, we know that we are interested in the total force on the object,
Fz,buoyant = Fbottom Ftop
= ⇢ g hbottom ⇡r2 ⇢ g htop ⇡r2
= ⇢ g ⇡r2 (hbottom htop )
= ⇢ g ⇡r2 (L)
= ⇢gV
This gives us the interesting result that the net buoyant force on the object is equal to the weight10 of the
water that is displaced by the object.
In the authors’ experience, most cylinders have some weight of their own. So to get the true net force
10 It is very important to note that weight is a force.

41
on the cylinder, we need to consider this weight.
Fnet = Fbuoyant Fweight
= ⇢gV mobject g
= ⇢f luid g Vobject ⇢object Vobject g
= g V (⇢f luid ⇢object )
This is an even more interesting result. If ⇢fluid > ⇢object , the net force is positive, so the object will tend
to accelerate upward. If ⇢object > ⇢liquid , the net force is negative, so the object will tend to accelerate
downward.11
For an object that is partially submerged, the fluid exerts an upward force equal to the weight of the
displaced fluid upon the object. In maths that is,
FB = ⇢w gVdisp . (2.16)
A floating object is static and therefore the forces on the object sum to zero. In this case the force balance
includes only gravitational force and buoyancy. Let the vertical direction be the z-direction. Then
X
Fz = Fg + FB = 0.
The force of gravity is the object’s mass multiplied with Newton’s gravitational constant
X
Fg = mobj g
So putting the two together we get
FB = ⇢w gVdisp = mobj g
The mass of the object in the case of the aircraft carrier is
2000 lbm 1 kg
mobj = 100, 000 tons ⇥ ⇥ = 90, 718, 474 kg
1 ton 2.2 lbm
and multiplying the mass by the gravitational constant gives the result
11 According to most 4-year-olds, this can be restated as, “Light things float, heavy things sink.”

42
FB = mobj g = 9.1 ⇥ 107 kg ⇤ 9.8 m/s2 = 8.9 ⇥ 108 N
We can also solve for the volume of water displaced by the ship as it floats we get
1029kg
Vdisplaced = 9.1 ⇥ 107 m3 ⇥ = 8.8 ⇥ 104 m3 (2.17)
m3
It’s a well known fact that physicists will do things like model a horse as a sphere to make the math
easier. Figure 2.9 is a graphical representation of this assumption.12
Figure 2.9: A physicist, a horse, and the sphere the physicist uses to represent the mass of the horse. Images
from Aki, Wikipedia and Wolfram respectively. Used without permission but in good fun.
Engineers, however, are responsible for the safety and well being of a ship’s passengers, and hence, can’t
make such absurd simplifying assumptions. Let’s take a closer look at the USS Gerald Ford and represent
it with a simpler geometry, as in Figure 2.10.
Figure 2.10: The floating USS Gerald Ford represented as a collapsable horse trough. Important dimensions
of the ship are length, L, draft, D, beam, B and dead rise, ✓.
Let’s see if the volume method, FB = ⇢water g Vobject , matches the method of integrating the pressures
along the hull.

12 It is left to the reader to judge the validity of this assumption.

43
First, the volume method,
FB = ⇢water g Vobject (2.18)

1
= ⇢water g BDL (2.19)
2
⇢gBDL
= (2.20)
2
Good.
Now let’s do it the real way.
One little di↵erential element will be
dF = P dA cos ✓ (2.21)
B/2
= P dA (2.22)
S}
| {z
cos ✓
B/2
= ⇢gh L ds
|{z} (2.23)
|{z} S
P dA
B/2
= ⇢g |s sin
{z ✓} L ds (2.24)
S
h
D B/2
= ⇢gs L ds (2.25)
S
|{z} S
sin ✓
(2.26)
Now we have a di↵erential force, all in terms of the length of the side of the hull. Now, and only now,
can we integrate. We will integrate from the surface down to the bottom of the hull, choosing s to go from
0 to S.
Z S
D B/2
FB = = ⇢gs L ds (2.27)
0 S S
Z S
BDL⇢g
= s ds (2.28)
2S 2 0
 S
BDL⇢g s2
= (2.29)
2S 2 2 0
BDL⇢g S 2
= (2.30)
2S 2 2
BDL⇢g
= (2.31)
4

44
So close, but yet so far. But wait! That’s only one side of the ship. If we consider that the ship has a
starboard and port side of the hull, we get the result we want,
BDL⇢g
Fb = (2.32)
2

45
2.5 Frank’s Tips for Fluids

13
In Frank White’s Fluid Mechanics book he o↵ers the following as a list of problem solving techniques:
1. Read the problem carefully and try to come up with your own summary of it
2. Gather necessary information from tables and charts. Write it down in your workspace
3. Make sure you understand what is being asked within the problem. If it seems unclear ask a classmate.
If still unclear ask an instructor
4. Make a detailed sketch of the system or control volume (we are not expecting Van Gogh, but if you
put in the e↵ort on your sketches this will aid greatly in problem solving).
5. List assumptions. If you decide the flow is incompressible note why. If you choose a control volume
analysis rather than a di↵erential equation indicate the reasoning.
6. Find algebraic solution (if possible) prior to making any calculations. This alleviates some calculator
mishaps and rounding issues.
7. Finally, report solution with correct significant figures, include vectors if needed and put a box around
that sucker to make it obvious to the grader
13 White 6th Edition

46
Chapter 3
Fluids in Motion
Figure 3.1: Ocean waves
3.1 Flow Descriptions
In thermodynamics we learned two very important concepts: Control Volume and Steady State. Let’s take
a closer look. Consider the nozzle shown in Figure 3.2,
and ask yourself a very important question.
@q
Is the flow through the nozzle steady? (By our usual definition of steady, that is @t , where q
is whatever quantity we want?)
Hopefully, after some debate, we find that the answer is “It depends.” Depends on what? It depends on

47
Figure 3.2: A steady-state nozzle operating at conditions ṁin = ṁout
your reference frame. Are you sitting at one spot in the flow watching to see if q changes? Or are you really
walking a mile in another fluid’s shoes and experiencing what that parcel of fluid experiences?
Perhaps we’re not quite ready to talk about steady yet.
3.2 Lagrangian and Eulerian Flow Field Descriptions
Let take a look at a chunk of fluid. Its size and shape is roughly that of a potato. We’ll even give it a name:
⌦. Further, we will call ⌦ our system, by the fluid mechanics definition of system.
Figure 3.3: Lagrange description - following particles as they travel downstream
One way of representing the fluid motion is by tracking fluid elements, particles or parcels. This viewpoint
or description is called the Lagrangian description1 . We can keep it separate from its peers because of its
name, ⌦. If we restrict the movement to an xy-plane, as seen in Figure 3.3, the coordinates of element ⌦ at
various times can be written as vector from the origin are,
X(⌦, t) = X(⌦, t)x̂ + Y (⌦, t)ŷ

1 Named after Italian mathematician Joseph Lagrange

48
From the Lagrangian view, we continue downstream with the particles that were in ⌦ from the very
beginning. Even as the particles break up and move in di↵erent directions we still must keep tabs on every
one of them. As the group of particles you are concerned with becomes larger this task becomes increasingly
daunting. The velocity, V⌦ , is always measured at one that one chunk. We don’t care where it is, we just
know that
dX(t)
V⌦ = V(t) =
dt
This is a nice approach for two reasons. First, the quantity of interest is only a function of time. I don’t
need to define q as a function of x, y and z because I am only measuring q at ⌦. Second, the kinematics are
easy. Newton’s Law, F = ma applies just like it always has.
But what if ⌦ breaks up into two pieces, which do we track? What about 10 pieces? What if it disperses
into a bunch of molecules, violating our precious continuum assumption?
Often, it easier and more practical to focus on specific locations in space rather than individual particles.
For instance, watching the air flow over a cyclist. We aren’t interested in the air that isn’t interacting with
the object, so why track it?
Figure 3.4: Euler description - watching a control volume
The Eulerian description focuses on a specific point or volume in space rather than an individual particle
or group of partiles. An example of this type of analysis would be a fixed thermometer that provides the
temperature of a specific point for some length of time. We measure the temperature in one location as the
fluid passes by - as the fluid passes through the box.
The box, in this example, is called the control volume and it is bounded by control surfaces. To define
the temperature everywhere in the box, we would have to define temperature as
T = T (x, y, z, t).

49
We would have to define a temperature field.
A vector quantity, velocity, for example, would be defined as the velocity field
V(x, y, z, t) = u(x, y, z, t)î + v(x, y, z, t)ĵ + u(x, y, z, t)k̂.
This is certainly more practical for a couple reasons. First, it is better suited for practical applications.
We want to know what the steam is doing while it is interacting with the turbine blade. What is the air
doing while it is flowing around the wing? What is the water doing while it is in the pipe? Second, it’s
much, much easier to make measurements at a fixed point than to make measurements at a specific parcel
moving downstream.
There is one small detail that will require some work...
How do we apply F = ma to a control volume?
The answer to this is a lecture of its own.
One way to think about the di↵erence between these two descriptions is to think of Euler as a fisherman.
He sits on the shore of a river, with this fishing pole and cooler, and he watches a fixed region of the river
for some time. For the Lagrangian description you concern yourself with a specific fish or school of fish, and
you travel down the river following them.
Figure 3.5: Euler the fisherman and the Lagrangian fish

50
3.3 Flow Regimes
3.3.1 Steady vs Transient
We’ve learned that steady and unsteady are a little trickier than we thought. We’ll define steady as no change
in properties at a point with respect to time. During steady flow, a property (velocity, density, concentration,
etc) can change from point to point, but at any given point within the flow, the property remains constant.
3.3.2 Compressible vs Incompressible
Whether a flow is compressible or incompressible not depends largely on whether you are an engineer or a
physicist. You may look at the air in the room and decide that density is varying (breath, corner by door,
air around the bulb on the projector). Hence, the flow in the room is compressible. You may look at a lazy
river and decide that density of the water sticks pretty darn close to 998 kg/m3 and is hence incompressible.
The question2 is
How much change in density are we willing to accept and still call the flow incompressible?
Generally, we’ll consider liquids to be incompressible. Even at 200 atmospheres, the density of water
only changes by 1%. Gases are a little trickier. Let’s define a special number called the Mach number. This
is a dimensionless number,
V
Ma = ,
c
where V in this case is the speed of the flow and c is the local speed of sound.3 If we are willing to accept
changes in density up to 5%, we can treat all flows below Ma¡0.3 as incompressible. Once we exceed Ma
= 0.3, we should probably consider compression, and all of the thermodynamic implications (namely work)
associated with compression.4
3.3.3 1- 2- and 3-dimensional Flows
3.3.4 Internal vs External Flows
Internal flows are bounded by walls on all sides. External flows flow over a surface of interest.
2A question that separates us from physicists.
3 Be aware that the speed of sound is heavily dependent on temperature, and hence changes with altitude. Given two jets
traveling at the same speed at di↵erent altitudes, one could be supersonic (M¿1)pand one could be subsonic (M¡1).
4 Careful observation shows another way to think about Mach number: Ma = K.E./u where K.E. is the kinetic energy and
u is the specific internal energy. When Ma ¿ 1, there is more kinetic energy than internal energy.

51
3.3.5 Viscous vs Inviscid Flow
In viscous flow, shear and viscosity are important, and possibly even dominate the dynamics of the flow. For
example, our moving plate and journal bearing problems are highly viscous. Moving objects are separated
from stationary objects by thin layers of viscous fluid.
Inviscid flows are those in which viscosity isn’t a big player. Everything is moving along at basically
the same speed. Without velocity gradients, there is no shear. Without shear, viscosity becomes irrelevant.
Hence, inviscid.
3.3.6 Laminar vs Turbulent Flow
Depending on the velocity of the flow, the size of the pipe or object, the density of the fluid and the viscosity
of the fluid, there are two very important flow regimes. These two regimes are laminar flow and turbulent
flow. Our experiments with moving plates and viscosity all relied on the assumption of laminar flow. The
word laminar itself comes from lamina, or laminates, or layers. The picture of laminar flow is that of neatly
organized layers of fluid. An important parameter in this sense is the Reynolds number,
⇢U L
Re = , (3.1)
µ
in which ⇢ is density, U is velocity, L is the important length, and µ is viscosity. The Reynolds number is
another dimensionless quantity that we will examine more carefully later. The numerator of the Reynolds
number could be described with words like big, fast and heavy. These all imply high momentum. The
denominator, µ, could be described as the stickiness of the flow. If you find these two descriptors to be at
odds with each other, you are right. If the flow has a lot of momentum, the slightest nick, scratch, bump or
wiggle in the flow could generate swirls, eddies and vortices that feed on each other and on the flow energy.
If the flow is very thick and viscous, the viscosity will quickly damp out the swirls, eddies and vortices.
For internal flows, the important length scale is the diameter of the pipe or height of a channel. When
the Reynolds number based on diameter, ReD , exceeds about 2300 [-], the flow is likely to be turbulent.
For external flows, the critical Reynolds number is around 5x(105 ). The important length scale for
external flows is the distance over which the flow has been in contact with the surface.
Think of a parade. Consider a marching band vs the Shriners on their little miniature cars or motorcycles.
What is their average velocity? What is their average kinetic energy?

52
3.3.7 Single phase vs multiphase Flow
In thermodynamics, you often considered boilers, evaporators and steam generators. Liquid water goes in,
steam comes out. We largely ignored what happens in between. In reality, there are several flow regimes
that involve two phases of water existing at the same time. Figure 3.6 shows a schematic of some of the
two-phase flow regimes that occur within the evaporator.
Figure 3.6: An evaporator with liquid refrigerant entering at the bottom left and vapor exiting the top right.
The boiling process results in several di↵erent two-phase flow regimes.
3.4 Flow Visualization
3.4.1 Timelines
Mark several adjacent points at one instant in time. Flow over a wing.
3.4.2 Streaklines
Mark one point in the flow at one instant and watch where it goes. Glowing ember out of campfire.
3.4.3 Pathlines
Look at all the particles that passed through a particular point. Smoke bomb in a light breeze.

53
3.4.4 Streamlines
A line that is tangent to the velocity at all points. Because it is always tangent, the slope must be
dy u
=
dx v

54
3.5 Reynolds Transport Theorem
We discovered that an Eulerian description of a flow, that is just watching what happens in a control volume,
is very convenient for instrumentation and allows us to focus on what’s interesting (e.g. the air flowing over
the cyclist.) We know from our physics and dynamics courses that we like to track a system, or chunk of
matter, because chunks of matter obey F = ma.
The Reynolds Transport Theorem allows us to link the two together mathematically. At it’s core, it tells
you something you already know:
IN OUT = STORED
We will consider the same nozzle we looked at during our steady-transient debate. The dashed line in
Figure 3.7 represents our control volume. The mass within the dashed line is our system. Remember, the
system is going to move downstream while the control volume is going to sit right where it is.
Figure 3.7: A converging nozzle with some fluid passing through at time, t. The control volume is shown
with dashed lines and the system is the cross hatched area within the control volume.
First, let’s make sure we can calculate the amount of stu↵ in the systems. In thermodynamics, we referred
to the total entropy in the system as S, and the specific entropy as s. The total enthalpy was H and the
specific enthalpy as h. As we develop the Reynolds Transport Theorem, we’ll refer to the Amount of stu↵ as
A, and the specific amount (amount per mass) as ↵. So what’s the total amount, A, of stu↵ in the control
volume? It is given by Equation 3.2,
Z Z
ACV = ↵ dm = ↵ ⇢ d–
V (3.2)
CV CV
Go through the units of Equation 3.2 and make sure they work out.
Now the system moves downstream of the control volume a little bit as in Figure 3.8,
Looking at the control volume, some new mass has entered and occupies Region I and some of the mass
in our original system has left the control volume and occupies Region II. Looking at the system, it is still

55
Figure 3.8: A converging nozzle with some fluid passing through at time t + t. Some of the system has left
the control volume.
the same collection of mass, but it now occupies a slightly di↵erent volume. The volume it now occupies is
V
–system,t+ t = CV Region I + Region II
So what is the amount of stu↵, A, in the system at t and t + t? Remember the mass in the system
remains the same, but we may have picked up or lost a little stu↵. These two quantities are
Asystem,t = ACV,t (3.3)
Asystem,t+ t = ACV,t+ t AI,t+ t + AII,t+ t (3.4)
We can see what changed by subtracting Equation 3.3 from Equation 3.4,
Asystem,t+ t Asystem,t = ACV,t+ t ACV,t AI,t+ t + AII,t+ t.
When we divide the di↵erence through by t, we get something that begins to look very much like a rate,
Asystem,t+ t Asystem,t ACV,t+ t ACV,t AI,t+ t AII,t+ t

= + .
t t t t
We then take the limit as t approaches zero to get
dAsystem dACV
= Ȧin + Ȧout . (3.5)
dt dt
You can stare at Equation 3.5 for a while and try to make sense of it, or rearrange just a bit to get Equation
3.6,
dAsystem dACV ⇣ ⌘
= + Ȧout Ȧin . (3.6)
dt dt CV

56
So what does this mean?
The rate of change of the amount of stu↵ in the system is the same as the rate of change of
the amount of stu↵ in the control volume plus the net flux of stu↵ out of the control volume.
In other words
If more stu↵ is leaving the control volume than is coming into the control volume (net outward
flux), the system must have gained stu↵ while it was in the system!
The geometry of our nozzle is very convenient. The inlet and outlet are essentially perpendicular to the flow,
so we could rewrite Equation 3.6 by breaking the in and out terms into density, specific amount, velocity
and area. This gives us

dAsystem dACV
= + (↵2 ⇢2 V2 A2 ↵ 1 ⇢ 1 V 1 A1 ) . (3.7)
dt dt
You may look at Equations 3.6 and 3.7, and you may think that 3.6 looks a little better than 3.7.5 And it
does, until you have a potato shaped control volume like that in Figure 3.9, with mass flowing in and out at
weird angles all over the control surface.
Figure 3.9: An inconveniently shaped control volume with an awkwardly distorted system and mass fluxes
in and out at random places at all di↵erent angles other than orthogonal to the control surface.
No we can see that Equation 3.6 is pretty, but only useful in very limited situations. More generally, we
could write the in and out terms in integral form as
Z
Ȧnet = ⇢ ↵ u · n̂ dS, (3.8)
CS
where the subscript CS denotes the control surface and dS is the di↵erential surface element6 . Notice that
5 You shouldn’t judge an equation by how it looks.
6 The use of dA for a di↵erential area is fine here. But to avoid mistakes within the alphabet soup we created, use of dS is
preferred over dA

57
since we dot the flow of stu↵, ⇢ ↵ u, with the normal vector, n̂. This captures the amount of stu↵ that is
actually going in the out direction. If this integral works out to be positive, there is a net outflow of stu↵.
A net outflow of stu↵ means the system is gaining stu↵ while in the control volume.
Just in case the stu↵ isn’t evenly distributed in the control volume, we should be more general and say
that the total amount of stu↵ in the control volume is
Z
ACV = ⇢ ↵ d–
V
CV
and then the rate of change of the amount of stu↵ in the control volume (an important term in Equation 3.6
Z
dACV d
= ⇢ ↵ d–
V
dt dt CV
Finally7 , we can put it all together to get the Reynolds Transport Theorem,
Z Z
dAsystem d
= ⇢ ↵ d–
V + ⇢↵u · n̂ dS. (3.9)
dt dt CV CS
One more way to think about Equation 3.9:
The rate of change of stu↵ in the system (the piece of mass we’re tracking) is the same as the
rate of change of stu↵ in the control volume (the box we’re watching), UNLESS..
• if there is a net flow of stu↵ out of the control volume, the system must be picking up stu↵
on its way through
OR
• if there is a net flow of stu↵ into the control volume, the system must be losing stu↵ on its
way through.
7 Finally.

58
3.6 Material Derivative
We’ll tackle the material derivative in three steps:
1. The general, mathematical description. Accurate, general, hard to follow
2. Physical intuition example Not accurate, very specific, easy to understand
3. The derivative of velocity Accurate, physically understandable, mind blowing.
3.6.1 Derivatives in Lagrangian and Eulerian Reference Frames
We’ve spent some time considering how to relate changes to a system (the potato shaped chunk of mass,
⌦) as it passes through the control volume. We agree that the Eulerian control volume approach is more
practical and more interesting, but the Lagrangian system approach is nice because F = ma applies to
chunks of mass, not control volumes. We can rewrite F = ma as
du
F=m
dt
where u is the vector velocity of our system. We need to take a pretty close look at how we take derivatives
in the Lagrangian and Eulerian sense. It’s probably easier to start with some scalar quantity. Since we’re
running out of convenient letters, we’ll just use f .
We want to consider some physical scalar quantity, f (r(⌦, t), t) ( where f could be temperature, concen-
tration, etc.), and investigate how this property changes following the system ⌦.
This is a little funny looking because time, t, shows up twice.
f( r(⌦, t) , t
|{z} ) = f (r, t)
| {z }
position of ⌦ The scalar field
changes with time
changes with time
We can start by taking the derivative with respect to time. Since r is also a function of time we must
include this term in our derivative and by use of the chain rule we are left with,
df (r, t) @f dr @f
= +
dt @r dt @t
The second term on the right hand side is simply how f changes over time. The first term is comprised
@f
of two parts. The first part, @r , represents how f changes along the domain of r,
@f @f @f @f
= î + ĵ + k̂ = rf.
@r @x @y @z

59
@r
The second factor of the first term, @t , is the velocity of the fluid parcel. If we let the position of the
fluid particle ⌦ be equal to the position in the field at the specific point in which the particle occupies8 , then
r(⌦, t) = x(t)
| {z } |{z}
Lagrangian Coordinates in
position vector Eulerian space
and
dr dx
= = u(x, t)
dt
|{z} dt
|{z}
Lagrangian Eulerian
velocity velocity vector
Putting this back together yields,
Df @f
= + u · rf (3.10)
Dt @t
This is called the Material or Lagrangian Derivative and it is so important in fluids that it is given its
D
own symbol Dt . This equation is what relates the Eulerian and Lagrangian descriptions because it accounts
for changes in field quantities both over time and space.
8 In other words, the origin for the Lagrangian description is conveniently in the same spot as the origin of the Eulerian
description

60
3.6.2 Some Physical Intuition
Let’s consider a more practical example. Imagine you are SCUBA diving in the Caribbean somewhere.
You’re in some shallow water on a sunny day.
You spot your favorite rare tropical sea urchin a few meters ahead and decide to take a photo. As you
approach the urchin, you startle a small octopus. The octopus swims away to the right while squirting out
a cloud of dark, purple ink. You can no longer see your rare sea urchin.
Figure 3.10: A SCUBA diver swimming through a cloud of octopus ink in an attempt to photograph a sea
urchin.
As the octopus escaped, it’s supply of ink decreased, so the concentration of Ink, I, in the water decreases
to the right. That is,

@I
= 0.25Inks / m
@x
Keep in mind, you are also swimming from left to right, that is
u = 0.5 m/s
You may not know that octopus ink is very photoreactive, that is, it breaks down very quickly in sunlight.
So as the time goes on, the ink is fading in place. This is good for you and your photograph, because
dI
= 0.1 Inks/s
dt
too.

61
Now you have two things going for you. You are swimming through a cloud of ink that is thinning to
the right and the cloud of ink is breaking down from the sunlight. You (the Lagrangian photographer)
experience a total change of ink concentration of
dI dI
change in ink = +u
dt  dx
ink ⇣ h m i⌘ ✓ 
ink
◆
= 0.1 + 0.5 0.25
s s m

ink
= 0.225
s
Since both of these terms are negative in our case, you’ll be shooting through crystal clear tropical water
before you know it.
In our case, we would write


DI ink
= 0.225 .
Dt s

62
3.6.3 Acceleration, du/dt, Lagrangian and Eulerian Senses
Let’s run through it one more time, using the velocity vector, u, as the quantity of interest. We know that
for a particle, or potato, or system that acceleration and velocity are related as
dusys
a= .
dt
Let’s picture our system as being pretty small relative to our control volume, so that our system doesn’t
just pass through, but can move around in the system. Considering our system to be a particle is probably
the way to go in this case9 . We know that as our particle moves around in the control volume, or field, the
velocity of the particle at any given time depends on its x-, y- and z-positions, all of which are functions of
time.10 But, as we learned in the octopus example, the entire field could be changing with time, too.
uP = u(xP (t), yP (t), zP (t), t).
The velocity field could be changing with time, and the particle’s velocity depends on where it is in the field,
also a function of time.11 This is the classic application of the chain rule. We remember12 that the total
derivative is the sum of all the partial derivatives,
duP @u @u dx @u dy @u dz
aP = = + + + . (3.11)
dt @t @x dt @y dt @z dt
Something very special happened here, did you notice it? What happened to the subscripts?
(long pause to allow brain dust to settle in room)
On the left side of Equation 3.11, we have the acceleration of the particle, potato, system, whatever. But
on the right hand of the equation we only have derivatives in terms of the field in our control volume. We
9 Subscript P means the value is for the Particle
10 The particle is, after all, moving.
11 In Calc 1 parlance, this is a function inside a function. Velocity is a function of time and position. Position itself is also a
function of time.
12 Maybe

63
have names for dx/dt and the like, so we can rewrite Equation 3.11 as
@u @u @u @u
aP = +u +v +w (3.12)
@t @x @y @z
| {z }
looks familiar
@u
= + u · ru (3.13)
@t
@u
= + u · ru. (3.14)
@t
Putting it together, we can write
Du @u
aP =
Dt
=
@t
+ | ·{z
u ru}
|{z} Advective/convective derivative
Temporal derivative

64
The convective terms looks like,
2 3
@u @v @w
 6 @x @x @x 7
6 7
u · ru = u v w ·6 @u
6 @y
@v
@y
@w 7
@y 7
4 5
@u @v @w
@z @z @z

= u @u @u @u
@x + v @y + w @z
@v
u @x @v
+ v @y + w @v
@z u @w @w @w
@x + v @y + w @z
Example: flow through a nozzle
We’ve looked at the example of a nozzle throughout this section so let’s revisit it one more time with out
new tools.
Figure 3.11: Steady through a nozzle with velocity vectors at inlet and outlet
The flow through the nozzle is still steady, but this time we are given a function of velocity with respect
to position inside the nozzle,


u = 3x2 0 0
Compute the acceleration of a particle as a function of position within the nozzle.

65
Looking ahead a little bit, we now have the tools to apply Newton’s second law to the fluid in a control
volume. We just finished the acceleration. We know the mass. What are the forces?
• Force due to gravity
• Force due to di↵erence in pressure
• Force due to viscous interactions

66
Chapter 4
Conservation Laws
Figure 4.1: SpaceX launches the Falcon Heavy rocket in April 2019
4.1 Conservation of Mass - Integral Form
We have spent some time looking at the Reynolds Transport Theorem as it applies to mass. Since the system
is defined as mass, the LHS must be exactly zero.
Z Z
dmsystem d
= ⇢ d–
V + ⇢u · n̂ dS. (4.1)
dt
| {z } dt CV CS
=0

67
leaving us with only two options. The one we have addressed so far is that of a steady flow,
Z Z
dmsystem d
= ⇢ d–
V+ ⇢u · n̂ dS
| dt
{z } |dt CV
{z } CS
=0 =0
Z
⇢u · n̂ dS = 0
CS
IN = OUT
So what if we consider a changing control volume? For instance, a balloon.
4.1.1 Example: Shrinking Balloon
If a balloon starts at R0 and I release the neck of the balloon, how long will it take to reach some predefined
smaller radius, Rs ?
(draw balloon)
Now it gets a little more interesting. We are back to two interesting terms,
Z Z
d
0= ⇢ d–
V+ ⇢u · n̂ dS. (4.2)
dt CV
| {z } | CS {z }
shrinking escaping
balloon air
The first term on the RHS is the shrinking balloon and the second term on the RHS is the escaping mass.
We’ll look at the CV first.

✓ ◆
dMCV d 4⇡r3
= ⇢ .
dt dt 3
And for the flux

Z
⇢u · n̂ dS = ⇢ u1 A1
CS

68
We know the two must balance, so

✓ ◆
d 4⇡r3
⇢ = ⇢ u 1 A1 .
dt 3
Solve for r
d(r3 ) 3
= A1 u 1
dt 4⇡
3
d(r3 ) = A1 u1 dt
4⇡
We’re left with a bit of a question, “What is d(r3 )?” We don’t know the variable of di↵erentiation, so we
need to use the generic form d(r3 ) = 3r2 dr.1 Using this, we write
Z Rs Z t
3
3r2 dr = A1 u1 dt
R0 0 4⇡
3
Rs3 R03 = A1 u 1 t
4⇡
4⇡ R03 Rs3
t=
3 A1 u 1
V0 V s
=
A1 u 1
(check units)
This is a problem you probably did during the Related Rates section in Calc 1, but now we see that you
can get to the same result using a much more flexible, and hence powerful, tool.
4.1.2 Example: Developing Flow over a Flat Plate
Imagine a steady flow of air approaching a flat plate that is aligned with the flow.
1 This goes way back to the definition of a di↵erential: If y = f (x), then dy = f ‘(x)dx. You learned this about the same
time you studied Related Rates.

69
The velocity of the air approaching the plate is called the far field velocity, U1 . After some distance
along the plate, the velocity profile becomes
 ⇣ ⌘ ⇣ y ⌘2
y
u = U1 2 .
h h
The flow is steady, but it is developing along the plate.
Is there any flow into or out of the top surface of the control volume? Which way is it (in or out)?
Again, we’ll start with RTT as applied to mass, also known as the conservation of mass. Since it is steady,
the time derivative vanishes as well.
Z Z
dmsystem d
= ⇢ d–
V+ ⇢u · n̂ dS. (4.3)
dt
| {z } | dt CV CS
{z }
=0 =0
We’ll split the flux term into four components: left, top, bottom and right.
Z Z Z Z Z
⇢u · n̂ dS = ⇢u · |{z}
n̂ dS + ⇢u · |{z}
n̂ dS + ⇢u · n̂ dS + ⇢u · |{z}
n̂ dS = 0
CS AL AT AB AR
= î = ? | {z } =î
=0
We don’t know much about the dimensions in and out of the page, so we’ll call that w, for width.2 The
di↵erential surface, dS, on the left and right sides becomes w dy, giving
Z Z h  ⇣ ⌘ ⇣ y ⌘2
y
⇢wU1 ( 1)h + ⇢u · n̂dS + ⇢wU1 2 dy = 0
AT 0 h h
Evaluating the integral for AR gives
Z
2
⇢wU1 h + ⇢whU1 + ⇢u · n̂dS = 0.
3 AT
Now we solve for the flux through the top surface and get
Z
1
⇢u · n̂dS = ⇢U1 wh
AT |3 {z }
di↵erence between
left and right
We know the RHS is positive, so the LHS must also be positive. This means that u · n̂ must be positive,
telling us that the net flow must be out of the top control surface.
2 NOT w for z-component of velocity.

70
4.1.3 Example: Heating Air
Air is flowing through a 100 mm diameter tube coiled inside a burner. It enters at 203 kPa (absolute), 20
C and 15 m/s. When it exits, the pipe is increased to 150 mm diameter, the pressure has dropped to 150
kPa but the temperature has increased to 75 C. What is the exit velocity?
We know by know that the system doesn’t change mass, so we’ll write our conservation law as if we
already know this,
Z Z
d
⇢ d–
V + ⇢u · n̂ dS = 0 (4.4)
dt CV CS
Equation 4.4, is often called the continuity equation. We will treat the pipe as rigid and the flow as steady.
We know we can just deal with average velocities at the pipe inlet and outlet, so we’ll skip the toil and just
write it as3
0 ⇢in Vin Ain + ⇢out Vout Aout = 0
Solving for Vout ,

⇢in Ain
Vout = Vin .
⇢out Aout
We can almost always treat air as an ideal gas, with
P
⇢= ,
RT
so
Pin
RTin Ain
Vout = Vin Pout
.
RTout Aout
This is nice because it saves us the hassle of looking up4 R because it cancels out. We will rewrite it as
Pin Tout Ain

Vout = Vin
Pout Tin Aout
We know that the area of the surfaces go with d2 , so the area ratio becomes
✓ ◆2
Ain din
=
Aout dout
3 Back in thermo, we just called this a steady state mass balance.
4 Which you will always remember is 287 J/kg-K anyway.

71
We have yet to plug in any values leaving us with a nice clean algebraic expression for velocity,
✓ ◆2
Pin Tout din
Vout = Vin = 10.7 m/s
Pout Tin dout
No need to carry through troublesome constants like R and ⇡.

72
4.2 Conservation of Momentum - Integral Form
One of the most interesting and important applications of fluid mechanics is the interaction between moving
fluids and a solid surface and the forces that result from this interaction. Steam turbines, airplanes, internal
combustion engines, heart valves, jet engines, sailboats... the list goes on and on.
Figure 4.2: The New Zealand based Te Aihe AC75 class sailboat. It converts wind motion into forward
thrust to move it through the water. As it moves through the water, foils convert the flow of water into a
vertical force to lift it out of the water. The rudder converts the flow of water into a lateral force to steer.
In e↵ect, horizontal wind is lifting a 7,600 kg boat out of the water at over 50 mph (20 m/s).
If we are interested in forces, we’ll look again at Newton’s definition of a force,
F = ma
du
=m
dt
d(mu)
= .
dt
We know that (mu) is momentum, so that means that a force is a change in momentum, as in Equation
4.5,
d(mu)sys
F =
|{z} . (4.5)
Force | dt
{z }
rate of change
of momentum
as long as what we’re tracking is a chunk of mass, like a potato.
So if we can track momentum in a fluid, we can calculate forces. The magic of fluid mechanics becomes
real.

73
We can generalize this to include multiple forces, like gravity and pressure, while considering all of the
momentum in the system and write,

X Z
d
F= ⇢u d–
V (4.6)
dt sys
Let’s make sure the units work out,
 
1 kg h m i 3
[N] = [m ]
s m3 s

kg-m
[N] = .
s2
We developed a general form of Reynolds Transport Theorem that uses ↵ as the specific amount of stu↵
in the system and control volume,
Z Z
dAsystem d
= ⇢ ↵ d–
V + ⇢↵u · n̂ dS. (4.7)
dt dt CV CS
If the “stu↵” we want to track is momentum, we can’t simply replace ↵ with (mu) because momentum
is not an intensive property, it depends on mass, so we need to be a little careful. On the left hand side, all
of the stu↵ we’re tracking is already in there, so we can replace Asys with (mu) . But on the right hand
side, the “specific momentum,” if such a thing existed, would be (mu)/m, or simply u. So on the right hand
side, we replace ↵ with u,
Z Z
d(mu)system d
= ⇢ u d–
V + ⇢u(u · n̂)dS. (4.8)
dt dt CV CS
The left hand side of this looks familiar. In fact, we just saw it as the right hand side of Equation 4.5,
which is equal to the sum of the forces. Combining Equations 4.5 and 4.8, we get the elusive application of
Newton to a control volume, the momentum equation,
Z Z
d
F
|{z} = ⇢ u d–
V + ⇢u(u · n̂)dS . (4.9)
dt CV
Sum of external | {z } | CS {z }
forces acting Time rate of change Net flux of
on the CV of momentum momentum out
in the CV of the CV
This is the impressive result of hundreds of years of physics and mathematics, but to the future fluid
dynamicist5 , it may be a little intimidating. We’ll look at a few simplified cases first.
5 You.

74
4.2.1 Frictionless Di↵erential Cylindrical Slug of Liquid Steadily Accelerating

Along a Streamline
Figure 4.3: A cylindrical slug of liquid accelerating along streamline
This is a very special, very simplified case, but we’ll take a look. We’ll split our forces into two categories.
Surface forces, FS , are those acting on the surface of the slug and body forces, FB , are those acting on the
body. We’ve already determined that the only body force we’re going to deal with is gravity.
Z Z
d
F S + FB = ⇢ud–
V + ⇢u(u · n̂)dS
dt CV CS
Since the flow is specified as steady, the time rate of change of the control volume is zero. We are
neglecting viscous shear stresses on the surface, and therefore the surface force is only due to the pressure
di↵erence on either end of the slug,
FS = P A (P + dP )A (4.10)
= A dP (4.11)
The body force is due to gravity,
FB = ⇢ gS d–
V = ⇢ g cos(✓)A ds
where gS is the component of gravity acting along the streamline, ✓ is the angle between the stream line and
y-axis (labeled in the figure), and ds is the di↵erential length along the streamline. From basic trig we can

75
see that cos(✓)ds is actually dz, if we define z as up. So our body force simplifies to
FB = ⇢ g A dz
Looking at the net momentum flux, we really have just an inlet and an outlet, so
Z
⇢u(u · n̂)dS = u( ⇢ u A) + (u + du)(⇢ u A) (4.12)
CS | {z } | {z }
in out
= ⇢uA du (4.13)
keeping in mind that the in term is the negative one.
The momentum equation allows us to set our net external forces equal to our net outward momentum
flux, so
A dP ⇢gA dz = ⇢uA du.
We can divide out the area now to get,
dP ⇢g dz = ⇢u du.
Finally, we can integrate from 0 to P, 0 to z, and 0 to u, to get a famous equation,
u2
P +⇢ + ⇢ g z = constant (4.14)
2
You will find that people, especially those who never endured a semester of fluid mechanics, will often
attribute everything to Bernoulli. “How does an airplane fly?” Bernoulli. “How does a sailboat work?”
Bernoulli. “Why does a curve ball curve?” Bernoulli. “Why is the sky blue?” Bernoulli. While you may
have scowled when you read the title of this section, Frictionless Di↵erential Cylindrical Slug of Liquid
Accelerating Along a Streamline, it is exactly that. Bernoulli only applies under very special conditions.
Of course, as engineers, we will sometimes neglect some friction. We will sometimes model flow through
a pipe as a slug. We will sometimes assume that the flow is along a streamline, even when we know it is
turbulent. When we do so, it is very, very important that we know what kind of assumptions we’re making.

76
4.2.2 Steady Flow with One Inlet and One Outlet
If flow is steady and uniform through the control volume, for instance, flow through a section of pipe, we
can simplify the flux term a little bit. The flux term,
Z
⇢u(u · n̂)dS
CS
includes the velocity, u, twice in the integrand. If this is constant, we are free to pull one of them out,
leaving a familiar looking integral,

Z
u ⇢(u · n̂) dS
| CS {z }
ṁ
So the net momentum flux simplifies to
Z X
⇢u(u · n̂)dS = u ṁ
CS
where each u ṁ is the momentum flux across a uniform inlet or outlet.
Now we can apply this to our momentum equation,
X
F = ṁ(uout uin )
or in scalar notation
X
F = ṁ(uout uin )

77
4.2.3 Summary of control volume analysis
The purpose of this type of analysis is often to determine the forces necessary to change the direction fluid
motion. The following is a general procedure that can be used to solve these problems:
1. Choose a good control volume
• Choosing a control volume wisely will often simplify the problem. Draw control volumes with
control surfaces perpendicular to the fluid stream lines.
2. Identify forces
• Body forces - gravity
• Surface forces - due to stresses such as pressure and shear
• Reaction forces - in control volume analysis there is often a reaction force applied from the
structure onto the fluid stream
• Moments - may also need to be considered
3. Identify changes in momentum
Changes in momentum occur due to:
• changes in mass flow
• changes in velocity magnitude
• changes in velocity direction
For processes with steady mass flow in and out of control volume, we must only concern ourselves with
changes in velocity.
4. Momentum/Force balance
Start with the general form of the linear momentum equation.
X Z Z
d
F = Fb + Fs + R = ⇢u d–
V + ⇢u(u · n̂) dS
dt cv cs
Cancel any terms that can be neglected. If the process is running at the steady state then the control

78
volume term goes to zero. We are then left with
X Z
F= ⇢u(u · n̂) dS
cs
If we’re dealing with a single streamline, then the mass flow of the in stream is the same as the mass
flow of the out stream.
X
Fx = ṁ(uout uin )
X
Fy = ṁ(vout vin )
X
Fz = ṁ(wout win )
5. Solve for remaining quantities
Gather expressions for unknown quantities which are in terms of given quantities.

79
4.2.4 Accelerating Control Volumes
In the case of accelerating control volumes our analysis changes slightly. The relative velocity and
relative accelerations must be considered in problems with accelerating control volumes. Let’s define
a new term,
uf /cv = Velocity of the fluid, relative to the control volume
and therefore,
uf = Velocity of the fluid = ucv + uf /cv
Now for the momentum balance.
X
F = ma
duf
=m
dt
d
=m ucv + uf /cv
dt
d d
= m (ucv ) + m uf /cv
dt dt
d d
= m (ucv ) + muf /cv
dt dt
Applying the Reynolds Transport theorem on the far right term,
Z Z
d d
muf /cv = ⇢uf /cv d–
V + ⇢uf /cs (uf /cs · n̂)dS
dt dt CV cs
Putting this together we get,
X Z Z
ducv d
F= mcv + ⇢uf /cv d–
V + ⇢uf /cs (uf /cs · n̂)dS (4.15)
| {zdt } dt
| cv
{z } | cs
{z }
Inertia of all mass Rate at which fluid momentum Rate of momentum
within CV in CV is changing with respect in/out relative to CS
to momentum of CV

80
Example: Falcon Heavy launch
Figure 4.4: Falcon Heavy rocket launch - SpaceX
The Falcon Heavy rocket and the fuel carried inside have an initial mass of Mr = 544200 [kg]. The ratio
Mf uel
of the mass of fuel to the total mass is Mr = .952. The engines are ignited and burn for 400 [s]. The
exhaust is expelled at a constant speed of 4500 [m/s] relative to the velocity of the rocket. Use average
mass flow for the exhaust stream6 Neglect drag resistance due to air and changes in gravity due to increased
elevation.
(a) What is the initial weight of the rocket?
(b) Consider the flux of momentum through the control surface of the rocket to be the thrust. This is the
final term on the RHS of Equation 4.15. Calculate the thrust at the moment the rockets are ignited.
(c) Calculate the maximum velocity reached by the rocket.
First things first, draw a good control volume around area of interest.
The weight is easy, but helps put some context with the problem.
Winit = Mr ⇤ g = 5.337 ⇥ 106 ĵ [N] (= 5, 337 [kN])
Let the thrust be defined as
Z
Fthrust = ⇢uf /cs (uf /cs · n̂)dS. (4.16)
cs
We need to make a very important observation here. Back in Equation 4.15, the sign on the last term was
positive. In our definition of thrust, Equation 4.16, we introduce a negative sign. Why is this? The flux
term, the last term on the RHS, is a measure of the net outward flow of momentum. If there is a positive
6 We’re assuming the rocket is single stage. The Falcon Heavy is actually a two stage rocket

81
Figure 4.5: Falcon Heavy rocket with control volume displayed in green
net outward flow, there must be some source of momentum inside the CV. We know from Newton’s second
law, that a source of momentum is also called a force.
But... This is a force acting on the fluid, “pushing” the fluid out of the nozzle. Our goal is just the
opposite, that the fluid pushes back on the rocket. So the quantity we really want, thrust, is the opposite
(negative) of the push on the fluid. Hence, the negative sign.
The good news is that by using average quantities we can skip the integral and we are left with
Thrust = Fthrust = ṁf ue
with ṁf is the mass flow of the exhaust and ue is the velocity of the exhaust. In this case the exhaust
velocity is
ue = 0î + ve ĵ + 0k̂
The average mass flow rate of the exhaust fluid stream is calculated by finding the mass of the fuel and
dividing it by the burn time.
Mf uel Mr ⇤ .952 518000 [kg]

ṁf = = = = 1295 [kg/s]
tburn tburn 400 [s]

82
Finally, the thrust felt by the Falcon Heavy 7 rocket is
⇣ ⌘
Fthrust = ṁf ue = (1295 [kg/s]) 4500 [m/s]ĵ = 5.83 ⇥ 106 ĵ [N] (= 5, 830 [kN])
The max velocity reached by the Falcon Heavy rocket occurs when the instant that the fuel has all been
burned up. To get start with a momentum balance in the y-direction.
X dvcv
Fy = M r ṁf ve
dt
dvcv
Mr g = M r ṁf ve
dt | {z }
PLUS a net
outward flux
in the
negative ĵ
with the mass of the rocket Mr equal to
Mr = Mr,i ṁf t
Plugging this back into the momentum balance yields,
dvcv
(Mr,i ṁf t)g = (Mr,i ṁf t) ṁf ve
dt
To get a function of velocity with respect to time, let’s isolate the derivative of velocity of control volume.
dvcv ṁf ve
= g
dt Mr,i ṁf t
The di↵erential equation we are left with is separable. Move the dt to the right hand side and integrate the
RHS from 0 to tburn .
Z vmax Z tburn Z tburn Z tburn

ṁf ve ṁf ve
dvcv = g dt = dt g dt
0 0 Mr,i ṁf t 0 Mr,i ṁf t 0
The first integral on the left hand side requires a visit from an old friend, U-substituion. Let’s focus only
the on this term for a moment. Let u be equal to,
u = Mr,i ṁf t
7 The actual thrust felt at engine ignition is around 7600 [kN]. Pretty close and has the correct sign.

83
Then get the di↵erential term dt by itself.
du = ṁf dt
du
dt =
ṁf
Let’s ignore the bounds until we substitute back in for u after the integral. The integral becomes
Z tburn
ṁf ve du
= ve ln (u) = ve ln (Mr,i ṁf t)
u ṁf 0
Let’s ignore the bounds until we substitute back in for u after the integral. The integral8 becomes
Z vmax Z tburn Z tburn tburn

ṁf ve ⇥ ⇤
dvcv = dt g dt = ve ln (Mr,i ṁf t) gt
0 0 M r,i ṁf t 0 0
✓ ◆
Mr,i
vmax = ve ln gtburn
Mr,i ṁf tburn
And finally we are left with the maximum velocity of the Falcon Heavy9 rocket,
vmax = 9730 [m/s]
8 ln(a/b) = ln(b/a)
9 The actual max speed of the Falcon Heavy is around 11 [km/s].

84
Example: Curved surface on moving cart (a.k.a. Turbine Blade) - Revisited
Figure 4.6: A steady 30 m/s jet being deflected by a curved surface mounted to a cart that is moving at a
steady10 m/s.
• ⇢ = 1000 [kg/m3 ]
• Solve for the reaction forces Rx and Ry .
Process for solving
1. Choose a good control volume
If the option is available, choosing a CV that has bounds which are perpendicular to fluid streamlines
is often best. In this case we get something that looks like,
Figure 4.7: Good control volume labeled with green dashed line
2. Identify forces
• Pressure is constant everywhere. Forces due to pressure F = P A have zero influence.

85
• Gravitational acceleration of the fluid will show up in the y component of our resultant force.
• Must be some force that maintains the cart velocity at 10 m/s. But this does not concern us.
• Reaction forces applied from the cart onto the fluid jet
Rx = Reaction force of cart on fluid in x-direction
Ry = Reaction force of cart on fluid in y-direction
3. Identify changes in momentum
Changes in momentum are indicated by:
• Changes in velocity magnitude
• Changes in velocity direction
• Changes in mass flow rate
In this case the velocity magnitude is the same for both the inlet and the outlet of the control volume.
Therefore the the change in fluid momentum is related to the change in fluid velocity direction. Also,
because the cart is moving we must use the change in relative velocity.
ujet = 30 [m/s] î
ucart = 10 [m/s] î
The velocity at the inlet of the CV is then calculated by
uin = [uin , vin ] = (30 10) [m/s] î + 0ĵ = 20 [m/s] î
The exiting velocity has both x and y components. They take the form of,
uout = [uout , vout ] = 20 cos(60o ) [m/s] î + 20 sin(60o ) [m/s] ĵ
= 10 [m/s] î + 17.3 [m/s] ĵ

86
4. Momentum and force balance
The change in momentum experienced by the fluid must be equal to the sum of forces applied to the
fluid. For both x and y components, a momentum balance is performed.
X
Fx = Rx = ṁ(uout uin )
X
Fy = R y M g = ṁ(vout vin )
In the y-component we must include the force of gravity acting down on the mass as it is accelerated
upward by the cart.
5. Solve for remaining quantities
The only unknown quantity in this case is the mass flow ṁ. Solving for these we get
ṁ = ⇢|u|Ajet = 60 [kg/s]
Plus this quantity into our momentum balance and solve for the reaction forces on the cart
Rx = ṁ(uout uin ) = 60(10 20) = 600 [N]
Ry M g = ṁ(vout vin ) = 60(17.3 0) = 1040 [N]
We are given the area of the jet stream but not given a length of the cart. If we say that the length
of the cart is relatively small, then the mass within the control volume tends toward zero. The term
associated with gravity goes to zero.
Mg = 0
The reaction forces from the cart onto the fluid stream is equal to,
R = Rx î + Ry ĵ = 600 [N]î + 1040 [N]ĵ

87
Example: Reactions on a U-shaped Pipe
Figure 4.8: A pipe bent into a 180 C flow reversal device.
• Steady state
• Pout = 0 [Pa] (Using gage pressure is convenient for these problems)
• Din = Dout = 20 [cm]
• ⇢ = 1000 [kg/m3 ]

88
Example: Reactions on a U-shaped Nozzle
Figure 4.9: A nozzle bent into a 180 C flow reversal device.
• Steady state
• Pout = 0 [Pa] (Using gage pressure is convenient for these problems)
• Din = 20 [cm]
• Dout = 5 [cm]

89
Example: Reactions on a 90 Nozzle
Figure 4.10: A nozzle accelerating and deflecting a stream of water downward.
• Steady state
• zin = zout
• Ain = 100 [cm2 ]
• Ain = 25 [cm2 ]
• Pin = 220 [kPa]

90
Chapter 5
Di↵erential Fluid Flow
Figure 5.1: Clouds flowing around a mountain generates Von Karman vortex street - NASA
In the previous chapter, we used integral forms of conservation equations to solve problems. The use
of di↵erential equations in our fluid analysis will often allow for us to solve the pressure, velocity and
acceleration profiles, throughout the flow field, rather than just at the inlet and outlet. The di↵erential
forms of the conservation equations can be applied to any location in the flow field and are not restricted to
a specific region.

91
5.1 Conservation of Mass - Di↵erential Form
Let’s start by examining a di↵erential fluid element as our control volume. Consider the system to be the
mass of fluid that occupies the volume at some fixed time, and the positive x-direction points to the right.
Figure 5.2: Fluid element with volume equal to V

– = dx dy dz.
For now, let mass flow in and out of the CV be restricted to the flux through the control surfaces whose
normal is parallel with the x direction, as seen in Figure 5.2. Apply Reynolds Transport Theorem for mass
on the fluid element.
Z Z
dMsys d
=0= ⇢ d–
V + ⇢(u · n̂)dS
dt dt cv cs
This expression simplifies in our case to
@⇢ @(⇢u)
0= dx dy dz + (⇢u + dx) dy dz ⇢u dy dz .
|@t {z } | @x {z }
Local change in mass Net flow of mass out
Notice the area of the both the left hand face and the right hand side control surface are equal to (dx)(dy).
Rearranging our new expression a bit,
@⇢ @(⇢u)
0= dx dy dz + (⇢u + dx)) dy dz ⇢u dy dz
@t @x
@⇢ @(⇢u)
= dx dy dz + dx dy dz.
@t @x
Divide both sides of the equation by the volume, dx dy dz, we are left with,
@⇢ @(⇢u)
0= + .
@t @x
If we include the changes in mass due to flux through the y and z direction surfaces, then the equation
becomes

92
@⇢ @(⇢u) @(⇢v) @(⇢w)

0= + + + . (5.1)
@t @x @y @z
We can rewrite equation 5.1 using vector notation as,
@⇢
0= + r · (⇢u). (5.2)
@t
This equation is the di↵erential form continuity equation. This equation can be thought of as a
stronger and more local application of conservation of mass. The first term indicates how density is changing
with respect to time. The second term indicates the net outward mass flux or the divergence of the mass.
Since the mass in any chosen system is constant the two terms must be exactly equal and opposite. This
should be somewhat intuitive. If the density of our system is decreasing, then mass that once lie within the
original control surfaces has moved outward and the volume our system occupies has increased. The rate
at which this volumetric expansion occurs depends on the velocity of the fluid. In scenarios like this, fluid
density is a function of both time and space ⇢ = ⇢(x, y, z, t).
5.1.1 Incompressible fluids
In many applications, the density of the fluid can be treated as constant. If this is the case, then we can
simplify equation 5.2 by eliminating the temporal derivative of density and removing it from inside of the
divergence term.
0
@⇢
0= + ⇢(r · u) = ⇢(r · u)
@t
Dividing both sides by the density we are left with,
0 = r · u. (5.3)
This result indicates any incompressible flow field will be divergence free. However, flow fields can be
divergent free without having constant density.
Example: Steady Incompressible 2-D Flow Field
Consider another, simpler, example of a flow field in the x-y plane,
u = Ax.

93
With your neighbor, try to
• Sketch some vectors representing the velocity of the fluid at di↵erent x-positions in the x-direction.
• Apply equation 5.3 to this flow field.
0
@u @v @w◆
7
0=r·u= + + ◆
@x @y ◆@z
@v
• Solve for @y
@u @v
+0=
@x @y
@v @u
= = (A)
@y @x
• Integrate to find v(y)
@v = A@y
Z v Z y
@v = A@y
0 0
v= Ay + f (x, t)
It is necessary to include some function of the other variables that could influence the velocity in the y
direction. However, since we are not given any other information we may assume the velocity is only
dependent on y.
• Sketch vectors representing the full 2-D flow field in the x-y plane.

94
Figure 5.3: Velocity field

95
Example: Waves
Water flowing in the x-direction, propagates a wave in the x-direction associated with the flow field. The
velocity of the flow field in the x and y is:
u = U sin(kx !t)
v=0
Where U is the velocity amplitude, 2⇡/k is the wavelength and 2⇡/! is the frequency of the wave. What
is the vertical velocity?
Figure 5.4: Wave propagating in the x-direction
How in the World are we going to get the vertical velocity component of this wave when this is all the
information we are given? Where there’s a will there is a way, friend. Lets apply our di↵erential continuity
equation,
@⇢
0= + r · (⇢u)
@t
Water is an incompressible liquid and so I know there will not be any changes with respect to time. This
means we can say
@⇢
0=
@t
which implies that
0 = r · (⇢u)
Given the incompressibility of water, we can also say it has no spatial dependence and pull it out from
the divergence operator.
0 = ⇢(r · u)

96
Dividing both sides by density,
@ @ @
0 = (r · u) = (
î + ĵ + k̂) · (uî + v ĵ + wk̂)
@x @y @z
@u @v @w
0= + +
@x @y @z
Given that the velocity of the in the y-direction is constant zero, we also know that
@v
= 0,
@y
leaving us with
@u @w
+ = 0,
@x @z
or
@w @u
=
@z @x
= kU cos(kx !t).
We can now integrate from z = 0 (bottom of the ocean) to the local value of z, yielding
w= kU z cos(kx !t).
This gives us a simplified, nonetheless beautiful1 representation of a wave propagating across the ocean’s
surface,
u = (U sin(kx !t))î (kU z cos(kx !t))k̂
1 Probably worthy of a tattoo.

97
5.2 Forces on Di↵erential Element
Even though our fluid element is small, it still has mass and therefore the it’s must obey Sir Isaac’s Laws.
In order to accelerate the fluid element a force or multiple forces must be applied to it. Both surface forces
and body forces act to accelerate the fluid element.
X du
F = Fb + Fs = ma = ⇢ dx dy dz (5.4)
dt
The body forces Fb will most often be gravity.
Fb = mg = ⇢ dx dy dz g (5.5)
The surface forces Fs will be comprised of stresses applied both normal and tangential to the surface
of our system, which means pressure and viscous stress. Let’s consider them one at a time starting with
pressure. The force due to pressure is visualized in Figure 5.5.
Figure 5.5: Force due to pressure on the x-direction faces
Again assuming the positive x-direction points to the right, the force due to pressure on our fluid...
@P @P
Fx,press = P dy dz î (P + dx) dy dz î = dx dy dz î
@x @x
Apply same analysis to the y and z directions. The force felt by our fluid element due to pressure in
vector form then looks like,
@P @P @P
Fpress = ( î + ĵ + k̂) dx dy dz = rP dx dy dz. (5.6)
@x @y @z
5.2.1 Inviscid Di↵erential Flow
If we choose to ignore viscous contributions to the surface forces and call the flow inviscid 2 , then the result
of the our force balance looks like,

2 Invis-kid or invissid?

98
X du
F = ⇢ dx dy dz g rP dx dy dz = ⇢ dx dy dz . (5.7)
dt
The system we have chosen is infinitesimally small and therefore acts like a Lagrangian particle moving
through a Eulerian flow field. To get the acceleration of the fluid element we apply the Material Derivative
on the right hand side
X 
@u
F = ⇢ dx dy dz g rP dx dy dz = ⇢ + u · ru dx dy dz .
@t
Now divide both sides by the volume of the element, dx dy dz, and the result is

@u
⇢g rP = ⇢ + u · ru (5.8)
@t
This equation is conservation of momentum for inviscid fluid and is often times called the Euler Equa-
tion34 . It’s important no notice that gravity, g, is a vector quantity. Generally, we think of gravity as
downward, so applying Euler’s Equation, 5.8, to fluid with constant density in the horizontal plane gives a
slightly more intuitive version,
rP = ⇢(u · r)u
Notice the way we have rearranged the advection term of acceleration. We can do this because
u · ru = (u · r) u
✓ ◆
@ @ @
= u +v +w u
@x @y @z
| {z }
Velocity Weighted Gradient Operator
✓ ◆
@ @ @
= u +v +w u
@x @y @z
Notice the velocity weighted gradient operator is a scalar operator. This new operator will be applied to all
3 components of the velocity, [u, v, w].
Since Equation 5.8 and its simplified cousin are vector equations, they can be broken into their compo-
3 Not to be confused with Euler’s Identity: ei⇡ + 1 = 0
4 Almost tattoo worthy.

99
nents. The three components of the incompressible version above are
@P ⇥ @u @u @u ⇤
=⇢ u +v +w (5.9a)
@x @x @y @z
@P ⇥ @v @v @v ⇤
=⇢ u +v +w (5.9b)
@y @x @y @z
@P ⇥ @w @w @w ⇤
=⇢ u +v +w . (5.9c)
@z @x @y @z
A quick check of the units,
  
Pa kg h m i m/s
=
m m3 s m
multiply both sides by [m3 ]

⇥ ⇤ hmi
[Pa] m2 = [kg] 2
s
This helps us see that Euler’s momentum equation is F = ma per unit volume.

100
5.2.2 Viscous Di↵erential Flow
Euler was a pretty sharp guy - one of the most famous scientists and mathematicians of all time. However,
it takes a lot of guts to neglect viscosity when working with fluids other than gases and cryogenic helium.
Neglecting viscosity may get you in the ballpark of the real answer, but if you want to suit up, take the
mound and really get after some fluid dynamics, then we should probably include viscosity contributions.
Let’s look at the same di↵erential element and work out the forces due to viscous stress.
So far, we have applied F = ma to our fluid element using pressure (a surface force) and gravity (a body
force) as the only force applied to cause our acceleration.
Now we need to take a look at the other surface force, shear.
We’ve already learned that shear is a result of viscosity. In our 2-D problems, shear forces acting in the
x-direction were a result in a velocity gradient in the y-direction. Specifically,
du
⌧x = µ .
dy
If we take a look at a three dimensional infinitesimal element, such as in Figure 5.6, we can see that the
forces acting in the x-direction come from shear on the y- and z-faces. But what is the shear on these faces?
Figure 5.6: Forces in the x-direction due to shear at the surfaces.
Let’s take a step back and remember our discussion of the gradient function. The gradient of a scalar
(stinky tuna can example) gave us a vector, since there could be di↵erent smell gradients in three directions.
It is this gradient vector, rS, that allowed us to find our way to the o↵ensive can.
The gradient of a vector, ru, had three directional derivatives on three components of velocity, giving

101
us a total of nine components. We wrote this as
2 3
6 7
6 @u @v @w 7
6 @x 7
6 @x @x 7
6 7
6 7
6 7
6 7
ru = 6
6 @u @v
7
@w 7 .
6 @y @y @y 7
6 7
6 7
6 7
6 7
6 7
4 @u @v @w 5
@z @z @z
And what about the units? The spatial derivative of velocity, du/dy for example, must be
 
m/s 1
=
m s
Our definition of a fluid is a material that continually deforms, or strains, under shear. In our discussion of
viscosity, we defined du/dy as a strain rate.
This all means that the tensor ru is actually a strain rate tensor. If we multiply a strain rate [1/s] by a
viscosity [Pa-s], we must wind up with a stress tensor. We’ll refer to this stress tensor, as we usually do, as
2 3
6 7
6 @u @v @w 7
6 @x 7
6 @x @x 7
6 7
6 7
6 7
6 7
⌧ = µ6
6 @u @v
7
@w 7 .
6 @y @y @y 7
6 7
6 7
6 7
6 7
6 7
4 @u @v @w 5
@z @z @z
The units of this stress tensor, as you would expect, are Pascals.
Back to the shear stresses from Figure 5.6. There is no short and simple way to consider all the shear
stresses. Let’s look from the top down at the cube from Figure 5.6. We’ll focus on the forces in the x-
direction due to the shear stresses on the z-surfaces. Remember that surfaces are defined by the direction
of their normal vector, so the z-surfaces are parallel to the x-y plane.
Let’s consider our point of interest, (x, y, z), to be in the center of the element as shown in the 2-D
drawing in Figure 5.7.

102
Figure 5.7: Top-down view of the cubical element from Figure 5.6. Forces in the x-direction due to shear on
the z-surfaces.
The shear force on the back surface would be
✓ ◆
@⌧zx dz
Fx,back = ⌧zx dx dy
@z 2
while the shear force on the front would be
✓ ◆
@⌧zx dz
Fx,front = ⌧zx + dx dy.
@z 2
Any acceleration is due to a net force, so we subtract the two and the ⌧zx disappears, leaving
Fx,z, net = Fx,front Fx,back

✓ ◆ ✓ ◆
@⌧zx dz @⌧zx dz
= dx dy + dx dy
@z 2 @z 2
✓ ◆
@⌧zx
= dx dy dz.
@z
The net force due to viscous stresses is related to the gradient of the stress tensor.
But remember, the stress tensor, ⌧, is itself a result of the gradient of the velocity field, ru. This means
that the net force due to viscous shear is somehow related to the second derivative of the velocity field,
Fviscous / r2 u
We could easily spend two lectures and several pages formally deriving this relationship, but we will leave

103
it as a conceptual understanding that the only viscous e↵ect on fluid acceleration, Du/Dt, must be due to
a di↵erence in stress, which must be from the second derivative5 of velocity.
(insert 5 pages of notes and 2 days of derivation here)6
The constant of proportionality for this relationship is the fluid viscosity µ. If we apply this force to our
di↵erential element the result is the following expression,
Fviscous = µr2 udxdydz
If we include the viscous force back into force original force balance,
X 
@u
F = ⇢ dx dy dz g rP dx dy dz + µr2 udxdydz = ⇢ + u · ru dx dy dz .
@t
Again we will divide both sides by the volume of the fluid element, do a little rearranging, and the result
is
h @u i
⇢ + u · ru = rP + µr2 u + ⇢g (5.10)
@t
This is called the Navier-Stokes Equation and is the di↵erential form of conservation of momentum.
The NS equation is one of the most famous equations in the world. It is used to describe the fluid motion
7
of viscous substances.
For an incompressible flow in Cartesian coordinates, we can break our key equations down into continuity
and the three components of momentum,
@u @v @w
+ + =0 (5.11a)
@x @y @z
✓ ◆ ✓ 2 ◆
@u @u @u @u @P @ u @2u @2u
⇢ +u +v +w = + ⇢ gx + µ + + (5.11b)
@t @x @y @z @x @x2 @y 2 @z 2
✓ ◆ ✓ 2 ◆
@v @v @v @v @P @ v @2v @2v
⇢ +u +v +w = + ⇢ gy + µ + + (5.11c)
@t @x @y @z @y @x2 @y 2 @z 2
✓ ◆ ✓ 2 ◆
@w @w @w @w @P @ w @2w @2w
⇢ +u +v +w = + ⇢ gz + µ + + (5.11d)
@t @x @y @z @z @x2 @y 2 @z 2
5 Also called the Laplacian, r2 , of velocity.

6 Or not.
7 If you have trouble remembering this you could have it tattooed on your right arm.

104
5.3 Canonical Flows
The following examples of flow regimes are called the Canonical Flows of fluid mechanics. These few
examples contain enough information to apply the laws of physics to fluid motion and can be applied to an
infinite number of scenarios. We have seen a lot of these already in homework, where we simply provide the
relationships for you. Here, we will use the continuity equation and the Navier-Stokes equation to solve for
flow properties.
5.3.1 Terms that Indicate Flow Characteristics
Prior to the canonical flows, let us spend just a moment discussing what each of these pieces of jargon mean.
The following terms provide vital information about the flow.
Fluids Buzzwords
@(...)
• Steady ! @t =0
The flow field velocity, temperature, pressure and all other properties are unchanging with time.
• Incompressible ! ⇢ = contant
The density of the fluid is constant at all points in time, and throughout all of the flow field.
@u
• Axisymmetric ! @✓ =0
For a given axis, and a perpendicular plane centered about the axis, and some distance from the center,
properties are constant for rotation through the plane at a fixed distance.
@u
• Fully Developed ! @z =0
If the fluid motion is in the z-direction and the flow is fully developed, then the velocity magnitude
and distribution is constant for all values of z.
• Parallel to surface ! u · n̂surf ace = 0
This is applied for flows within a constrained space and nonporous material boundaries. The component
of velocity that is perpendicular to the surface is zero. In other words, the velocity is only perpendicular
to the unit normal of the surface and therefore their dot product is zero.
• Isothermal ! T = constant
Temperature throughout the flow field is constant for all of time.

105
5.3.2 Couette Flow - Steady, Incompressible, Shear -driven, Isothermal, Two-

Dimensional Flow between Two Plates
Examples:
• Flow within a lubricated journal bearing
• Flow between a skate and ice
• Taylor-Couette flow
In general, our flow is represented by the two plates in Figure 5.8
Figure 5.8: Flow between two plates.
The flow is horizontal, so v and w are both zero. This leaves us with only two remaining equations,
@u @v @w
+ + =0
@x @y @z
✓ ◆ ✓ 2 ◆
@u @u @u @u @P :
⇠ 0 @ u @2u @2u
⇢ +u +v +w = ⇢ g⇠
+⇠ x+µ + + .
@t @x @y @z @x @x2 @y 2 @z 2
Since it is steady, we eliminate any time derivatives. Since there is no flow in the y-direction (parallel
flow) or z-direction (2-D), v and w go to zero. Further, since the flow is 2-D, we can also get rid of any
z-derivatives. Finally, there is no externally applied pressure to this flow8 , so we can eliminate the pressure
gradient. These simplifications give us

8 Think of the ice skate example.

106
0 0
@u @v @w◆
7
+ + ◆ =0
@x @y ◆@z
0 1 0 1
0 0 0 0 0
B @u @u 7
◆
@u ✓
@u C @P7
◆ B @ 2 u @ 2 u @ 2 u◆
7C
⇢B ◆ +w C ◆ + ⇢ gx + µ B ◆ C.
@ @t + u @x + v @y @z A
=
◆@x @ @x2 + @y 2 + ◆
@z 2 A
◆
But our simplifications to the continuity equation
@u
=0
@x
eliminate any x-derivatives, so must also be zero. This leaves a considerably more manageable momentum
equation,
✓ ◆
@2u
0=µ .
@y 2
It’s easier to look at the other way around, so
✓ ◆
@2u
µ =0 (5.12)
@y 2
Two important things come out of this. First, the velocity, u, is only a function of y, allowing me to
ditch the partial derivative. Second, I can just look at the viscous part of that equation and integrate9 it a
couple times.
d2 u
µ =0
dy 2
Z 2 Z
d u
µ 2 dy = 0 dy
dy
du
µ =C
dy
Z Z
du
µ dy = C dy
dy
u = Cy + D
We know that at the lower plate, y = 0, no slip applies and we have u = 0. Making this substitution,
0 = C(0) + D ! D = 0.
9 The word integrate here is a little dangerous. It’s an indefinite integral, which really is an antiderivative. In this case, a
tool to solve a di↵erential equation.

107
No-slip also applies at the upper plate, y = h, where the velocity must be U , so
U
U = Ch ! C= .
h
Finally, we have the velocity profile for shear-driven Couette flow,
U
u= y. (5.13)
h
This is admittedly a little uninteresting. We maybe even could have guessed it. However, we have carefully
started with a very general momentum equation, whittled it down according to our application, then solved
it to get an expected result. A very important first step in our study of di↵erential fluid mechanics.

108
5.3.3 Steady, Incompressible, Pressure-driven, Isothermal, Two-Dimensional

Flow between Two Plates
Examples:
• Brake fluid leaking past a piston
• Leak in any narrow gap
• Dyson “Vortex” hand dryer
Again, we have two parallel plates, but this time the plates are stationary and the flow is driven by an
applied pressure. We will make all the same assumptions, but this time we can’t ditch the pressure term.
The momentum equation winds up looking like
0 0
@u @v @w◆
7
+ + ◆ =0
@x @y ◆@z
0 1 0 1
0 0 0 0
B @u @u 7
◆
@u ✓
@u C @P B @ 2 u @ 2 u @ 2 u◆
7C
⇢B ◆ +w C + ⇢ gx + µ B ◆ C.
@ @t + u @x + v @y @z A
=
@x @ @x2 + @y 2 + ◆
@z 2 A
◆
And then rearranged to get

✓ ◆
@2u @P
µ = (5.14)
@y 2 @x
But wait. Take a close look at Equation 5.14. If I could shrink myself down to an infinitesimal swimmer,
I could swim in the y-direction at will, without changing x-position, and visa versa. So if we have an
x-derivative equal to a y-derivative, they must both be equal to the same thing: a constant.
✓ ◆
@2u @P
µ = = Constant
@y 2 @x

109
Again, I can ditch the partial derivatives and integrate10 with respect to y.
d2 u @P
µ =
dy 2 @x
Z 2 Z
d u @P
µ 2 dy = dy
dy @x
du @P
µ = y+D
dy @x
Z Z
du @P
µ dy = ( y + D) dy
dy @x
1 @P 2
µu = y + Dy + E
2 @x
1 @P 2 D
u= y + y+E
2µ @x µ
We’ll apply some known boundary conditions to solve for the constants. Starting with no slip at the lower
plate, y = 0, we get
1 @P 2 D
0= 0 + 0+E ! E = 0.
2µ @x µ
More interestingly, since the upper plate is no longer moving, we can apply no-slip there and get
1 @P 2 D 1 @P
0= h + h ! D= h.
2µ @x µ 2 @x
With our constants known, we can write out our velocity profile,
✓ ◆ ⇣ ⌘ ⇣y⌘
h2 @P y 2
u= (5.15)
2µ @x h h
5.3.4 Navier-Stokes in Cylindrical Coordinates
We will save you the toil and provide the cylindrical equations of fluid motion. Pumps and pipe flow are
two critically important concepts that a mechanical engineer needs to understand. Pipes, typically round,
are much better suited to cylindrical coordinates, where continuity is described by
@⇢
0= + r · (⇢u)
@t
The radial, tangential and axial directions are r, ✓ and z, respectively. In this coordinate system the velocity
is expressed as a sum of vector components in each of the r, ✓ and z directions. The velocity field will again
10 Indefinitely again.

110
be divergence free if the fluid is incompressible.
ur @ur 1 @u✓ @uz

r·u= + + + =0 (5.16)
r @r r @✓ @z
The vector form of the Navier-Stokes equation still looks quite compact.
h @u i
⇢ + u · ru = rP + µr2 u + ⇢g
@t
Breaking the vector equation of Navier-Stokes down into cylindrical components yields the following11 :
• r-momentum
✓ ◆
@ur @ur u✓ @ur u2✓ @ur
⇢ + ur + + uz =
@t @r r @✓ r @z
 ✓ ◆
@P 1@ @ur ur 1 @ 2 ur 2 @u✓ @ 2 ur
+ ⇢ gr + µ r 2
+ 2 + (5.17)
@r r @r @r r r @✓2 2
r @✓ @z 2
• ✓-momentum
✓ ◆
@u✓ @u✓ u✓ @u✓ ur u✓ @u✓
⇢ + ur + + + uz =
@t @r r @✓ r @z
 ✓ ◆
1 @P 1@ @u✓ u✓ 1 @ 2 u✓ 2 @ur @ 2 u✓
+ ⇢ g✓ + µ r 2
+ 2 + (5.18)
r @✓ r @r @r r r @✓2 2
r @✓ @z 2
• z-momentum
✓ ◆
@uz @uz u✓ @uz @uz
⇢ + ur + + uz =
@t @r r @✓ @z
 ✓ ◆
@P 1@ @uz 1 @ 2 uz @ 2 uz
+ ⇢ gz + µ r + 2 2
+ (5.19)
@z r @r @r r @✓ @z 2
5.3.5 Poiseuille Flow
The first of the canonical flows is the Poiseuille Flow 12 shown in Figure 5.9. Here we consider a horizontal
flow through a circular pipe. The fluid moves from left to right through the pipe at a steady rate. High
pressure Phigh is on the left hand side and low pressure is on the right hand side Plow . The fluid has some
11 A hideous three-headed beast
12 High probability that at least one of the instructors botches this spelling

111
viscosity µ, and is incompressible with density of ⇢. The flow is fully developed in the direction of motion,
isothermal and adiabatic. Solve for the velocity profile.
Figure 5.9: Horizontal flow through a circular pipe
We’ve already seen the parabolic velocity profile in the form of
✓ ◆
r2
uz (r) = Umax 1
R2
But does this match our description now that we have actually included some physics? Sure doesn’t look
like it will.
Let’s start with Navier-Stokes equation and immediately we can set a few terms equal to zero. The
0
h @u i
* 0
⇢ + u · ru = rP + µr2 u + ⇢g
@t
The flow is axisymmetric about the z-axis, and therefore the velocity is constant at given radius for all
values of ✓. The flow is fully developed in the z-direction. Therefore there is no z-dependence for any of
the velocity components. The fluid motion is parallel with the material boundary, which implies there is no
velocity in the ✓ or the r directions. So any term multiplied one of these velocity components also goes to
zero. Suddenly our expression begins to relax and we are only left with,
2 0 30
0 00 ✓ ◆ ⇢
>
h >
⇢
@u⇢
z ⇢⇢
u✓ @u>
z @u⇢>
⇢
z
i @P 61 @ @uz 1 @ 2⇢
uz @ 2 u✓
z7
⇢ ur⇢ + ⇢ + uz⇢ = + µ4 r + 2⇢ 2 + 5
⇢ @r ⇢
r @✓ ⇢ @z @z r @r @r ⇢
r @✓ @z 2
 ✓ ◆
@P 1@ @uz
0= +µ r
@z r @r @r
Only the z-momentum equation survives the cancellation event13 . After a bit of rearranging, we are left
13 The All Powerful Red Pen of Cancellation

112
with a governing di↵erential equation,
✓ ◆
@ @uz @P r
r = (5.20)
@r @r @z µ
Integrating over r removes the derivative on the right hand side.
Z ✓ ◆ Z
@ @uz @P r
r dr = dr
@r @r @z µ
@uz @P r2
r = + C1
@r @z 2µ
where C1 is a constant of integration. At this point divide both sides by r and integrate again to get
the velocity in the z-direction as a function of the radius. At this point we also realize that velocity in the
z-direction only depends on the radius, which turns partial derivative into total derivatives. The same holds
true for the pressure gradient term.
duz dP r C1
= +
dr dz 2µ r
Z Z
duz dP r C1
dr = + dr
dr dz 2µ r
dP r2
uz = + C1 ln(r) + C2
dz 4µ
Boundary conditions are necessary solve for the constants of integration. In our case the boundary
conditions become
1. No-slip at the wall ! uz (r = R) = 0
2. Solution must be finite ! |uz (r = 0)| < 1
No-slip velocity at fluid interfaces has been discussed previously and so we will assume you believe us
about the first boundary condition. The second boundary condition only limits the solution to be finite. As
you move closer to the center of the pipe, the natural log term plummets to 1. Therefore the constant
attached to the natural log term must be exactly zero.
C1 = 0

113
Next, we’ll apply the no-slip boundary condition at the wall.
dP R2
uz (r = R) = 0 = + C2
dz 4µ
We can now solve for the second constant of integration,
dP R2
C2 =
dz 4µ
Putting this back together yields and rearranging produces the following
dP r2 dP R2
uz (r) =
dz 4µ dz 4µ
@P 1
= R2 r 2
@z 4µ
✓ ◆
@P R2 r2
uz (r) = 1 (5.21)
@z 4µ R2
The maximum velocity occurs at the center of the pipe where r = 0.
@P R2
Umax =
@z 4µ
So I guess this stu↵ checks out.

114
Journal Bearing You may have left the last lecture thinking, “Now I know how a journal bearing works.
Cool!” but a few minutes later, you asked yourself, “But how does the oil get in there?” A pump, of course.
Now we need to add an applied pressure gradient to the shear driven flow. Figure 5.10 shows a schematic
of this, with a few possible velocity profiles shown.
Figure 5.10: Flow between two infinite plates with applied pressure gradient.
As a reminder, here are the Navier-Stokes equations in Cartesian form.
@u @v @w
+ + =0 (5.22)
@x @y @z
✓ ◆ ✓ 2 ◆
@u @u @u @u @P @ u @2u @2u
⇢ +u +v +w = + ⇢ gx + µ + + (5.23)
@t @x @y @z @x @x2 @y 2 @z 2
✓ ◆ ✓ 2 ◆
@v @v @v @v @P @ v @2v @2v
⇢ +u +v +w = + ⇢ gy + µ + + (5.24)
@t @x @y @z @y @x2 @y 2 @z 2
✓ ◆ ✓ 2 ◆
@w @w @w @w @P @ w @2w @2w
⇢ +u +v +w = + ⇢ gz + µ + + (5.25)
@t @x @y @z @z @x2 @y 2 @z 2
1. With your neighbor, agree14 on some simplifying assumptions.
2. Write down your simplified version(s) of continuity and momentum.
d2 u 1 @P
= (5.26)
ddy 2 µ @x
@P
= ⇢g (5.27)
@y
@P
=0 (5.28)
@z
3. Integrate to get an expression for the appropriate velocity? Be careful with constants in this case.
14 Not necessarily easy.

115
1 @P 2
u= y + Cy + D
2µ @x
4. State your boundary conditions and justify them.
No slip at top and bottom plate.
u(0) = 0
u(h) = U
5. Find an expression for the velocity profile in the oil.
1 @P
0= + C(0) + D ! D=0
2µ @x
1 @P U 1 @P
U= + C(h) ! C= h
2µ @x h 2µ @x
Uy 1 @P 2
u= + (y hy)
h 2µ @x

116
Body forces, coordinate systems and shear. Oh my!15
A thin film of water is flowing down an inclined plate, as shown in Figure 5.11.
Figure 5.11: Thin (h) film of water flowing down an inclined plate.
As a reminder, here are the Navier-Stokes equations in Cartesian form.
@u @v @w
+ + =0 (5.29)
@x @y @z
✓ ◆ ✓ ◆
@u @u @u @u @P @2u @2u @2u
⇢ +u +v +w = + ⇢ gx + µ + 2 + 2 (5.30)
@t @x @y @z @x @x2 @y @z
✓ ◆ ✓ 2 2
◆
@v @v @v @v @P @ v @ v @2v
⇢ +u +v +w = + ⇢ gy + µ + 2+ 2 (5.31)
@t @x @y @z @y @x2 @y @z
✓ ◆ ✓ 2 ◆
@w @w @w @w @P @ w @ w @2w
2
⇢ +u +v +w = + ⇢ gz + µ + + (5.32)
@t @x @y @z @z @x2 @y 2 @z 2
6. Carefully define a coordinate system.
see figure
7. Write down your simplified version(s) of continuity and momentum.
Assume 2-D, fully developed

@v
=0
@y
15 Hint, hint.

117
d2 u
0 = ⇢gx + µ (5.33)
dy 2
@P
0 = ⇢gy (5.34)
@y
@P
0= (5.35)
@z
Let’s take a close look at the gravity terms16 . Since we have a tilted coordinate system there will be both
x and y components of gravity.
g = g sin(✓)î g cos(✓)ĵ
m
g = 9.81 [ ]
s2
If we took our plate and tilted it so that ✓ = ⇡/2, then all of g will be in the positive x-direction. Let us
plug in the expression for gx into the momentum equation for x-direction.
d2 u
0 = ⇢[g sin(✓)] + µ
dy 2
8. Integrate to get an expression for the appropriate velocity? Be careful with constants in this case.
d2 u ⇢g sin ✓
=
dy 2 µ
⇢g sin ✓ 2
u= y + Cy + D
2µ
9. State your boundary conditions and justify them.
No slip at plate and no shear at the water-air interface
u(0) = 0
du
=0
dy h
16 I’ll try not to botch it this time.

118
10. Find an expression for the velocity profile in the water.
D=0
⇢g sin ✓
C= h
µ
✓ ◆
⇢g sin ✓ y2
u= hy
µ 2

119
5.3.6 Strategy for di↵erential fluid analysis
The purpose of this type of analysis is often to determine the velocity profile. The following procedure will
aid in your endeavors.
1. Choose a good coordinate system
• Choosing a coordinate system wisely will often simplify the problem. For flows in a single direction,
choose coordinate system where one axis is parallel with the direction of the flow.
2. Apply continuity equation
Continuity equation is our mass balance. The full form is:
@⇢
0= + r · (⇢u)
@t
If the fluid is incompressible then the density is constant. This simplifies the expression to
0 = 0 + r · (⇢u)
0 = ⇢ (r · u)
@u @v @w
0=r·u= + +
@x @y @z
3. Apply Navier-Stokes equation
Vector form of Navier-Stokes is:
h @u i
⇢ + u · ru = rP + µr2 u + ⇢g
| @t {z } | {z
F
}
ma
Remember that the left hand side is mass times acceleration per unit volume, and the right hand side
is force per unit volume. Break the vector equation down into components, and kill terms that are
equal to zero. The fluid motion may be in a single direction, in which case the direction of the flow
will be the component of Navier-Stokes that is most interesting to us. Use the list of Fluids Buzzwords
to help simplify the expressions.
4. Solve expression
From the simplified Navier-Stokes expression, isolate variable of interest and solve the expression. This
may require one or two integrations that will leave constants of integration. Solve the constants of

120
integration using boundary conditions.
5. Boundary conditions
• No-slip at a material boundary (plates, channel or pipe walls, fluid-fluid interface, etc.)
• Equal shear at fluid-fluid interface
• No shear at free surface, i.e. zero velocity gradient (as in the case with the inclined plate)
• Finite solution (natural log terms containing r must have coefficients that are exactly zero)

121
Chapter 6
External flows
Figure 6.1: CFD simulation with steamlines over airplane wing - ANSYS
To this point, we have used the integral form of the momentum equation to find reaction forces based
on known flow fields. We made some pretty big assumptions and simplifications regarding the flow field,
but it allowed us to solve some interesting problems. Then we used the di↵erential form of the momentum
equations to actually find the flow field based on some known boundary conditions - namely walls.
Now we will take a look at some very simple external flows. The first, and simplest will be the incom-
pressible flow over a flat plate.

122
6.1 von Karman Drag Integral
In 1921, Theodore von Karman1 developed a method to calculate the drag on a flat plate. Figure 6.2 shows
a free stream at U1 approaching and then flowing over a sharp flat plate.
Figure 6.2: Free stream flow of U1 approaching a sharp flat plate. The region of influence of the plate, the
boundary layer, grows as the flow moves down the plate.
A fluid with velocity U1 approaches a flat plate with a sharp edge. The no-slip condition begins at the
leading edge of the plate. As the fluid flows down the plate, the viscous forces retard the flow farther and
farther from the plate. This flow is obviously not fully developed, as the velocity profile is changing as we
move downstream.
One of the first and most important questions we must answer about this flow is What is the drag force
on the plate? There is a velocity gradient at the plate, the fluid has some viscosity, so we would expect some
shear. This shear, multiplied by the area of the plate, would result in some drag force.
Compared to our fully developed, wall-bounded flows, this is tricky. What are the boundary conditions?
Or for that matter, where are the boundary conditions? von Karman came up with a clever way to use the
integral form of the momentum equation to come up with the drag force. Figure 6.3 shows the same flow
over the same plate, but with von Karman’s control volume.
All the flow entering the left control surface enters with velocity U1 . Because the viscous forces retard the
flow within the boundary layer, the same mass needs a larger area at the exit. The upper control surface is a
streamline that is mostly U1 , but includes a little vertical velocity to account for the decreasing x-velocity.
Because this upper streamline is just outside the boundary layer, the streamline just below is also mostly
U1 plus a little vertical. Since there is no velocity gradient at our upper control surface, any forces in the
x-direction must be the drag, FD at the plate. If we call b the dimension in and out of the page,
1 While at CalTech, von Karman mentored a student named Andrew Fejer, who went on to do pioneering work in unsteady
aerodynamics and turbulence. While at the Illinois Institute of Technology, Fejer took on a graduate student named Hassan
Nagib, who became one of the world’s experts on turbulent boundary layers. Still at IIT, Nagib took on a struggling graduate
student who went on to be the tallest graduate student in the lab.

123
Figure 6.3: Boundary layer over flat plate with von Karman’s control volume. CV is bounded on the left
and right by vertical surfaces at x = 0 and x = x. Bottom surface is the flat plate itself. Top surface follows
a streamline just outside the boundary layer.
X
Fx = FD
X Z Z
Fx = ⇢ u(x = 0)(u · n̂) dA + ⇢ u(x = L)(u · n̂) dA
CS,in CS,out
Z h Z
=⇢ U1 ( U1 )b dy + ⇢ u(x = L)(u(x = L))b dy
0 0
we can at least evaluate the first integral, to get

Z
2
FD = ⇢U1 bh ⇢b u2 dy |x=L (6.1)
0
So now we have our drag on the plate, FD , in terms of h and , but we don’t know h and . We do know that
any mass entering on the left CS must exit on the right CS. How do we know this? Because the upper CS
is a streamline. We know streamlines are parallel to the flow, so flow can’t cross a streamline. Conserving
mass then,
Z h Z
⇢ U1 b dy = ⇢ ub dy
0 0
Z
U1 h = u dy |x=L (6.2)
0
We put Equation 6.2 back into the expression for the drag force, Equation 6.1, and get von Karman boundary

124
layer momentum integral,

Z
FD = ⇢b u(U1 u) dy |x=L . (6.3)
0
Von Karman, like most fluid dynamicists, liked dimensionless parameters.2 Related to this momentum
integral, he defined the momentum thickness, ⇥,3
Z ✓ ◆
u u
⇥= 1 dy. (6.4)
0 U1 U1
It may not be obvious why this is called a momentum thickness. Dimensionally, it does have the right
units, but beyond that, it doesn’t make much sense. If we look back to how it fits in with Equation 6.3, it
leads to a very useful result regarding the total drag force,
2
FD = ⇢bU1 ⇥ (6.5)
The drag force becomes a function of
• Fluid density: makes sense
• Fluid velocity: also makes sense
• “Thickness” of the boundary layer: the thicker the boundary layer grows, the more drag it created as
it grew!
There is one more way to look at the momentum thickness. It is the thickness of a layer of fluid flowing
at U1 that has the same momentum lost as the boundary layer developed.
This seems like a lot of work4 to get the drag force. Why didn’t we just integrate the shear from the
start of the plate to L? We could have.
Z L
FD = b ⌧w (x) dx,
0
where ⌧w is the wall shear at location x and b is still the distance in and out of the page. Alternatively,
dFD
= b⌧w . (6.6)
dx
2 Much more on these in a couple weeks.
3 THeta, for THickness.
4 And it is

125
If we go back and di↵erentiate Equation 6.5, we get
dFD 2 d⇥
= ⇢bU1 . (6.7)
dx dx
Oooooh... This is getting interesting. By equating Equations 6.6 and 6.7, we get the momentum integral
relation for boundary layers,

d⇥
⌧w = ⇢U 2 (6.8)
dx
Equation 6.8 tells us that the local shear is the rate (spatially) at which the momentum thickness is
growing.
We’re losing sight of what’s going on here, for sure. What does this all mean? Hang in there, we’re
almost ready to look at an example.
6.2 Prandtl Boundary Layer Equations
As long as we’re taking a trip down the Fluids History Lane, let’s make one more stop. Theodore von
Karman’s PhD advisor was a guy named Ludwig Prandtl5 Prandtl was a genius, but not an impractical guy.
He took a look at the fluid in and around the boundary layer and decided a number of things:
• v⌧u
@u @u
• @x ⌧ @y
@v @v
• @x ⌧ @y
He gave these observations to his ME 363 students and asked them to simplify the steady, 2-D y-momentum
equation. Most students turned in something that looks like
0 0 it 1
small 1
small forget meh
2◆
7
B ✓
@v @v7
◆ C @P B @ v @ 2 v◆
7C
⇢ @u +v◆ A = + µB ◆ + ◆2 C
@x ◆@y @y @◆@x2 @y A
◆
which e↵ectively means that there isn’t much of a wall-normal6 pressure gradient,
@P
⇡ 0.
@y
5 You’ll learn a lot about him in Heat Transfer.
6 Perpendicular to the wall, in this case, y-direction.

126
The y-momentum equation goes away except to tell us that the pressure gradient must only be in the
x-direction,
P = P (x).
Depending on what the flow is, a wing or a golf ball for instance, there may be some change in velocity. We
can rearrange Bernoulli7 to show that the pressure gradient in the free stream could be
dP du
= ⇢U1 .
dx dx
This is basically stating that if the flow in the free stream is accelerating (U1 dU1 /dx) it must be due to a
pressure gradient (dP/dx).
Finally, we’re left with a simplified set of equations that describe8 flow in the 2-D incompressible boundary
layer. These are basically the continuity equation and a very simplified x-momentum equation.
@u @v
+ =0 (6.9)
@x @y
@u @u dU1 1 @⌧
u +v ⇡ U1 + (6.10)
@x @y dx ⇢ @y
where (6.11)
@u
⌧ =µ
@y
6.3 Blasius Boundary Layer Solution
But we still don’t know the size or profile within this boundary layer. Prandtl had another student named
Paul Blasius. Blasius recognized that even after his advisors simplifications, we’re still left with a set of
non-linear elliptic partial di↵erential equations. If only he could shu✏e things around and get these in terms
of one variable, instead of four.
Blasius proposed just this. He defined a new dimensionless variable, ⌘,
✓ ◆1/2
⇢U1
⌘=y (6.12)
µx
7 But not right here
8 If not quite govern

127
and he also showed that the relative velocity in the boundary layer, u/U1 , is only a function of ⌘. Specifically,
u
= f 0 (⌘)
U1
and then he rewrote Prandtl’s simplified equations, Equations 6.9 and 6.10, as a single function, f , in terms
of ⌘,
d3 f d2 f
3
+ f 2 = 0. (6.13)
d⌘ d⌘
We still have to apply the same two boundary conditions: no slip at the wall and no shear where the
boundary layer meets the free stream.
At y = 0, u = 0 (u/U = 0) ! f 0 (0) = 0 (6.14)
At y = 1 u = U ! f 0 (1) = 1 (6.15)
Equation 6.13 still can’t be solved analytically, but it at least allows us to tabulate values of u/U at
di↵erent values of ⌘.
Table 6.1: Tabulated Laminar Boundary Layer Solution. After values of ⌘ over 5.0, the boundary layer
velocity, begins to very slowly approach the free stream velocity, U .
q
⇢U u
⌘=y µx f f0 = U f 00
0 0 0 0.3321
0.5 0.0415 0.1659 0.3309
1.0 0.1656 0.3298 0.3230
1.5 0.3701 0.4868 0.3026
2.0 0.6500 0.6298 0.2668
2.5 0.9963 0.7513 0.2174
3.0 1.3968 0.8460 0.1614
3.5 1.8377 0.9130 0.1078
4.0 2.3057 0.9555 0.0642
4.5 2.7901 0.9795 0.0340
5.0 3.2833 0.9915 0.0159
5.5 3.7806 0.9969 0.0066
6.0 4.2796 0.9990 0.0024
6.5 4.7793 0.9997 0.0008
7.0 5.2792 0.9999 0.0002
7.5 5.7792 1.0000 0.0001
8.0 6.2792 1.0000 0.0000
We may want to know how thick the boundary layer is. We could look in the tabulated data and find

128
that u/U is 0.99 when ⌘ is 5.0.9 Putting these back in the definition of ⌘, Equation 6.12,
✓ ◆1/2
U⇢
5.0 = 99
µx
We can rearrange this to

99 5.0
=⇣ ⌘1/2 (6.16)
x ⇢U1 x
µ
The denominator on the right had side looks familiar. We could rewrite this again to get an explicit form of
the 99 boundary layer thickness,

5.0 x
99 =p
Rex
where Rex is the Reynolds number defined with length along the plate as the length scale of importance10
⇢U1 x
Rex = .
µ
There is some more interesting information buried in here. Say we want the wall shear, which is where
this all started.
du
⌧w = µ
dy y=0
s
U 1 ⇢ d2 f
= µU |
µx d⌘ 2 ⌘=0
r
⇢µU1
= 0.332 U
x
2
0.332⇢U1
= p (6.17)
Rex
But WAIT!!! This looks a whole lot like Equation 6.7. Let’s make Equation 6.17 look a little bit more like
Equation 6.7,
✓ ◆
0.332 2
⌧w = p ⇢U1 (6.18)
Rex
Hold on, hold on. If we put a factor of 1/2 in here, this will look very familiar.
✓ ◆✓ ◆
0.664 1 2
⌧w = p ⇢U
Re 2 1
| {z x } | {z }
coefficient K.E.
9 It’s important to remember here that ⌘ has absolutely no physical meaning within the flow. We use ⌘ to find what we
want, then work backwards to get physical parameters.

10 More on this, soon.

129
This is a very important result. We can get the local wall shear by multiplying the kinetic energy of the free
stream by some coefficient. We’ll call this the wall shear stress coefficient or local skin friction coefficient,
cf .
If we want the drag force of the whole plate, we just integrate our expression for ⌧ from x = 0 ! L.
After some algebra, we get

1.29
Cf = p (6.19)
ReL
where ReL is the Reynolds number based on the entire length of the plate. If it appears that the numerator
p
simply doubled, you’re right. This is a result of integrating 1/ Re

130
6.4 Turbulent Boundary Layers
As you walked to class this morning, you were (literally) surrounded by fluid dynamics. From the flow
through your heart valves to your breath condensing in the cold air. The wind, the clouds, the snow falling,
air flowing around the bus driving by, the fuel, air oil and coolant flowing throughout the engine of the bus
driving by. We should consider two ideas:
• To understand fluids is to understand the world around you.
• Most of the flow in and around you is turbulent. Not laminar.
Hence, a brief discussion of turbulence and turbulent boundary layers.11 We figured out that a laminar
boundary layer has some thickness. We can describe the thickness in a few di↵erent ways, that give us
di↵erent insight into what is physically happening:
• Boundary Layer (or Distrubance) Thickness, 99
• Momentum Thickness, ⇥ ! The distance the actual surface must be displaced, so that if we had
an ideal fluid, then the rate of momentum flow within the new boundary would be the same as it is
for the real fluid.

Z ✓ ◆
u u
⇥= 1 dy. (6.20)
0 U1 U1
• Displacement Thickness, *
! The distance the actual surface must be displaced, so that if we had an ideal fluid, then the mass
flow within the new boundary would be the same as it is for the real fluid.
Z ✓ ◆
⇤ u
= 1 dy. (6.21)
0 U1
We glossed over many of the details, but we were able to get a feel for what Blasius did, and that his
boundary layer thickness took the form

⌘x
=p
Rex
where the specific value for ⌘ comes from Table 6.1. For example if we are interested in ⌘ at which u/U is
0.99, we choose 5.0.

11 To be clear, many, many brilliant people have dedicated their entire careers to turbulent boundary layers. Many textbooks
have been written. Courses taught. Partnerships forged. Friendships ended. To cover “turbulent boundary layers” in 50
minutes is comically unrealistic. Yet we try.

131
Figure 6.4: Displacement thickness and momentum thickness
Years of experiments have led to some similar relationships for turbulent boundary layers.
0.38x
x = 1/5
(6.22)
(Rex )
0.059
Cf,x = 1/5
(6.23)
(Rex )
These equations can be used for Reynolds number in the following range:
5 ⇥ 105 < Rex < 1 ⇥ 107
Figure 6.5: Laminar and turbulent boundary layer thicknesses, 99 , in air at 20 C and 100 kPa along a flat
plate with no pressure gradient.

132
Figure 6.6: Laminar and turbulent viscous drag coefficients in air at 20 C and 100 kPa along a flat plate
with no pressure gradient.
Figure 6.7: Cartoon showing boundary layer concepts as the transition from laminar to turbulent occurs.
Not to scale. Image source uncertain.
Figure 6.8: Turbulent boundary layers begin as laminar boundary layers and a transition zone before they
become fully turbulent.

133
Figure 6.9: Fluid draining through a rectangular channel
Figure 6.10: Fluid draining through a circular pipe

134
Figure 6.11: Cross flow past a sphere (or cylinder)
• Trace where you would expect the streamlines on the left hand side to go as they move by the sphere.
• Indicate regions where you think the velocity is zero and where the velocity is greater than the free
stream velocity (u > U1 ).
• Using Bernoulli
1 1
P1 + ⇢u21 + ⇢gz1 = P2 + ⇢u22 + ⇢gz2
2 2
relate the velocities that you determined in the previous part, to the pressure at di↵erent points along
⇡ 3⇡
the sphere. Consider at least points of ✓ = (0, 2, ⇡, 2 , 2⇡)

135
Figure 6.12: Cross flow past a strange looking shape
• Trace where you would expect the streamlines on the left hand side to go as they move by the airfoil.
• Indicate regions where you think the velocity is zero and where the velocity is greater than the free
stream velocity (u > U1 ).
• Using Bernoulli
1 1
P1 + ⇢u21 + ⇢gz1 = P2 + ⇢u22 + ⇢gz2
2 2
relate the velocities that you determined in the previous part, to the pressure at di↵erent points along
the air foil.
• Do you notice anything about the pressure on top of the airfoil compared to the pressure on the bottom
of the air foil? What will this di↵erence in pressure result in?

136
Figure 6.13: Pressure distribution on a cambered airfoil. Note that the majority of the lift comes from the
leading edge of the airfoil.

137
6.5 Lift and Drag
To truly understand lift, we would need to study topics including potential flow, conformal transformations,
complex integration, vorticity and circulation. We’ll touch on potential flow in a couple weeks, but we won’t
have time for the fancy math. Nonetheless, we want to do a little better than the explanation in Figure
6.14.12
Figure 6.14: Typical explanation for powered flight.
12 Figure 6.14 is one small step short of crediting Bernoulli with powered fight.

138
Let’s channel our inner Statics student and sketch some of the key forces on this Cessna 172N.
Figure 6.15: Side view of a Cessna 172N
And some of the key equations:
X
Fx = 0 ! Thrust = Drag (6.24)
X
Fy = 0 ! Lift = Weight (6.25)
X
M=0 ! Stability (6.26)
We talked about drag, and two general forms of drag for fully immersed bodies:
• Viscous Drag: Von Karman integral, wall shear, skin friction coefficient, etc.
• Pressure Drag: Uneven pressure distribution, wake, separation, adverse pressure gradient, golf ball
For a streamlined body such as an airfoil, what type of drag do you think dominates?
As such, in our non-dimensionalized drag force, which L2 or area do you think matters most?
FD
CD = 1 2A
(6.27)
2 ⇢u
Let’s take a look at some airfoil geometry.

139
Figure 6.16: Some geometric definitions of a symmetric airfoil. Image from White.
The area of interest for an airfoil is called the planform area.
Apl = bc (span x chord) (6.28)
The drag coefficient is

FD
CD = 1 2A
2 ⇢U
We can define a similar coefficient of lift, CL . As you might guess
FL
CL = 1 2
.
2 ⇢U Apl
Once we start playing with the angle of attack, ↵, we see that both FD and FL are
CD = f (Re, ↵)
CL = f (Re, ↵)

140
(b) Inverted image of smoke lines over an airfoil at a high

(a) Inverted image of smoke lines over an airfoil at low angle of attack. Note the net deflection of the stream-
angle of attack. Screenshot from NCFMF. lines downward. Screenshot from NCFMF.
Figure 6.17: Smoke lines around airfoils for low and high angles of attack
Figure 6.18: Lift and drag on a NACA 0009 symmetric airfoil. As lift increases, so does drag.

141
Figure 6.19: Cessna above Lake Michigan, making the 100 mile flight from Milwaukee, WI to Grand Haven,
MI.
You and a buddy are out in your Cessna flying above Lake Michigan. You’re flying at a height of 12,000
ft. and the trip is 100 miles. You reach a little over half way across the lake when the engine fails. What the
heck are we going to do? Maybe if we tip the nose of the plane down just a bit, we can glide at a constant
velocity safely to the other side of the lake.
We’re still moving at constant velocity, no acceleration and therefore and the forces must still balance to
zero.
X
Fx = 0 = FD cos(✓) + FL sin(✓) (6.29)
X
Fy = 0 = Fg + FD sin(✓) + FL cos(✓) (6.30)
We know our drag force is

1 2
FD = ⇢U CD Apl ,
2
and the lift force is

1 2
FL = ⇢U CL Apl ,
2
Take the x-direction force balance and equate the two terms
FD cos(✓) = FL sin(✓)

142
Figure 6.20: Cessna now gliding (hopefully to safety) after the engine has failed.
Now, plug in the terms for drag force and lift force, and cancel like terms
1 2 1
⇢u CD Apl cos(✓) = ⇢u2 CL Apl sin(✓)
2 2
CD cos(✓) = CL sin(✓)
We can rearrange one more time to get
sin(✓) CD
= tan(✓) = (6.31)
cos(✓) CL
Equation 6.31 is called the glide ratio and applies to non-propelled aerial motion (often called gliding). What
are some things we can do to maximize glide ratio? Or do we even want to?
Figure 6.21: Some more airfoil geometry, including camber. This airfoil is no longer symmetric, and would
behave di↵erently if flying upside down. Image from, you got it, Wikimedia Commons.

143
Including camber, the coefficient of lift, CL is predicted to be
✓ ◆
h
CL ⇡ 2⇡ sin ↵ + 2
c
where h/c is the ratio of camber, h, to chord length, c.13 For a foil to produce lift, it needs either a positive
angle of attack or a positive camber, or both.
But lift doesn’t come for free. Figure 6.18 shows that with increasing angle of attack, both lift and drag
increase.
If one wanted to get the most lift (leading edge) with the least drag (planform), you would make a wing
with a high aspect ratio. That is a very long span and a very short chord. Figure 6.22
Figure 6.22: The Rutan Voyager and the P-51 Mustang. The two aircraft had very di↵erent missions and
hence very di↵erent wing designs. Photo credits NASA (Voyager) and Ron Fernuik (P-51). Printed under
fair use.
13 Another handy dimensionless ratio.

144
Chapter 7
Dimensional Analysis and Similitude
Figure 7.1: Proper scaling between model and full scale must be done to accurately correlate model results
and full-scale e↵ects.
7.1 Introduction
Throughout the semester, we have emphasized “units” as a way to check our work. For instance, when we
look at Newton’s Law, we can check our units to make sure we got it right,
F = ma (7.1)
hmi
[N ] = [kg] 2 (7.2)
  s
kg m kg m
= . (7.3)
s2 s2

145
Check. We got everything there.
We all pretty much have F = ma committed to memory by now, but when we work with something a
little trickier, for instance pipe flow, we may miss something. You may recognize
1 dP 2
u= (r R2 )
4µ dx
as the velocity profile of pressure driven pipe flow. It might be easier to recognize if we do a little rearranging,
✓ ◆✓ ⇣ r ⌘2 ◆
R2 dP
u= 1 (7.4)
4µ dx R
hmi 
[m2 ] Pa
= ([-] [-]). (7.5)
s [Pa-s] m
⇣ ⌘
r 2
We factored out R2 , leaving a fairly convenient 1 R at the end of the equation. Why is this conve-
nient? Because it doesn’t have units. It answers the question
How far, relatively, am I from the center of this pipe?
It doesn’t matter if this is a hypodermic needle or the Alaskan pipeline. If your are half-way from the
center to the wall, you’re half way from the center to the wall. Just by algebraic simplification, you already
discovered the value of dimensionless parameters. In this case, r/R.
Keeping track of units also can give you insight into a problem. Consider a cantilevered beam, such as
the one shown in Figure 7.2.
Figure 7.2: A cantilevered beam deflecting by some amount, , under some load, F .
You know that the sti↵ness of the beam, k [N/m], is probably a function of three parameters, the second
moment of area, I [m4 ], the Young’s modulus of the material, E [Pa], and the length of the beam, L [m]. But
you aren’t sure if it is linear in I and quadratic in E, or visa versa, or what. Using some simple dimensional
analysis, you can figure out how this works.
k = f (L, I, E) = La I b E c

146
hNi h ih ih i
= ma (m4 )b Pac
m
a= 2
b=1
c=1
So the result is,

EI
k=
L3
7.2 Experimental Savings
Imagine you are working for Boeing1 and you need to figure out the drag on a new aircraft. Your knowledge
of fluid mechanics tells you that the primary variables for the drag force are the length, L, of the aircraft,
the velocity, U , of the air, the density, ⇢, of the air and the viscosity, µ, of the air. So the drag force, FD , is
Fd = f (L, U, ⇢, µ).
The shape of the plane is complex enough that you can’t just use an integral momentum equation or the
di↵erential Navier-Stokes equations, so you’ll run an experiment. To be really confident in your results, you
need ten data points for each parameter. That is, ten lengths of aircraft models. For each of the ten models,
you need ten velocities. That’s one hundred experiments so far. For each of those, you need ten densities,
and....
You plan to use the NDF in Chicago to run these experiments. It costs $500 to push the start button,
so your experiments will run you a cool $500(104 ).2
We will see that the drag force does, in fact, depend on L, U , ⇢ and µ, but very specifically as
FD ⇢U L
= g( )
⇢U 2 L2 µ
or if we define
FD = CF ⇢U 2 L2 ,
1 In which case, you may want to dust o↵ your resume about now.
2 That’s 5 million dollars.

147
we get
CF = g(Re).
We can develop a very general drag coefficient that is only a function of Reynolds number. Ten runs, five
grand, done.

148
7.2.1 Comparing Data
We have looked briefly at the laminar boundary layer and how the velocity, u, varies with distance from a flat
plate. Thousands of people at dozens of labs around the world perform experiments in turbulent boundary
layers. This is valuable research in drag reduction which could have significant global energy implications.
But how do we compare results from a wind tunnel in Stockholm to a wind tunnel in Melbourne? We us
dimensional analysis.
Figure 7.3 shows some velocity profile data from experiments done in Stockholm. The measurements
were performed at di↵erent x-positions along the plate and at di↵erent free stream velocities. The data are
plotted as you would expect. The distance from the wall, y, in meters, is plotted on the abscissa and the
velocity, u, in meters per second, is plotted on the ordinate.
Figure 7.3: Some velocity profiles in a turbulent boundary layer from the MTL wind tunnel at the Royal
Institute or Technology in Stockholm, Sweden. The Reynolds numbers are between 105 and 108
Clearly, the boundary layer changes at di↵erent locations di↵erent free stream velocities. However, if we
define two new non-dimensional in terms of wall units3 variables,
u
u+ = (7.6)
u⌧
r
⌧w
u⌧ = (7.7)
⇢
y⇢3/2
y+ = (7.8)
µ⌧ 1/2
The logarithmic Law of the Wall proposed by von Karman in 1930 says that
1
u+ = ln y + + B (7.9)

3 Wall units are way of expressing lengths in a dimensionless manner.

149
where  is the von Karman constant and B is a constant depending on the roughness of the wall. Experimental
studies have led to the conclusions that
 = .4 B=5
Plugging in B gives us
1
u+ = ln y + + 5 (7.10)

Now, let’s plot the exact same data but in terms of wall units, shown in Figure 7.4.
Figure 7.4: Same velocity profile data as plotted in Figure 7.3, plotted in non-dimensional wall units showing
that y + = f (u+ ).
The boundary layer profiles all seem to fall on top of each other. That is, they collapse onto one curve.
In other words, there is one fundamental relationship that relates the non-dimensional velocity, u+ , to the
non-dimensional distance from the plate, y + . Now researchers from Chicago, Princeton, New Hampshire,
Stockholm and Melbourne can compare their results and hopefully figure out how to reduce the drag on the
aircraft.

150
7.3 Dimensional Analysis
When designing new automobiles, airplanes, rockets and aircraft carriers, experiments are necessary to
determine the prototype performance. Small models of the prototypes can be used in the lab. The working
conditions of the machine can be simulated in the lab and useful information can be gained from the
performance of the small scale model. The order of magnitude is drastically di↵erent between the small scale
and real scale versions. In order to properly compare the two settings, we will need to consider dimensionless
quantities.
Dimensional analysis is a mathematical tool that we will use to organize variables of interest into di-
mensionless groups. Dimensional analysis is derived from dimensional homogeneity, which states that units
must be consistent for any added terms in an equation. This fact implies that the units on either side of
the equal sign must also be the same. Physical variables can be described in terms of primary dimensions4
: mass M , length L and time T .
7.4 Buckingham ⇧ Theorem
The scientist Edgar Buckingham was a daring pioneer in the field of dimensional analysis. He was among
the first to formalize the technique. His idea was simple: to search for a dimensionless quantity that could
relate variables of interest within a specific expression.
7.4.1 Procedure
1. Define physical variables
Determine the number n of variables that influence the physical scenario. Variables that can be
described as a combination of other variables are not included in this count. We then determine the
number m of primary dimensions that are used in the n variables. By subtracting the number of
primary dimensions used from the number of variables yields the number of the dimensionless ⇧ terms
that describe physical phenomena.
n = Number of variables
m = Number of primary dimensions used in the n variables
⇧ = (n m) = Dimensionless group or number that describes a phenomenon

4 The other primary dimensions are ampere, kelvin, mol, and candela.

151
2. Select the repeating variables
From the n key variables, choose m of them that collectively contain all m primary dimensions. The
one variable is not in this collection is called the q group.
3. ⇧ terms
The dimensionless ⇧ term will be expressed as each of variables within the m collection raised to an
unknown exponential power, multiplied by q. There could be several combinations to choose from. In
general choose the easiest. For some phenomena that is a function of q, x, y and z we have determined
that all m primary dimensions are present in a combination of the m number of terms: x, y and z.
Then the ⇧ group will look like:
⇧ = xa y b z c q (7.11)
4. Dimensional Analysis
Plug in the base units for all variables in the expression. Then replace base units with primary
dimensions5 mass M , length L and time T . Since we want ⇧ to be dimensionless, we know that the
result of the multiplied term must have mass M , length L and time T all raised to the zeroth power.
⇧ = xa y b z c q = M 0 L0 T 0 (7.12)
Set the exponents of each dimension equal to zero separately. Then consider the contribution from
exponents a, b, and c for each of the separated dimensions.
5 Force F , length L and time T also works.

152
7.4.2 Examples
If you aren’t a little woozy right now, you’re a genius and you should start pursuing your Nobel prize. For
the rest of us, some examples will help.
Drag Force on a Plane
We deduced that the drag force on a plane, or the wing, or the tail, or whatever, is
Fd = f (L, U, ⇢, µ).
So let’s see if we can figure out some non-dimensional groups that explain this relationship.
1. Define physical variables
We have five variables, Fd , L, U , ⇢ and µ in which Mass, Length and Time are all represented.
Therefore, we expect
5 3=2 ⇧ groups.
2. Select the repeating variables
This requires a little subtlety, but in general, don’t mess with the variables we are most interested in.
Certainly, I want to know Fdrag , so I’ll leave that out. Most of the common non-dimensional numbers6
common in fluid mechanics include a density, a velocity and a length, so I’ll pick those three. This
leaves viscosity, µ, as my other non-repeating variable.
3. Pi Terms
Now I know how I’m going to get my two non-dimensional Pi groups.
⇧1 = ⇢a U b Lc (Fd ) (7.13)
= M 0 L0 T 0 (7.14)
and (7.15)
⇧2 = ⇢d U e Lf (µ) (7.16)
= M 0 L0 T 0 (7.17)
6 Or in other words, based on experience.

153
4. Dimensional Analysis
We then set the exponents on M, L and T each equal to zeros and solve. For the first Pi group we get
a= 1 b= 2 and c= 2,
yielding our first Pi group as

Fd
⇧1 = .
⇢U 2 L2
This group is commonly called the force coefficient.
For the second Pi group we get
a= 1 b= 1 and c= 1,
yielding our second Pi group as

µ
⇧2 = .
⇢U L
This a familiar looking group of variables. It is the Reynolds number, Re, upside down.
A formal application of the Buckingham Pi theorem would require us to start with
g(⇧1 , ⇧2 ) = 0,
and from there, end up with
⇧1 = G1 (⇧2 ).
That is, any Pi group can be written as a function of the others.
We will skip the mathematical formalities7 and believe that ⇧1 , the non-dimensional force, can be expressed
as a function of ⇧2 , the upside down Reynolds number.
As long as we’re at it, basic algebra tells us that if we can write ⇧1 as a function of ⇧2 , we could also
write it as a function of ⇧2 1 , so finally, we have
✓ ◆
Fd ⇢U L
=f .
⇢U 2 L2 µ
7 Which is painful for both instructors, but we don’t have time to learn it and teach it.

154
Now, we can run a bunch8 of experiments, and come up with a plot something like that of Figure 7.5.
Assuming we’re working in a typical wind tunnel and we only have one scale model, we would vary the
velocity as a means to vary the Reynolds number.
Figure 7.5: Some made up data relating the the drag coefficient, CD , to the Reynolds number, Re.
But, we don’t want any Reynolds number. We want the conditions at which our plane will be flying.
But the beauty of our non-dimensional data is that it doesn’t matter how big or small the wing is, for any
given Re, we know exactly9 what the CD will be. Figure 7.6
Figure 7.6: A specific Reynolds number corresponding to a specific drag coefficient. Any wing operating at
this specific Re will experience this specific CD .
Since we know the conditions (velocity, altitude, pressure, density, temperature, etc) at which our real
plane will fly, we know the Reynolds number for the actual flight conditions. Lucky for us, our experimental
8A bunch is an informal way of saying, “Enough to get a good looking line.”
9 To within experimental uncertainty.

155
data tells us what the drag coefficient will be. Since any wing at a given Re experiences the same CD , we
can write
if
(Re)model = (Re)actual
then
✓ ◆ ✓ ◆
Fd Fd
= .
⇢U 2 L2 model ⇢U 2 L2 actual
We could certainly write this as
Fd, model Fd, actual

2 2 = 2
⇢model Umodel Lmodel ⇢actual Uactual L2actual
2
Fd, model ⇢model Umodel L2model
= 2
Fd, actual ⇢actual Uactual L2actual
2
⇢model Umodel L2model
Fd, model = Fd, actual (7.18)
⇢actual Uactual L2actual
2
which allows us to directly calculate model forces from actual forces using the relative flow conditions.

156
Small Tube Flow
–˙ , depends on the physical variables radius,

For laminar flow through a small tube, the volumetric flow rate, V
R, viscosity, µ and pressure gradient, dP/dx. So we start with the unknown relationship
✓ ◆
dP
–˙ = f
V R, µ, .
dx
We have four variables, so n = 4. We’ll take a look at the primary dimensions,
✓ ◆
dP
–˙ = f
V R, µ, (7.19)
dx
 ✓   ◆
m3 kg kg
= f [m], , (7.20)
s m-s m-s2
All three of them, mass, length and time, are in there. So m = 3. We have four variables, so n = 4. This
leaves us with ⇧ = n m = 1 Pi group.
–˙ as our non-repeated variable, leaving R, µ, and dP/dx as our repeated variables. This means
We’ll use V
✓ ◆c
dP
⇧1 = R a µ b Q
dx
 b  c 
a kg kg m3
= [m]
m-s m2 -s2 s
= [kg]0 [m]0 [s]0 (i.e. dimensionless)
We collect all the unknown constants and set them equal to zero to find that
Length 0=a b 2c + 3
Mass 0=b+c
Time 0= b 2c 1=0
a= 4
b=1
c= 1

157
with makes our first and only Pi group

µ
⇧1 = –˙
V
R4 dP
dx
It may be more useful to look at the Pi group as a ratio,
–˙
V
⇧1 = R4dP
= const.
µ dx
or rearranged a little,
R4 dP
–˙ = (const.)
V .
µ dx
Looking back at our pipe flow work, we can see that the constant is ⇡/8.
Wow, that was a lot of work to find something we already knew. Let’s apply the same principle to
something we don’t already know.
Centrifugal Pump
You are designing a pump system to distribute water throughout a high rise apartment building in Chicago.
–˙ , impeller diameter, D, pump speed,

You know the pump power, Ẇ , is a function of volumetric flow rate, V
⌦, and fluid properties ⇢ and µ.
Total Drag on a Sailboat Moving through the Water
You get a job with an America’s Cup team. You know from fluids class that the drag will be a function of
⇢, µ and U . The boat’s length, L, is obviously a factor, too. But, your observations indicate that the boat
leaves a wake on the surface of the water. These waves are moving up and down under the e↵ect of gravity.
Therefore, you need to include g.
Define the variables:
7.5 Some Classic Pi Groups
In general, we like to compare the inertia of a flow to other stu↵ going on. In fact, that’s what the Navier-
Stokes equations do:
✓ ◆ ✓ ◆
@u @u @u @u @P @2u @2u @2u
⇢ +u +v +w = + ⇢ gx + µ + 2 + 2 (7.21)
@t @x @y @z @x @x2 @y @z

158
We know this is ma = F per unit volume, so let’s take a closer look at the inertial term. The left hand side
is mass times acceleration, or

ML
ma = ,
T2
but it’s convenient for us to work in terms of ⇢, which is in mass/volume, so we’ll replace the M with ⇢–
V
which is ⇢L3 , so
⇢L3 L
ma =
T2
⇢L2 L2
=
T2
⇢V2 L2
=
T2
= ⇢U 2 L2
O.K. So inertial forces, that is something that comes from ma and has units of force, is ⇢U 2 L2 .
What about viscous forces?
du
⌧A = µ A (7.22)
dx
U
= µ L2 (7.23)
L
= µU L. (7.24)
So if we want to compare inertial forces to viscous forces, we get
inertial ⇢U 2 L2
= (7.25)
viscous µU L
⇢U L
= (7.26)
µ
= Re (7.27)
Now is as good a time as any10 to make an observation. The ratio
µ
⇢
10 And none of them are very good.

159
shows up so frequently, that we have a special name for it. That is the kinematic viscosity, ⌫.
µ
⌫= .
⇢
There are two ways to think about viscosity:
• Viscosity is the stickiness of the fluid. How intertwined are the molecules that make up the fluid? It
is a measure of how well it resists shear. This is the first definition we learned. This is called dynamic
or molecular viscosity, µ. It’s units are stress time, Pa-s, or kg/m-s.
• Viscosity is the rate at which information about acceleration propagates through the fluid. If I jerked
one of the two flat plates in my viscosity experiments, how long would it take for the that new influx
of momentum to di↵use to the other plate. It turns out this relies on the molecular stickiness and
density. We call this the kinematic viscosity, µ/⇢. It has units m/s2 .11
The Reynolds number is hence defined in two very common ways,
⇢U L UL
Re = = .
µ ⌫
What about pressure forces? Forces result from a di↵erential in pressure, so
PA = P L2
We often want to compare pressure forces to inertial forces.
pressure P L2
= (7.28)
inertial ⇢U 2 L2
P
= (7.29)
⇢U 2
but since the denominator is so close to dynamic pressure, we throw in the 1/2 to make it
P
Eu = 1 2
, Euler Number
2 ⇢U
Gravitational forces are easy,
mg = ⇢L3 g
11 You will learn next semester that all di↵usion coefficients have units m/s2 .

160
And of course, we will compare gravitational forces to inertial forces,
inertial ⇢U 2 L2
= (7.30)
gravitational ⇢L3 g
U2
= (7.31)
gL
For some reason the father-son tag team of Bill and Bob Froude12 liked the square root of that ratio better,
and came up with

U
Fr = p , Froude Number
gL
7.6 Entrance Length
Figure 7.7: Flow development for laminar flow in the x-direction.
We’ve seen the boundary layer growth for free stream passing over a flat plate. For internal flows, we say
that the flow is fully developed when
@u
=0 For flow in x-direction
@x
The distance it takes for the flow becomes fully developed in a flow channel is called the entrance length,
and is given the variable Le . In Laminar flows this occurs when the upper and lower boundary layers meet
one another. This occurs in the center of the two boundaries and so we have,
99 = D/2.
Using the same equation we used for the Blasius solution we can say,
12 Rhymes with “dude”.

161
99 D 5.0
= =⇣ ⌘1/2 (7.32)
Le 2Le ⇢U1 Le
µ
Let’s now rearrange the expression and solve for the entrance length.
D 5.0
=⇣ ⌘1/2
2Le ⇢U1 Le
µ
D 5.0 1/2
=⇣ ⌘1/2 Le
2Le ⇢U1
µ
✓ ◆1/2
D ⇢U1
L1/2
e =
10 µ
1/2
✓ ◆1/2
D ⇢U1 D
L1/2
e =
10 µ
Finally taking the square of both sides of the expression leaves us with the laminar flow entrance length.
D
Le,laminar = Re (7.33)
100
Equation 7.33 is the entrance length for flow between two plates with a very small aspect ratio (H/W ).
For laminar flow in a pipe we get a similar expression of
D
Le,laminar = Re (7.34)
20
If we have a rectangular channel with an aspect ratio that is not close to zero, we can use hydraulic
diameter Dh of the channel. This is calculated by
4A
Dh = (7.35)
P
where A is the cross sectional area of the channel and P is the perimeter of the channel. For turbulent
flows through a pipe, the development length can be given as
Le,T = DRe1/4 For Re  107 (7.36)
This expression may look slightly di↵erent depending on which textbook you use. Frank White states in

162
his book that he has used a power of one sixth for the last several decades. But recent PIV data suggests
13
the power of one fourth is more appropriate .
7.7 Drag Coefficients
We have spent some time working on non dimensional numbers. Many of the Pi groups we found can be
lumped into two categories:
• Numbers: Someone’s name attached to a Pi group that compares two important flow phenomena, often
forces
• Coefficients: One interesting phenomenon, often a force, that’s non-dimensionalized by some other flow
parameters
One of the Pi groups that keeps coming up is the Reynolds number. With good reason. The balance of
viscous to inertial forces is very important in fluid flow. Another one that comes up a lot is a force coefficient,
or drag coefficient.
FD
CD = .
⇢U 2 L2
Remember, these Pi groups are really functional relationships. They don’t give us any constants that
might be built in. This allows us to add our own constants, if we want. We could rewrite CD as
FD
CD = 1 2
.
2 ⇢U A
where A is the “area” of interest. Area is left in quotes because we don’t know exactly what area we need.
During our discussion of boundary layers, we discovered that the viscous shear on a flat plate, gave us a
13 Even MJ missed a bucket now and then.

163
coefficient of friction, Cf , of
⌧w
Cf = 1 2
2 ⇢U
0.664
=p .
Rex
We then integrated the shear over the entire area of the plate, and it gave us
1.33
CD = p .
ReL
In our discussion of golf balls and turbulent boundary layers, we decided that a turbulent boundary layer
is better at bringing high momentum fluid down close to the plate, and this results in higher shear at the
plate. This results in a higher shear, resulting in a slightly di↵erent drag coefficient,
0.0742
CD = 1/5
.
ReL
What’s the point of all this? There are two points, really
• We have unknowingly come up with several important non-dimensional numbers already.
• The question, “What formula do I use?” has no place in Fluid Dynamics.
O.K. So we have the drag coefficient for a flat plate. We have a few drag coefficients actually. If we plot
them, they look like the curves shown in Figure 7.8
Looking at these drag coefficients, we can see that we need to know the regime of flow (laminar vs
turbulent) before we can figure out the correct drag coefficient.
Further, these drag coefficients are all based on wall shear. As long as wall shear is parallel to the flow,
it contributes to drag. But what if we turn the plate so that it blocks the free stream, such as in Figure 7.9.
Now the shear really doesn’t have anything to do with the flow. The drag is a result of pressure di↵erences

164
Figure 7.8: Variation of drag coefficient with Reynolds number for flow over a flat plate.
Figure 7.9: Flow over and around a flat plate oriented perpendicular to the free stream.
on the upstream and downstream sides of the plate, or
Z
FD = P dA
surface
where P is the pressure and “surface” includes both upstream and downstream sides of the plate.
For now14 , drag force can really be broken down into two sources.
1. Skin friction due to viscous drag
2. Form drag or pressure drag
Unfortunately15 , we aren’t able to develop an analytical expression for the drag coefficient. In fact, there
is a long list of shapes and objects for which we can’t do this.
Years of experiments have led to the table of drag coefficients in Figures16 7.11 and 7.12
14 There is also lift induced drag and something called wave drag
15 Or fortunately, if you’re getting sick of deriving things.
16 Actually a table, but copied from White.

165
Figure 7.10: Sources of drag for common geometries. Taken from Wikipedia.

166
Figure 7.11: Drag coefficients of several 2-dimensional objects. Table taken from Frank White, Fluid Me-
chanics, McGraw-Hill, 2008

167
Figure 7.12: Drag coefficients of several 3-dimensional bodies. Table taken from Frank White, Fluid Me-
chanics, McGraw-Hill, 2008

168
Jumping out of airplanes
What diameter of parachute would you need to fall through the air at only 12 mph.
If we are falling at a constant velocity of 12 mph, there is no acceleration and therefore the net force on
the skydiver must zero. The force of drag must balance the force of gravity.
Fg = FD
1
mg = CD ⇢u2 A
2
The drag coefficient for a parachute can be found in Figure 7.12.
CD = 1.2
The area used in the drag force calculation is the projected area of the parachute, which is simply a
cirlce.
⇡ 2
A = ⇡r2 = D
4
Plugging in these terms we get,
1 ⇡
mg = CD ⇢u2 D2
2 4
8mg
D2 =
CD ⇢u2 ⇡
And therefore the diameter of the chute is
✓ ◆1/2
8mg
D=
CD ⇢u2 ⇡

169
Example: Baseball
In 2010 while playing for the Cincinnati Reds, a young pitcher by the name of Aroldis Chapman threw the
fastest pitch in Major League Baseball history. His pitch left his hand at an insane 105.1 mph (47 m/s).
Calculate the force of the drag on the baseball the moment it leaves Chapman’s hand.
Some background information about a baseball:
D = 74 mm
m = 145 g
The force of drag is calculated by

1
FD = CD ⇢u2 A
2
The drag coefficient for a spherical baseball can again be found in Figure 7.12. The ball is an ellipsoid
with the ratio of length to diameter equal to one.
For this geometry, the table provides a drag coefficient for laminar flow and another for turbulent flow.
This means we will have to calculate the Reynolds number to determine which regime we are in. We can
use the density and viscosity of air at T = 20 [o C].
⇢uD
Re = = 228581
µ
For external flows, the transition from laminar to turbulent happens around 5 ⇥ 105 , which means for
our case the flow is still laminar. Looking back at the drag coefficient table in Figure 7.12 we get that
CD = .47
The area in the drag force equation is the projected area of the shape that is normal to the fluid motion
so again we are just left with the area of a circle.

170
⇡ 2
A = ⇡r2 = D
4
Now we can calculate the drag force.
1
FD = CD ⇢u2 A = 2.7 [N]
2

171
Chapter 8
Pipes & Pumps
Figure 8.1: CFD simulation through a pump - Siemens
8.1 Pipe Flow Losses
Engineers tend to make fun of each other. Things like
• You can’t spell “geek” without a E.E.
• Industrial Engineering (IE) is “imaginary engineering”
• Civil engineers are “dirt engineers,” and other sewage related jokes

172
• Mechanical Engineering (ME) is “miscellaneous engineering”1
On top of that, aerospace engineers often criticize undergraduate fluid mechanics classes in mechanical
engineering as “3 credits of pushing oil through a pipe.” Well, from an infant’s heart and circulatory system,
to the plumbing and HVAC systems in a building, to the Alaskan pipeline. Pushing stu↵ through a pipe is
pretty important. Fortunately, it isn’t too hard. We’ll spend a couple days on it. Prepare yourself for your
friends making fun of you.
8.1.1 Thermodynamics Review
Let’s take a moment and review the energy equation as we learned it in thermodynamics. Of course, energy
is conserved, so for pipe flow,
✓ ◆ ✓ ◆
1 1
ṁin h + U 2 + gz = ṁout h + U 2 + gz , (8.1)
2 in 2 out
where h in this case is specific enthalpy of the fluid. Enthalpy, as we learned, is an algebraic convenience
combining internal energy and work. We can expand the enthalpy and mass flow in Equation 8.1 a little bit
as
✓ ◆ ✓ ◆
˙ P 1 2 ˙ P 1 2
⇢V
– in u + + U + gz = ⇢V
– out u + + U + gz , (8.2)
⇢ 2 in ⇢ 2 out
For an incompressible fluid, the volumetric flow rate and density are constant on both ends of the pipe. We
can then divide through by g and get
✓ ◆ ✓ ◆
u P U2 u P U2
+ + +z = + + +z . (8.3)
g ⇢g 2g in g ⇢g 2g out
Equation 8.3 looks a little familiar. It used to be an energy balance, but like an aging Hollywood star,
it’s had one too many facelifts and is no longer recognizable. Hopefully, you trust the algebra to get from
Equation 8.1 to 8.3.
What you should investigate a little bit is the dimensional homogeneity of the equation. What are the
units of each of these terms?
Each term in parentheses in Equation 8.1 has units energy per mass, which is a really nice unit if you’re
into thermodynamics. Civil engineers don’t usually take thermo, so they like to measure the “energy” of
their fluids by how high they are. Water behind a dam, pressure measured by water columns, etc. For some
1 To be fair, they’re correct about the miscellaneous engineering thing.

173
reason, the Civil’s won the war on energy equations, so we leave the fluid mechanics energy balance in meters
instead of Joules per kilogram.
Now, we need to do a little comparison between thermodynamics2 and fluid dynamics. In thermo, we
paid close attention to what all the devices do, but largely neglected what happens between the devices. In
fluids, thus far, we have completely neglected internal energy, or much of anything to do with temperature,
for that matter.
Let’s rearrange Equation 8.3 just a little bit, collecting the internal energy terms,
✓ ◆ ✓ ◆ ✓✓ ◆ ✓ ◆ ◆
P U2 P U2 u u
+ +z = + +z + (8.4)
⇢g 2g in ⇢g 2g out g out g in
| {z } | {z } | {z }
Pressure, kinetic and Pressure, kinetic and Internal energy
gravitational potential gravitational potential lost along
energy (inbound) energy (outbound) the way.
So what exactly happened here? Viscous shear, that’s what happened. As the fluid traveled along
the pipe, it was constantly being sheared. Shear, multiplied by some area, is a force. Force, over some
distance, is energy. Shearing this viscous fluid along the length of the pipe required some energy. The
viscous dissipation shows up as a veryslight increase in temperature, and drop in pressure.3 But, because we
choose to use meters, e↵ectively gravitational potential energy, to quantify this energy loss. The term used
by civil engineers, and adopted by mechanical engineers, is head, h.4 The traditional letter used to represent
head is h. Consequently, we won’t consider enthalpy for the remainder of the course.
8.1.2 Pipe Losses
We will now rewrite Equation 8.3 in terms of the lost head due to friction,
✓ ◆ ✓ ◆
P U2 P U2
+ +z = + +z + hf . (8.5)
⇢g 2g in ⇢g 2g out |{z}
| {z } | {z } friction head
Pressure, kinetic and Pressure, kinetic and
gravitational potential gravitational potential
energy (inbound) energy (outbound)
The term P/⇢g is called pressure head, the term U 2 /2g is called velocity head, and the z term is called,
as it should be, head.
Applying Equation 8.5 is a lot like applying Bernoulli’s equation, except we need to figure out what the
frictional losses are. The theory is steeped in turbulent boundary layer theory, but is fortunately, easily
summarized by yet another non-dimensional group, the Darcy friction factor, f .5 The friction factor is
2 Which the author still contends should be called “Comparative Thermostatics.”
3 But of course, thanks to the dirt engineers, we measure this energy loss in meters.
4 Don’t confuse the enthalpy h with the head h.
5 Named after French engineer Henry Philibert Gaspard Darcy. It is sometimes called the Darcy-Weibach friction factor,

174
defined as
hf
f= L U2
(8.6)
d 2g
⇣ ✏⌘
= f Re, (8.7)
d
where
✏ = roughness (8.8)
and d is the diameter of the pipe. This makes the relative roughness, ✏/d, a relatively easy dimensionless
group to understand. Relative roughness ✏ can be thought of as the average size of the protuberances from
the material surface, illustrated in 8.2
For laminar flow, the friction factor f is

64
f= (8.9)
Re
For turbulent flows we have two options:
1. Use the Moody chart in Figure (also extremely likely to be inside the back cover of your book)
2. Use a correlation
The correlation that describes the curves on the Moody chart for turbulent flows is called the Colebrook
equation.
✓ ◆
1 ✏/d 2.51
p = 2.0 log10 + p (8.10)
f 3.7 Re f
Notice this function is not explicit in terms of the friction factor. You can guess and check with the
Reynolds number in an iterative manner, or you can EESily use an equation solver6 .
Now, with this handful of dimensionless groups, it should be easy to figure out the frictional losses in
pipe flow. If only we knew something about the roughness of pipes. Fortunately, we have the table in Figure
8.3 to guide us.
With the table in Figure 8.3 in hand, we determine the absolute roughness. From there, we divide by the pipe
diameter, d, to get a relative roughness. Now that you have two dimensionless groups, Reynolds number and
relative roughness, you can find use the Moody Chart in Figure 8.4 to find the appropriate friction factor.
tacking on some credit to German mathematician and engineer, Julius Ludwig Weisbach. Much like the Newton-Leibniz debacle,
the German guy gets the short end. Funny how history works.
6 See what we did there?

175
Figure 8.2: Roughness shown for laminar, transitioning and turbulent flows
Based on the definition of friction factor, you could solve to get
L U2
hf = f (8.11)
d 2g
which works out to be in meters.7
Figure 8.3: Roughness values for common pipe and duct materials
7 Because f and L/d are dimensionless and U 2 /2g, the velocity head, is in meters.

176
Figure 8.4: Moody Chart
8.2 Head Loss
Let’s talk about what the term“head loss” means.
• Some textbook specify hL as the pressure lost due to viscous e↵ects.
• Some books di↵erentiate between viscous or friction losses, hv or hf , and the other losses, such as
minor losses (pipe fittings) and elevation changes.
• The authors’ experience in the HVAC and plumbing worlds (big buildings) is slightly di↵erent than
the textbooks, and based on experience we tend to believe that “head” is simply a proxy for pressure.
I have several calls out to my friends who are engineers in the building and pumping worlds. No one has
called back yet. I suspect they view it like this:
✓ ◆ ✓ ◆
P U2 P U2
+ +z = + +z + hf (8.12)
⇢g 2g in ⇢g 2g out |{z}
| {z } | {z } head loss
Total head, in Total head, out
✓ ◆ ✓ ◆
Pin Pout U2 U2
= +z +z + hf (8.13)
⇢g ⇢g 2g out 2g in |{z}
| {z } friction loss
Total head loss

177
In other words, for horizontal pipes of constant diameter, “head loss” is simply viscous losses.
But... Once you start incorporating elevation changes (a big deal in big buildings) the “head loss” then
includes the hydrostatic e↵ect.

178
Head Loss in Water Pipe
Water at 17 C is pumped 75 m through a 6 cm stainless steel pipe. There is no elevation gain along the
path of the pipe. The flow rate is 5.5 L/s. What is the head loss along the pipe? (Hint: ⇢=998 kg/m3 , µ=
1.08 mPa-s)
Solution:
U = 1.945m/s, Re = 108000, ✏/d = 0.00003, f = .017, hf = 4.1

179
Flow Rate
Crude oil flows 100 m through a 150 mm pipe with a roughness of 0.06 mm. The pump can only generate a
lift 8 of 7.5 m. What is the average velocity in the pipe? You may assume that the kinematic viscosity, ⌫, of
6
the crude oil is 40(10 m2 /s.
We don’t have a velocity, so we can’t get a Reynolds number. What do we know?
We know neither U nor f .
s r
L U2 2hf dg 0.220725
hf = f ! U= =
d 2g fL f
We need a Reynolds number, but it requires a velocity,
uD
Re = = 3750u
⌫
I have no idea what the actual Re is, so I’ll guess 30,000 (nice and turbulent, but not too turbulent)
Grinding through the process now, I get f = 0.0245.
With this, I can go back and get velocity. U = 3.00 m/s
With this, I can check my Reynolds number. I get Re = 11,300.
With the new Reynolds number, I get a new f of 0.031
With the new f, I get U = 2.66 m/s, and Re of 10,000.
With Re of 10,000, I get f=0.0313 (pretty close to the last one). Good enough.
This iterative process is a bit of a hassle. When you have an equation solver available (which, you do)
you can use an implicit relationship between roughness, Reynolds number and friction factor developed by
Colebrook,
✓ ◆
1 ✏/d 2.51
p = 2.0 log10 + p (8.14)
f 3.7 Re f
8 Civil engineers strike again.

180
Minimum Duct Diameter
Heater air at 1 atm and 35 C needs to be moved 150 m in a long circular duct. The volumetric flow rate,
–˙ , is 0.35 m3 /s. The head loss must be no greater than 20 m.

V
What is the minimum diameter of the duct?
Solution:
5
⇢ = 1.145 µ = 1.895(10 ).
First, we need the Reynolds number...
Well, that’s not going to work.
Final answer: dmin = 0.271

181
Chapter 9
Streamlines & Potential Flow
Figure 9.1: Pressure field and streamlines of external flow over air foil
At this point, we have covered internal flows extensively and looked briefly at external flows. Regarding
external flows, we were able to investigate the viscous shear on a flat plate, and from that, develop a coefficient
of drag. We then took it on faith that foils and plates can generate lift at some angle of attack. We will look
briefly at how fluids flow over and around immersed bodies.
We also know that there is a thin1 boundary layer between the free stream, u = U1 , and the surface,
u = 0.
For many external flows over a body, we can approximate the flow as being inviscid (no viscous e↵ects)
and irrotational (no vorticity, particles are not rotating) for all of the flow field but a small layer around the
1 By thin, a boundary layer of millimeters on a body measured in meters.

182
body.
We have discussed viscosity and viscous e↵ects at some length. Vorticity will take a little more explana-
tion. Imagine looking top down on a laminar flow through a 2D channel. Now imagine two popsicle sticks
glued into an x-shape that is moving with the flow, somewhere o↵ the center line, as shown in Figure 9.2.
Figure 9.2: Two-dimensional channel flow with a floating vorticity gauge made of popsicle sticks. The
vorticity gauge is translated by the flow and rotated by the velocity gradient.
The end of the popsicle stick closer to the centerline will always move faster than the end closer to
the wall. This di↵erence in velocity results in a rotation of the vorticity gauge. Even though the flow is
laminar, and all the streamlines are moving straight left-to-right, the flow still has some rotational quality
to it. This rotational quality is called vorticity and is usually represented by the Greek letter omega, !.2
Mathematically, vorticity is given by the curl of the velocity field, as shown by Equation 9.1,
! =r⇥u (9.1)
Calculating Vorticity
With your neighbor, or working solo if that’s your thing, find the vorticity for a 2-D flow, [u, v].
2 Omega is chosen because of its relationship to angular velocity, also represented by !, but not to be confused with velocity
in the z-direction, w.

183
We know from our study of the momentum equation that the velocity gradient is a result of viscosity
and the no-slip condition at the wall. In the absence of viscosity, there can be no vorticity. Hence, the flow
field is irrotational, then the vorticity of the entire flow field is zero,
! = r ⇥ u = 0.
We can do this because the viscous e↵ects are really limited to small layer around the body called the
viscous boundary layer. In more advanced fluids classes, you can take the information learned in this chapter
and apply more thorough analysis to include e↵ects of the viscous boundary layer.
9.1 Stream Function
A streamline is a line that is tangent to the velocity field at every point along the line. Consider a 2-D
flow field where the flow of particles is steady and all particles follow streamlines. As one particle is leaving
its location, another particle is simultaneously moving in to fill the space. This leads us back to our familiar
definition of velocity,
dx dy
=u , = v. (9.2)
dt dt
Isolate dt for both of these expressions.

dx dy
dt = = (9.3)
u v
If the stream lines move generally in the x-direction but have some variance in their y position along the
path, it is convenient to solve for an expression that represents this.
dy v
= (9.4)
dx u
Equation 9.4 represents the shape of a streamline as you move in the x-direction. To solve for the exact
y-position for any x-position along the streamline, we can integrate this expression to obtain some y = f (x)
for an individual streamline.
Consider a particle on a streamline at some time t0 , labeled as the red dot in Figure 9.3. Let S be the
streamline and let ds be some infinitesimal change along the direction of the streamline.
ds = dx i + dy j (9.5)

184
Figure 9.3: Infinitesimal change in the direction of S, is denoted as the vector ds.
ˆ that is normal to the vector ds. This means that

At each point along S there exists a vector dn
ˆ · ds = 0
dn
To characterize each streamline, we will define a function for such that r is in the same direction as the
ˆ The dot product between r
normal dn. and s will also be zero.
0 = r · ds
✓ ◆
@ @
0= i+ j · (dx i + dy j)
@x @y
@ @
0= dx + dy
@x @y
This expression means that the gradient of the stream function in the streamline direction is exactly zero.
Rearranging again gets us,

dy @ @
= (9.6)
dx @x @y
Looking back on equation 9.4 we can now write that
dy @ @ v
= = (9.7)
dx @x @y u
dy
where dx indicates the derivative is evaluated along a single streamline. The stream function (x, y)
is constant along any one streamline. Finally we can end up with the following statements,
@ @
u= , v= , (9.8)
@y @x

185
where the velocity is given from the velocity vector u = [u, v]. Equation 9.8 provides a definition for the
stream function which can be used to find individual stream lines throughout the flow field.
To picture this, consider the flow in the same channel as Figure 9.2, but this time inviscid as shown
in Figure 9.4. Since the flow is inviscid, the velocity profile is uniform. All of the velocity is left to right.
Streamlines are evenly spaced and arbitrarily given values of -2 through 6.
Figure 9.4: Two-dimensional inviscid, irrotational channel flow with arbitrary values of stream functions
assigned to streamlines.
The values of stream function increase in the y-direction. Hence,
@
= u > 0.
@y
That is, the flow is moving left to right, in the positive x-direction. There is no gradient in stream
function in the x-direction, hence
@
= v = 0.
@x
meaning that there is no velocity in the y-direction.

186
9.2 Potential Function
If the flow field is irrotational, the vorticity of the velocity vector must be zero. This means that we can
define the velocity vector as a scalar gradient function.
u=r (9.9)
We call the scalar velocity potential and it is defined by,
@ @
u= , v= (9.10)
@x @y
where again the velocity is given from the velocity vector u = [u, v].
There is some analogy between pressure and velocity. In general (but certainly not always) fluid flows
from areas of high pressure to low pressure.3 Flow potential, , is the opposite. Fluid flows from areas of
low potential to areas of high potential. Best of all, there is no exception.
Figure 9.5 shows some flow characteristics to help understand the relationship between and in a
slightly more physically intuitive manner.
Figure 9.5: Two-dimensional inviscid, irrotational converging-diverging elbow. Vertical velocity is low at the
inlet, where the streamlines are far apart. As streamlines converge, velocity increases. Values of and
are arbitrary, but the relationship between them is correct.
3 The adverse pressure gradient on the back side of a golf ball is one important exception.

187
9.3 Properties of and
We can replace velocity components u, v with the definitions provided by the stream function and the
velocity potential function . Let’s have a look at what continuity looks like when we replace u, v with
, . Remember we in this analysis we are assuming that constant density so continuity simplifies to zero
divergence of the velocity field. Let’s look at the stream function first.
@u @v
0=r·u= +
@x @y
✓ ◆ ✓ ◆
@ @ @ @
= +
@x @y @y @x
0=0
And for the potential function.
@u @v
0=r·u= +
@x @y
✓ ◆ ✓ ◆
@ @ @ @
= +
@x @x @y @y
2 2
@ @
= + 2
@x2 @y
This is a significant result. That means that the continuity equation in term of the potential function
becomes,
0 = r · r = r2 (9.11)
4
Equation 9.11 says that the velocity potential function obeys Laplace’s equation. This relationship is
often helpful in solving for lines of constant potential.

4 Equations that obey Laplace’s equation are called harmonic equations.

188
9.3.1 Uniform Flow
For a uniform flow in the x direction, we know
@ @
=U and =0
@x @y
But we want the potential function, , not its derivative. So we integrate the two to get
= U x + C,
but since the actual values of the potential are arbitrary5 , we can just ditch the C.
= U x. (9.12)
Similarly, we can get the stream function,
@ @
=U and =0
@y @x
leading us to
= U y. (9.13)
9.3.2 Sources & Sinks
For a 2-D flow field, a source is a point in the field that expels fluid in all directions. Jets in a hot tub and
fuel injection are two physical examples of sources. A sink is a point in the flow field in which the fluid is
flowing into from all directions and is aptly named. These flows are axisymmetric and therefore have no
dependence on ✓. Starting from Laplace’s equation of the potential function in polar coordinates,
0
1 @ h @ i >
⇢
1 @ 2⇢
r2 =0= r + 2⇢ 2
r @r @r ⇢
r @✓
This leaves us with
@ h @ i
0= r
@r @r
5 For instance, the score of the most recent jai alai match you saw.

189
Applying the antiderivative6 on both sides of the expression get
m @
=
r @r
where m is a constant. Applying another antiderivative on both sides yields the stream function,
(✓, r) = m ln(r) + C (9.14)
where again C is another constant of integration. If m is positive then the flow is a source with fluid
motion outward from the origin. If this coefficient is negative, then the flow is a sink in which al flow points
in to the origin.
Figure 9.6: Streamlines for source and sink flow fields.
Notice that C does not have an e↵ect on the velocity profile.
m
u=r = r̂
r
The stream function and potential function for source or sink is:
= m✓ = m ln(r) (9.15)
6 Subtle di↵erence from saying integrate here.

190
9.3.3 Vortex flow
We’ve already seen an example of the vortex flow field in a homework assignment.
k
ur = 0 , u✓ =
r
Figure 9.7: Pure vortex flow in the clockwise direction with streamlines drawn in blue
By the definition of the potential function we can use the velocity profile to get the potential function.
k ˆ @ 1@ ˆ
u= ✓=r = r̂ + ✓
r @r r @✓
Here k is a constant that determines the magnitude of the flow velocity. For tropical storms, hurricanes
and other cyclones this value is on the order of hundreds of thousands to even millions.
k 1@
=
r r @✓
@
k=
@✓
Applying the antiderivative on the either side of the expression with respect to theta.

191
9.4 Super position of potential functions
Perhaps one of the most convenient attributes of the stream and potential functions is the characteristic of
super position.
X
total = i = 1 + 2 + 3 + ..... (9.16)
9.4.1 Dipole & Doublets
The easiest example of this is the combination of source and sink. Imagine we have a source next to a sink
with the distance of 2 from one another. The magnitude m is equal for both the source and the sink.
This flow field is called a dipole and is shown in Figure 9.8. Magnetic field around a magnetic dipole has
magnetic flux lines that mimic those of our dipole streamlines.
1 y 1 y
= source + sink = m tan ( ) m tan ( ) (9.17)
x+ x
1 1
= source + sink = m ln (x + )2 + y 2 m ln (x )2 + y 2 (9.18)
2 2
Figure 9.8: Dipole flow field due to source at ( , 0) and a sink at ( , 0)
Removing the labels and mirroring the flow about the x-axis we get the full 2D flow field of the dipole,
seen in Figure 9.9.

192
Figure 9.9: Full 2D dipole flow field

193
9.4.2 Doublet
Now we can make the following adjustments to the flow field:
• !0
• m!1
• 2m ! K = strength of doublet
K sin(✓)
= (9.19)
r
Applying the same thing to the potential function we get,
K cos(✓)
= (9.20)
r
The streamlines for a doublet are shown in Figure 9.10.
Figure 9.10: Streamlines of a doublet
9.4.3 Half-body
We can combine two simple flows, such as a source and uniform flow, such as Figure 9.12.
In this case, the total stream function becomes the sum of the source and the uniform flow.
= uniform flow + source (9.21)

m
= U r sin ✓ + ✓ (9.22)
2⇡

194
Figure 9.11: The flow around a half-body: (a) superposition of a source and a uniform flow; (b) replacement
of streamline = ⇡bU with solid boundary to form half-body. Figure from Munson, et. al. Printed under
fair use.
9.4.4 Rankine Oval
We could also combine a uniform flow with a source and a sink.
Figure 9.12: The flow around a Rankine Oval: (a) superposition of source - sink pair and a uniform flow;
(b) replacement of streamline = 0 with solid boundary to form a Rankine oval. Figure from Munson, et.
al. Printed under fair use.
By tweaking the distance between the source and sink, the strength of the source and sink, and the
uniform velocity, we can adjust the shape of the oval.
We can add a vortex, essentially rotation, to the flow over a Rankine circle, and simulate the e↵ect of a
spinning ball.

195
Figure 9.13: E↵ect of oval parameters on shape. Figure from Munson Printed under fair use.
Figure 9.14: E↵ect of rotation on streamlines around a cylinder. Figure from White. Printed under fair use.

MorseDressler FluidMechanics v1 Comp

Uploaded by

Copyright:

Available Formats

MorseDressler FluidMechanics v1 Comp

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MorseDressler FluidMechanics v1 Comp

Uploaded by

Copyright:

Available Formats

1

December 17, 2019

c 2019 - Morse and Dressler - All Rights Reserved

c 2019 - Morse and Dressler - All Rights Reserved

At some point we’ll write something really nice about

• Prof. Arganthael Berson

c 2019 - Morse and Dressler - All Rights Reserved

0.1 Some mechanics of this class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

c 2019 - Morse and Dressler - All Rights Reserved

5 Di↵erential Fluid Flow 90

6 External flows 121

c 2019 - Morse and Dressler - All Rights Reserved

8 Pipes & Pumps 171

9 Streamlines & Potential Flow 181

c 2019 - Morse and Dressler - All Rights Reserved

1 So.... Fluid mechanics is uninteresting but really hard? . . . . . . . . . . . . . . . . . . . . . .

1.1 Flow visualization of velocity field for a turbulent jet . . . . . . . . . . . . . . . . . . . . . . . 1

2.1 Shipwreck in Lake Michigan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.1 Ocean waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

c 2019 - Morse and Dressler - All Rights Reserved

4.1 SpaceX launches the Falcon Heavy rocket in April 2019 . . . . . . . . . . . . . . . . . . . . . 66

c 2019 - Morse and Dressler - All Rights Reserved

8.1 CFD simulation through a pump - Siemens . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

c 2019 - Morse and Dressler - All Rights Reserved

c 2019 - Morse and Dressler - All Rights Reserved

inspiring, challenging and rewarding.

class, so it really shouldn’t be easy. But there may be some misconceptions...

Figure 1: So.... Fluid mechanics is uninteresting but really hard?

or professional pursuit. For instance,

c 2019 - Morse and Dressler - All Rights Reserved

spin the generator. Google, Gates, etc.

• Just about everything Elon Musk does

0.1 Some mechanics of this class

0.1.1 Grade calculation

Your grade in this class will be based on:

• 40% Friday Quizzes (every other Friday)

• 35% Homework - due on Wednesdays and Fridays

• 15% Final exam

• 5% Lecture Preview Quizzes

There are four ways earn your full credit in participation:

c 2019 - Morse and Dressler - All Rights Reserved

something in class, ask.

• Answer questions we ask in class.

• Correct mistakes we make on the board. They happen, help us out.

0.1.3 Homework Grading

however. In general, your homework will be graded on:

0.1.4 Goals for this class

We hope that by the end of the semester, you have

– pumps and losses

• And a basic understanding of the equations that govern

• And an elementary understanding of

c 2019 - Morse and Dressler - All Rights Reserved

A fluid is a substance that continuously deforms under a shear force.

(always? Under gravity?)

What are some examples of fluids?

• Most of your body

(Newtonian and non-newtonian will come later)

0.3 Systems and Control Volumes

• Fixed chunk of mass - maybe expands, maybe compresses“Piston cylinder device?

Control volume: Arbitrary volume in space, bounded by a control surface