You are on page 1of 122

LABORATORY MODELLING OF OPEN

CHANNEL FLOW PAST EMERGENT


VEGETATION

A dissertation submitted to the University of Manchester


for the degree of Master of Science
in the Faculty of Engineering and Physical Sciences

2009

James McNaughton
School of Mathematics
Contents

Abstract 10

Declaration 11

Copyright Statement 12

Acknowledgements 13

1 Introduction 14
1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.2 Theory of Open Channel Flow . . . . . . . . . . . . . . . . . . . . . . 15
1.2.1 Basic Equations . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.2.2 Non-Dimensional Analysis . . . . . . . . . . . . . . . . . . . . 18
1.2.3 Resistance to Flow . . . . . . . . . . . . . . . . . . . . . . . . 19
1.3 Normal Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2 Literature Review 23
2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.1 Modelling Flow Past Vegetation . . . . . . . . . . . . . . . . . 24
2.2.2 Laboratory Methods . . . . . . . . . . . . . . . . . . . . . . . 25
2.2.3 Field studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2.4 Numerical Methods . . . . . . . . . . . . . . . . . . . . . . . . 28
2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2
2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

3 Experimental Procedure 36
3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Apparatus and Methods . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.1 The Flume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.2 Test Specimen . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2.3 Taking Measurements . . . . . . . . . . . . . . . . . . . . . . 38
3.2.4 Calculating the Flow . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Description of Experiments . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3.1 Series A - Flow past different arrangements of equal φ . . . . . 41
3.3.2 Series B - Controlled depth changes past arrays of varying φ . 42
3.3.3 Series C - Normal flow past arrays of varying φ . . . . . . . . 42
3.3.4 Series D - Ordered arrangements for equal φ . . . . . . . . . . 43
3.3.5 Series E - Flow past flexible rods . . . . . . . . . . . . . . . . 44
3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4 Analysing Results and Interpreting the Data 46


4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2 The Results of Series A . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3 Calculating resistance to the flow – Test B3 . . . . . . . . . . . . . . 49
4.4 Normal Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

5 Results 64
5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2 Solid Volume Fraction and Flow Rate . . . . . . . . . . . . . . . . . . 65
5.3 Resistance by Solid Volume Fraction . . . . . . . . . . . . . . . . . . 68
5.4 Normal Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.5 Flexible Rods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.6 Comparison of results with other authors . . . . . . . . . . . . . . . . 85

3
5.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

6 Conclusions 88

A Nomenclature 95

B Fortran 90 Program 97
B.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
B.2 Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

C Results 103

Word count 16823

4
List of Tables

1.1 Typical roughness heights (Chanson, 1999) . . . . . . . . . . . . . . . 20

3.1 Investigated Solid Volume Fractions . . . . . . . . . . . . . . . . . . . 40


3.2 Test parameters for Series A . . . . . . . . . . . . . . . . . . . . . . . 41
3.3 Test parameters for Series B . . . . . . . . . . . . . . . . . . . . . . . 42
3.4 Test parameters for Series C . . . . . . . . . . . . . . . . . . . . . . . 43
3.5 Test parameters for Series D . . . . . . . . . . . . . . . . . . . . . . . 44
3.6 Test parameters for Series B . . . . . . . . . . . . . . . . . . . . . . . 45

4.1 Functions fitted to laboratory data and their errors for Test B3 (hup ≈
8 cm) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

5.1 Abbreviations of controlled depths used in results . . . . . . . . . . . 65


5.2 Actual upstream flow depths for tests in horizontal channel . . . . . . 66
5.3 Error in volume for different rods per row (m) . . . . . . . . . . . . . 75
5.4 Strouhal number and predicted frequency of vibrations for Test E2 (hup4 ) 85

C.1 Test B1 hup1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103


C.2 Test B1 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
C.3 Test B1 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
C.4 Test B2 hup1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
C.5 Test B2 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
C.6 Test B2 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
C.7 Test B3 hup1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
C.8 Test B3 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

5
C.9 Test B3 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
C.10 Test B4 hup1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
C.11 Test B4 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
C.12 Test B4 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
C.13 Test D1 hup1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
C.14 Test D1 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
C.15 Test D1 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
C.16 Test D2 hup1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
C.17 Test D2 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
C.18 Test D2 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
C.19 Test D3 hup1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
C.20 Test D3 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
C.21 Test D3 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
C.22 Series C Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
C.23 Series D Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
C.24 Test E1 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
C.25 Test E1 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
C.26 Test E1 hup4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
C.27 Test E1 hup5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
C.28 Test E2 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
C.29 Test E2 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
C.30 Test E2 hup4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
C.31 Test E2 hup5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
C.32 Test E3 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
C.33 Test E3 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
C.34 Test E3 hup4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
C.35 Test E3 hup5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
C.36 Test E4 hup2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
C.37 Test E4 hup3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
C.38 Test E4 hup4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

6
C.39 Test E4 hup5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

7
List of Figures

1.1 Cross-section of a rectangular prismatic channel . . . . . . . . . . . . 15


1.2 Horizontal cross-section of open channel flow with slowly varying depth 16
1.3 Cross-section of normal flow . . . . . . . . . . . . . . . . . . . . . . . 21

2.1 Flow Depth along channel (Bixio, 2007) . . . . . . . . . . . . . . . . . 30


2.2 Drag Coefficient vs Velocity ( m s−1 ) (Fathi-Maghadam and Kouwen,
1997) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.3 Cd vs Re (= uh/ν) (Järvelä, 2002) . . . . . . . . . . . . . . . . . . . 32
2.4 Drag Coefficient vs Rep (Tanino and Nepf, 2008) . . . . . . . . . . . 33
2.5 Vegetal Drag Coefficient vs Rec (Wu et al., 1999) . . . . . . . . . . . 33

3.1 Diagram of the laboratory flume used for all experiments . . . . . . . 37


3.2 Plan view of plastic board (depth 11 mm) . . . . . . . . . . . . . . . 38
3.3 Repeated patterns for Series D . . . . . . . . . . . . . . . . . . . . . . 44

4.1 Flow past an array with φ = 0.027 . . . . . . . . . . . . . . . . . . . 47


4.2 Surface depth for Tests A5 and A6 (φ = 0.041) . . . . . . . . . . . . . 48
4.3 Laboratory data (points) and the linear functions (lines) fitted to cen-
tral portions for Test A3 . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.4 Absolute depth gradient against flow rate for varying φ. . . . . . . . . 50
4.5 Control Volume of first five rows of rods of Test B3 . . . . . . . . . . 51
4.6 Cd vs Rep for a single cylinder and the region behind it. (Taken from
Acheson, 1990, p. 150) . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.7 Forces acting in the channel for Test B3 (hup ≈ 8 cm) . . . . . . . . . 56

8
4.8 Surface profiles of Test B3 (hup ≈ 8 cm) . . . . . . . . . . . . . . . . 57
4.9 Cd vs x (m) - Calculated from lab data and fitted functions of Test B3
(hup ≈ 8 cm) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.10 Cd vs Rep - Calculated from lab data and fitted functions of Test B3
(hup ≈ 8 cm) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.11 u vs x (m) - Calculated from lab data and fitted functions of Test B3
(hup ≈ 8 cm) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.1 Q vs φ for various hup in Series B . . . . . . . . . . . . . . . . . . . 65


5.2 Comparison of similar φ’s for patterned and un-patterned arrangements 67
5.3 Cd vs Rec for Test B2 with different upstream depths . . . . . . . . . 68
5.4 Cd vs x for Series B tests and upstream depth hup3 . . . . . . . . . . 69
5.5 Cd vs Rec for Series B tests and upstream depth hup3 . . . . . . . . . 70
5.6 Comparison of Cd vs Rec for equal φ but repeated rows. . . . . . . . 71
5.7 Cd vs Rec for Tests B2, B3 and D3 . . . . . . . . . . . . . . . . . . . 72
5.8 Drag coefficient for tests of Series C . . . . . . . . . . . . . . . . . . . 74
5.9 Cd vs Rec for α = 1 ◦ and hN1 . . . . . . . . . . . . . . . . . . . . . . 76
5.10 Cd vs Rec for normal flow (φ = 0.027) . . . . . . . . . . . . . . . . . . 77
5.11 Cd vs Rec for normal flow (φ = 0.041) . . . . . . . . . . . . . . . . . . 78
5.12 Laboratory observations at the front of the array of Test E1 with dif-
ferent upstream depths . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.13 Cd vs x (m) for Test E1 . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.14 Cd vs Rec for Test E1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.15 Cd vs Rec for Tests E1, E2 and B1 with upstream depths hup2 and hup3 81
5.16 Cd vs Rec for Tests E3 and D1 (pattern P2) . . . . . . . . . . . . . . 82
5.17 Cd vs Rec for Test E4 (φ = 0.024) . . . . . . . . . . . . . . . . . . . . 83
5.18 Cd vs Rec for Tests E4 and D3 . . . . . . . . . . . . . . . . . . . . . 83
5.19 Comparison of Cd vs Rep for Series B (h1 ) and results from Tanino
and Nepf (2008) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

9
The University of Manchester
James McNaughton
Master of Science
Laboratory Modelling of Open Channel Flow Past Emergent Vegetation
September 7, 2009

The study of flow past emergent vegetation is of great importance in studying


river dynamics and the drainage of flood plains. Rigid and flexible rods were studied
in a laboratory flume to investigate their resistance and how it is affected by the
solid volume fraction of the array, array arrangement, flow rate, channel slope and
rod flexibility. The results show correlations between these parameters and the drag
coefficient, Cd . An increase in solid volume fraction, flow rate and slope all lead to an
increase in Cd . Repeated rows of the same pattern of rods reduces Cd dramatically
as does the introduction of flexible rods. Specific patterns of rods were found to
give more drag than denser arrays of non-specific design whilst arrays of equal solid
volume fraction could be found to lower the drag by using specific patterns of the
same repeated row.

10
Declaration
No portion of the work referred to in this dissertation has
been submitted in support of an application for another
degree or qualification of this or any other university or
other institute of learning.

11
Copyright Statement

i. Copyright in text of this dissertation rests with the author. Copies (by any
process) either in full, or of extracts, may be made only in accordance with
instructions given by the author. Details may be obtained from the appropriate
Graduate Office. This page must form part of any such copies made. Further
copies (by any process) of copies made in accordance with such instructions
may not be made without the permission (in writing) of the author.

ii. The ownership of any intellectual property rights which may be described in
this dissertation is vested in the University of Manchester, subject to any prior
agreement to the contrary, and may not be made available for use by third
parties without the written permission of the University, which will prescribe
the terms and conditions of any such agreement.

iii. Further information on the conditions under which disclosures and exploitation
may take place is available from the Head of the School of Mathematics.

12
Acknowledgements

I would like to thank my supervisor Dr Gregory Lane-Serff, without whom my project


would surely have remained as thousands of numbers and scribbled notes in a lab
book. I would also like to thank Phil Oakes for his continued help with the laboratory
equipment that was the backbone of my project, his willingness to help at a moments
notice was a true asset to my work.
As well as the above I would also like to thank Dr Paul Johnson for his expert
and friendly teaching of valuable skills that have given me the opportunity to analyse
and present my research in such a professional manner.
Finally, I am eternally grateful to my parents for their continued love and support
throughout my academic studies and helping me to follow my passion.

13
Chapter 1

Introduction

1.1 Overview

Vegetation causes significant resistance to flow and as a result its study is vital for
understanding and managing rivers, flood plains, wetlands and any similar aquatic
environments. A common practise in modelling these environments is with open
channel flow. In this project open channel theory and laboratory experiments have
been used to investigate approaches to modelling a range of these environments. The
project aims to find the key parameters that effect the resistance of the vegetation,
and, where possible, investigate the manner in which this can be minimised. The
next chapter discusses approaches from other authors practising similar work as in
this project and the results that they have found. Chapter 3 then introduces the
laboratory equipment that was used and the methods that would be used in obtaining
results, an outline of the experiments to be completed is also included here. Chapter
4 presents selected results from the experiments and demonstrates the steps required
in preparing them for analysis. The results for all of the tests are then compared in
Chapter 5. This first chapter introduces the theory of open channel flow which will
be relevant to both the modelling and analysis of the chapters described above.

14
CHAPTER 1. INTRODUCTION 15

1.2 Theory of Open Channel Flow

To discuss the theory of open channel flow it is first essential to define the channel.
The shape of a channel is its geometric shape through a cross-sectional view, for
simplicity the most commonly studied channel is one where this cross-section does
not change shape or alter dimensions; such a channel is said to be prismatic. The
laboratory equipment that was available was a rectangular prismatic channel and so
the theory introduced is based on such a channel.
The channel, seen in Fig. 1.1, has a constant width, b, and at any point the depth
of water in the channel is given as h. The wetted perimeter, PW , is defined as the

Figure 1.1: Cross-section of a rectangular prismatic channel

length of the channel in contact with the flow in the cross-section, hence

PW = 2h + b. (1.1)

The cross-sectional area, Ac , is clearly defined as

Ac = h × b.

These two values are then used to define the hydraulic diameter,
Ac
DH = 4
PW
bh
=4 , (1.2)
b + 2h
CHAPTER 1. INTRODUCTION 16

which was first introduced by Buat (1779) and is used throughout open channel
theory as a parameter met in the later equations.
When considering flow in a horizontal channel it is quite probable that the depth
will not remain constant because of various forms of resistance (that are introduced
later). These changes in depth are likely to be gradual in relation to the length they
are considered over; such flows are defined to be slowly-varying flows. Fig. 1.2 shows
flow over a horizontal solid surface which can be used to model such a flow. Taking
axes as shown in the diagram the flow is parallel to the x-axis. The bottom of the
channel is at z = 0 and the depth of the fluid is at z = h. The fluid is defined as
having a free surface, meaning that at z = h the fluid meets with the atmosphere.

h1
Free Surfa
ce
h2

u1
u2

x x1 x2

Figure 1.2: Horizontal cross-section of open channel flow with slowly varying depth

1.2.1 Basic Equations

Considering a fluid with density, ρ, and kinematic viscosity, ν, travelling in the x-


direction, there will be no flow induced changes to the pressure or temperature af-
fecting the density and so we can assume ρ = constant. The flow is considered to be
fully developed so there is no time dependency, hence the continuity equation can be
written as
div u = 0, (1.3)

where u = (u, v, w) is the velocity vector. Considering a control volume from x1 to


x2 with respective depths, h1 and h2 , and velocities, u1 and u2 , as in Fig. 1.2 (with
CHAPTER 1. INTRODUCTION 17

constant channel width b), then by definition of equation (1.3) there is no rate of
change of the volume flow and so the flow rate, Q, is expressed as
Z Z
Q=b u1 dz = b u2 dz,
h1 h2

which, if assuming uniform flow, integrates to give

Q = u1 h1 b = u2 h2 b. (1.4)

We can then find the rate of change of momentum between points x1 and x2 using

Rate of Change of Momentum = ρQ(u2 − u1 ). (1.5)

Assuming that gravity, g = (0, 0, −g), is the only external force on the open
channel system, Bernoulli’s equation along a streamline states that

u2 p
+ gz + = constant,
2 ρ

where p is the pressure, which can be expressed in differential form as

dp u2
+ gdz + d .
ρ 2

As the flow that is considered here is defined above to be gradually varied then we
can assume no acceleration, du = 0, and so the above equation can be integrated to
give
p = patm − ρg(z − h),

where the constant of integration is determined to be atmospheric pressure, patm , at


the surface of the fluid, h. To find the pressure force, Fp , then the above equation is
integrated over the cross-sectional area, Ac , to give

1
Fp = ρgbh2 .
2

To find the net pressure force in the control volume we need only consider the pressure
force at x1 and x2 as hydrostatic pressure is assumed within the control volume.

1
Net Pressure Force = ρgb h21 − h22 .

(1.6)
2
CHAPTER 1. INTRODUCTION 18

1.2.2 Non-Dimensional Analysis

Non-dimensional numbers useful in analysis as they give critical and limiting values
that define certain types of flow, they also provide an essential tool in comparing
results from different pieces of work that will have been produced with different
dimensions and parameters.
The Reynolds number is perhaps the most important non-dimensional number in
fluid dynamics. It was presented by Osborne Reynolds (1883) and shows the ratio of
inertial to viscous forces
UL
Re = (1.7)
ν
where U is a typical velocity and L a length scale. In the case of open channel flow
the hydraulic diameter is used as the length scale giving

UDH
Rec = (1.8)
ν
Ubh
= . (1.9)
ν(b + 2h)
Reynolds described how by using the hydraulic diameter as the length scale then
the critical values are similar to his pipe studies and so we expect laminar flow for
Rec < 2000 and turbulent flow for Rec > 2000 − 3000.
The Reynolds number can also be calculated for cylinders, Rep , by using the
cylinder’s diameter, d, as the length scale which gives

Ud
Rep = . (1.10)
ν
Another important non-dimensional number is the Froude number, credited to
several authors but named after William Froude, it shows the ratio of inertial forces
and the fluid mass
u
Fr = √ . (1.11)
gL
The Froude number has its critical value at one, indicating critical flow. Either side of
this value we have sub- and super-critical flow respectively. For a rectangular channel
the characteristic length scale is taken to be the flow depth and so

u
Fr = √ . (1.12)
gh
CHAPTER 1. INTRODUCTION 19

The coefficients of drag, Cd , and skin friction, Cf , are also important when looking
at the resistance to the flow as in the next section. Such coefficients are defined as
Fd
Cd = 1/2ρu2 A
, (1.13)

and
τ
Cf = 1/2ρu2
, (1.14)

where Fd is the drag force, and A the facial area of the object in contact with the
flow. In both cases the velocity term, u, is taken to be the upstream velocity.

1.2.3 Resistance to Flow

There are numerous ways in which open channel flow can encounter resistance. The
first of these comes not from an obstruction but from the channel itself. The channel’s
walls and bottom have shear stress, τ , which as a result leads to shear stress. There
are several methods of finding the shear stress, which typically rely on a value that
relates to the surface roughness. The shear stress is related to the skin friction
coefficient by equation (1.14), so to find the shear stress we require the skin friction
coefficient. There are three main resistance coefficients; the Chézy coefficient, Ć;
Manning’s nm ; and the Darcy- Weisbach friction factor, f , all of which can be used
to determine Cf . These resistance coefficients can all be defined in terms of velocity
2/3


1 DH
U= sin α, (1.15)
nm 4
r r
8g DH
U= sin α, (1.16)
f 4
r
DH
U = Ć sin α, (1.17)
4
where α is the angle of channel slope (such as in Fig. 1.3 and discussed in the next
section). As the velocity can be expressed in terms of all three of these coefficients
using α and DH then they easily interchangeable. All have their advantages and
disadvantages which are discussed further when comparing literature in the next
chapter. For the analysis in this project the Darcy- Weisbach friction factor was used
for its non-dimensional appeal (nm has units s m−1/3 and Ć m1/2 s−1 ).
CHAPTER 1. INTRODUCTION 20

Table 1.1: Typical roughness heights (Chanson, 1999)


ks (mm) Material
0.01–0.02 PVC (Plastic)
0.02 Painted pipe
1–10 Riveted steel
0.25 Cast iron
1–1.5 Rusted cast iron
0.3–3 Concrete
3–10 Untreated concrete
0.6–2 Planed wood
5–10 Rubble masonry
3 Straight uniform earth

The Darcy-Weisbach friction factor is four times the skin friction coefficient (Chan-
son, 1999):

f = 4Cf
f 2
and so τ= ρu . (1.18)
8

Formulae for finding the Darcy-Weisbach friction factor use the Reynolds number as
defined in (1.9) and are generally based on experimental data. For 750 < Rec <
25, 000 the Prandtl-von Kármán equation (Prandtl, 1935),

1 p
√ = 2.0 log10 (Rec f ),
f

was originally developed for pipe flow but was found to give a good comparison for
open channels. Colebrook (1939) took the basis of this equation further to give the
Colebrook-White formula which is valid for turbulent flow,
 
1 ks 2.51
√ = −2.0 log10 + √ , (1.19)
f 3.71DH Rec f

where ks is defined as the sand the roughness height; this is not dimensionless and is
given in millimetres, typical values are given in Table 1.1.
As mentioned above, equation (1.19) is only valid for turbulent flow, if the flow
is laminar then the roughness of the channel is inconsequential (Graf and Altinakar,
1998) and the friction factor is given by

64
f= , (1.20)
Rec
CHAPTER 1. INTRODUCTION 21

as shown by Moody (1944).


So in order to find resistance given by the shear stress we must first find the
Reynolds number to determine if the flow is laminar or turbulent. Then either equa-
tion (1.19) or (1.20) determines the Darcy-Weisbach friction factor. Equation (1.18)
can then be used and plugged into (1.14).
The other resistance to the flow is an obstruction in the channel; namely the
vegetation in this project. In order to work out the drag coefficient we resolve forces
over a control volume to find the drag force and then use equation (1.13).

1.3 Normal Flow

A common occurrence in open channels is that the channel will be sloped at an


angle, α, to the horizontal. In such a situation the flow can vary in height as in the
previous section, however, it is also quite possible for the flow to have a constant
depth; such an occurrence is known as normal flow and is shown in Fig. 1.3. In the
case of normal flow the constant depth means that there is no net pressure force over
a control volume and also no rate of change of momentum, this can be easily seen
as equation (1.6) uses the difference of depth over the control volume and equation
(1.5) uses the difference in velocity (which is reliant on depth for constant flow rate).

h
L

u
z

x x1
mg

α x2

Figure 1.3: Cross-section of normal flow

As a result the force from the weight of the fluid will be equal to the channel’s
shear stress and the drag on the cylinders. The stress and drag are parallel to the
CHAPTER 1. INTRODUCTION 22

flow direction which is down the slope and so it is easier to resolve down the slope
and take the force from the fluid’s weight as it’s mass multiplied by the component
of gravitational acceleration in the direction of flow. By taking the fluid mass as
m = ρV where V = hbL is the volume of fluid from x1 to x2 in Fig. 1.3.
Resolving down the slope (at an angle α to the horizontal) then the force from
the fluid’s mass, Fm , is
Fm = mg sin α. (1.21)

By choosing to resolve in this direction the shear stress and drag forces met in the
previous section are still parallel to the flow and can be calculated using the same
steps of the previous chapter.

1.4 Discussion

Here we have introduced the theory of open channel flow alongside the equations
which are crucial in its study. The following chapter presents a review of literature
which has used this theory in performing studies relevant or similar to the work
carried out in this project, this theory is then used in later chapters to analyse the
results found in the laboratory.
Chapter 2

Literature Review

2.1 Overview

Open channel flow is found in rivers, canals, flumes and so forth and as a result
there is a wide range of literature studying various aspects of flow and in particular
resistance that may occur to the flow. This chapter presents a review of literature
that focusses on flow past vegetation.
Vegetation can be used as a tool in maintaining the shape of rivers (Stoesser et al.,
2003) as well as preventing coastal erosion (Wilson et al., 2003) and the breaking down
of wave energy (Nepf, 1999). Whilst these scenarios use vegetation to resist the flow
and strengthen barriers the flow is also very important in improving and maintaining
water quality (Bixio, 2007) and in the feeding by filtration of small aquatic organisms
as described by Ayaz and Pedley (1999). This last problem assumes a fluid of more
than one composition and these multiphase problems are valuable in modelling the
flow and dispersion of pollution and sediment (López and Garcı́a, 1998). As well
as sediment flow the idea of larger objects carried by the flow are also considered,
particularly leading from dead or uprooted vegetation which can lead to log jams in
the channel (Hickin, 1984). This last model is certainly realistic, as experienced by
Järvelä (2002) when some of the plants in his laboratory model died and broke off
mid experiment.
The focus of this chapter is to first give a comprehensive review of the methods in

23
CHAPTER 2. LITERATURE REVIEW 24

studying open channel flow past vegetation including the problems and advantages
of the different aspects in each. Results are then presented that will be relevant to
this specific research and so will be a valuable tool for comparison toward the end of
the project.

2.2 Methods

While some areas of study in fluid dynamics lend unkindly to the use of laboratories
or field work, open channel flow studies thrive on results obtained from such places.
Moreover, the building bricks of fluid dynamics’ theory are still here, making use of
the equations of continuity and motion in creating models through book work and
theory alone. The modern approaches of computational fluid dynamics (CFD) are
also brought into the subject.

2.2.1 Modelling Flow Past Vegetation

The standard problem for the physical modelling of open channel flow through veg-
etation is certainly complex, López and Garcı́a (1998) discuss how a basic problem
can be regarded as a function of many variables, the fluid itself has density, velocity
and viscosity whilst the obstructing plant life has a diameter, height and flexibility
and this is all before considering more complex models such as with multiple plant
and with sediment transport. For open channel flow we look at averaging the various
operators in the flow and non-dimensionalising the parameters as described previ-
ously. Typically the Reynolds number is high in open channel flow and can lead to
turbulence (Finnigan, 2000; Järvelä, 2002) however the physical nature of some prob-
lems outlined in the introduction involve slow moving sludge like flows (Hickin, 1984)
and others such as Lee et al. (2004) consider the Reynolds number at zero where the
inertial terms are negligible.
Due to the nature of the research finding the drag coefficient is very important
and studying its behaviour within the vegetation is often an author’s main aim.
The studies of flow through vegetation use the theory as described in the previous
CHAPTER 2. LITERATURE REVIEW 25

chapter with a split between those favouring Manning’s nm (Bixio, 2007; Wu et al.,
1999) and those the Darcy-Weisbach friction factor (Järvelä, 2002; Velasco et al.,
2003); in studying the velocity profiles to find a power law Cheng (2007) compares
values from both to reach a solution sooner. Some authors (such as Klopstra et al.,
1997) also choose to use the Chézy coefficient, though this is much less common
in the studied literature. Fathi-Maghadam and Kouwen (1997) suggests the non-
dimensional advantage of the Darcy-Weisbach factor is currently outweighed by the
widespread use of Manning’s nm . Neary (2003) summarises that as f , nm and Ć
can all be related by equations (1.15), (1.16) and (1.17) it is suitable to choose the
parameter that aids the analysis of results in the most advantageous manner.

2.2.2 Laboratory Methods

As laboratory research allows for controlled circumstances the results often lead to
more obvious correlations between parameters than field studies would. For modelling
plant life cylinders are used as the standard model shape, this allows for use of known
solutions for flows around cylinders that already exist in the world of fluid dynamics.
Naot et al. (1996) also suggest that by using perfectly vertical, rigid, cylindrical rods
then any vertical components of motion in the flow will be proportional to the skin
friction on the rods and pressure drag can be disregarded in the direction parallel
with the vegetation. The arrangement of the cylinders is important to the results
gained and the problems are formulated by considering the density of cylinders per
unit section through the channel giving the solid volume fraction, φ,

area of cylinders
φ= . (2.1)
total area

This type of formulation allows for the averaging formulation of the transport and
momentum equations (as demonstrated in the review article by Finnigan, 2000) which
are used by various authors in their methods of solving these types of problems.
The arrangement of the cylinders allows for random (Tanino and Nepf, 2008) and
ordered (Naot et al., 1996) arrays. In the latter of these Naot describes three designs
that relate closely to the problem descriptions outlined in the introduction; cylinder
CHAPTER 2. LITERATURE REVIEW 26

orientations are used to study how vegetation could be affecting flood plain flow,
flow at the banks of rivers, and flow around corners. In randomly assorting the
arrangement of cylinders Nepf (1999) uses a numerical program to generate a pattern
for the cylinders to be placed in and then repeats this five times over creating a semi-
random flow obstruction. This method is again repeated by Tanino and Nepf (2008)
where many random arrays are used with the density as described above being varied.
The main difference between ordered and randomly designed arrangements is that
in the natural world the plants will be of a more random arrangement1 . However,
it would be impossible to produce models for all such random arrangements, which
is why the solid volume fraction of plants is much more important as a parameter.
Naot et al. (1996) presents models of similar solid volume fractions all producing
similar results arguing in favour of ordered models over random designs. Conversely
it could be argued that however slight the chance any specifically set pattern of
the cylinders could be doctoring the results in one way or another. Similar densities
but completely different designs are easily obtainable by using commercially available
software; Tanino and Nepf (2008) used MATLAB’s built in random number generator
for such a task.
Whilst an excellent basis of results can be gained from the aforementioned exper-
iments it is also important to remember that the vegetation is unlikely to be in the
form of these solid rigid rods. Wilson et al. (2003) constructed rubber type plants
of the same biomechanical properties as real aquatic plant life which would have the
same bend subject to forces as the real plant life would, this study also went on to
fix plant type fronds to these rubbery rods to create even more realistic vegetation.
Järvelä (2002) took the next step by using real plants in the laboratory but this led to
a problem with them not surviving the unnatural laboratory conditions as described
in the introduction. Wu et al. (1999) used a mattress of rubberised horsehair to sim-
ulate vegetation although they admit they had not done physical tests to compare
this material against actual plants their elementary analysis indicated it was similar,
1
Although some biologists present research into the arrangement of plant growth being particu-
larly ordered this is beyond the scope of this research.
CHAPTER 2. LITERATURE REVIEW 27

if not more rigid, than aquatic plant life.


In the laboratory the flow can be controlled to exact specifications. The exact
influx of water can be measured and straighteners used upstream of the obstacles to
give steady flow (Nepf, 1999). Turbulence can be ensured if required, as described in
the methodology of Järvelä (2002), by sending the upstream flow through chunks of
rock. The laboratory also allowed for the water to be of controlled composition; in
the field this would not be possible. Although the controlled composition of the fluid
is unrealistic in the field, it is unusual for one environment’s flow composition to be
dramatically changed from one time length to the next. This type of problem is more
concerned with sediment transport which is studied in the field by Hickin (1984). The
actual studying of flow in the laboratory varies on what is to be observed. The veloc-
ity profiles behind the arrays of obstacles can be measured using Acoustic Doppler
Velocimeters (ADV) as demonstrated by Ghisalberti and Nepf (2004). The height
of the flow is also observed to change during flow and this can be measured using
gauges that register change dependant on the depth they are submerged at (Tanino
and Nepf, 2008). Elementary observation methods can also be employed such as the
injection of dye into the flow to monitor the flow direction and dispersion which was
performed by Nepf (1999). With the exception of Järvelä (2002) all of the laboratory
tests discussed use cylindrical obstructions as a representation of vegetation, Järvelä’s
study thus allowed an observation which other authors often dismiss, that being the
external leaves on the plants; Wilson et al. (2003) also showed an insight into the
problem with the attachment of fronds to his cylindrical plant mechanisms. This
type of study shows laboratory experiments are restricted in terms of realism and
indicate the need for field studies.

2.2.3 Field studies

Field studies take away the need for modelling the actual plant canopies but as a
result sacrifice many of the desired controlled conditions. The flow itself is certainly
more complicated and even in moderately straight channels there will be swirling
CHAPTER 2. LITERATURE REVIEW 28

effects (Järvelä, 2002). Furthermore, unlike in the laboratory, the wind direction
must be considered and the effect that it will have on both the water currents and the
vegetation (Worcester, 1995). When taking the water height into account Worcester
took measurements at several areas, which gave an average for the area of study. As
in the laboratory, dye was used to observe the flow, with Nepf (1999) repeating his
injecting model. Worcester took a more basic approach and used a dye filled balloon
which was pierced in the flow and then photographed from above to show the flow
patterns.
Jadhav and Buchberger (1995) presents a study that differs from all others dis-
cussed so far in that it uses time as one of its main parameters. Detention times are
compared against plant densities in order to investigate the efficiency of vegetation
aiding the removal of pollutants in wetlands. Focussing on the same problem, but
with a different methodology, Bixio (2007) constructed a flume in the field and took
depth measurements in order to investigate the vegetation’s resistance.
In the field, the dimensions and parameters of the vegetation are more realistic
but certainly harder to control, Nepf (1999) considered areas where the grass was
cylindrical to coincide with his laboratory studies and the solid volume fraction was
computed by counting the numbers of plants in unit squares. As well as these, Nepf
also collected samples of the plants from the field in order to calculate an average
leaf count. Bixio (2007) categorised the plant life into three types which covered the
average vegetation dimensions in the wetlands she was working in.

2.2.4 Numerical Methods

Numerical methods allow for a large number of calculations to be carried out and
also for a greater amount of similar orientations than lab work can afford time wise.
Numerical solvers also have the ability to control the extreme conditions that affect
field work. It is not surprising then that the study of open channel flow past vegetation
has moved into this realm of study. Koch and Ladd (1996) used lattice-Boltzmann
simulations for modelling the drag against the Reynolds number; owing to the freedom
CHAPTER 2. LITERATURE REVIEW 29

described above, these numerical methods they were able to perform calculations on
very large numbers of random and ordered cylinders arrays. Numerical research is,
however, very unpredictable for solving new problems as it is very unlikely that the
program will run perfectly first time, Koch and Ladd show the strength of numerical
solvers however, by comparing their results with existing laboratory data.
López and Garcı́a (1998) investigate the transportation of sediment through veg-
etation using a turbulent k −  model. This model is particularly interesting as it
usually neglects the viscous terms in the formulation of the equations found for mo-
mentum, kinematic energy and dissipation. This cannot be assumed in their model as
flow in vegetation holds high importance on inertial diffusion. Alternately to López
and Garcı́a, Neary (2003) uses a k − ω turbulence model for the transport equa-
tions. In his model, Neary varies the height and densities of submerged vegetation
and solves the problem for various Reynolds numbers. Both Neary and López and
Garcı́a’s models produce similar results meaning that the k − ω model is validated
alongside the latter’s k −  model.
The above turbulence models are created for solving specific open channel flow
problems. López and Garcı́a draw attention to the fact that there are very few
physically based models that can assess the influence of different flow and vegetation
properties. Stoesser et al. (2003) present a 3D solver for a large scale river which is
built to specifications in particular around the river Rhine but could be adapted to
other similar models. The main difference of this model to the previous ones is the
large scale scenario that is involved. Rather than considering shallow depth flow, or
indeed scaled down laboratory experiments, Stoesser et al. look at the big picture.
This means that the usual roughness models cannot be used as the authors look at
larger scale flow through forests2 . As there is no large scale methodology of finding
a large scale roughness coefficient Stoesser et al. adjust the value in their model to
give the physical observed hydraulic behaviour.
2
It is perhaps important to point out that in this case the flow is not affected by the small
vegetation (Wu et al., 1999).
CHAPTER 2. LITERATURE REVIEW 30

Research in flow through porous materials considers the permeability of an ob-


ject rather than the density that is considered in the vegetated open channel. By
closing the spacing between the cylinders to a very small amount and indeed have
the cylinders of similar thickness as shown by Ayaz and Pedley (1999) we are now
essentially monitoring the permeability rather than the density of the array; Ayaz
uses this technique for the simulation of aquatic organism’s feeding.

2.3 Results

In studying the depth profile along the channel Bixio (2007) presents a useful set
of results as shown in Fig. 2.1. This graph shows the depth along the x-coordinate
for four different controlled downstream depths, each with three different flow rates
(where green < blue < red). From these we see that with a low downstream control
depth the flow will converge to this depth toward the end of the channel. As would
be expected, the higher flow rates lead to an increased flow depth. We can also note
that for a low downstream control the surface tends to have a curved shape and as
the downstream depth increases the surface displacement become more linear.

Figure 2.1: Flow Depth along channel (Bixio, 2007)

Various studies yield similar results for the drag coefficient despite the models be-
ing for different purposes. Several studies show a high drag coefficient for laminar flow
CHAPTER 2. LITERATURE REVIEW 31

which decreases and approaches zero as the flow becomes turbulent. Fig. 2.2 shows
the drag coefficient against the square of velocity, taken from by Fathi-Maghadam
and Kouwen (1997). This particular graph represents non-submerged pine saplings
as the test specimen and shows that for increasing velocity the drag coefficient de-
creases from it’s peak of Cd ≈ 2 at u2 ≈ 0 m s−1 to Cd ≈ 0.5 at u2 ≈ 0.6 m s−1 .
The different series represented by the key are for test runs with different upstream
flow depths and whilst the above general trend of results is observed the decreasing
drag coefficients are not monotonic for any of these series. Fig. 2.3 and Fig. 2.4 show
the drag coefficient against the Reynolds number, produced by Järvelä (2002) and
Tanino and Nepf (2008) respectively. The first of these uses the flow depth as the
length scale in the Reynolds number (i.e. Re = (uh/ν) whilst the second uses the
cylindrical Reynolds number, Rep , defined in eqn (1.10). Despite figures 2.2,2.3 and
2.4 all using a different parameter to compare the drag coefficient against their results
are quite similar in terms of Cd range and the decrease with this chosen parameter.
None of the studies concluded here use varying viscosity so the two Reynolds numbers
are increasing with the inertial term which explains this pattern. Fig. 2.4 also

Figure 2.2: Drag Coefficient vs Velocity ( m s−1 ) (Fathi-Maghadam and Kouwen,


1997)

employs lines of fits to the data as the experimental results did not fully cover the
CHAPTER 2. LITERATURE REVIEW 32

Figure 2.3: Cd vs Re (= uh/ν) (Järvelä, 2002)

range of Reynolds numbers required for full comparison by the authors. Fig. 2.5 uses
the Reynolds number for an open channel, Rec , and defines a new drag coefficient,
Cd 0 = λCd where λ is a vegetal area coefficient which depends on the properties of the
vegetation and defined over a unit length and as a result is non-dimensional. Over a
similar range of Reynolds numbers, this graph from Wu et al. (1999) shows a different
Reynolds and drag coefficient dependence. Furthermore, it is the models’ arrival
at a state of homogeneous turbulence that can lead to the most interesting results,
this is often triggered by increasing the solid volume fraction which in turn leads to
secondary currents in the flow (Naot et al., 1996), agreeing with (Tanino and Nepf,
2008).
The aforementioned studies focus on unsubmerged flow, which Wu et al. (1999)
compares with submerged flow using his laboratory models of varying slopes. In these
Wu finds that an increased slope leads to an increased drag coefficient, although for
fully submerged flow the drag is much more closely influenced by the solid volume
fraction. Wu also shows the roughness coefficient is not related to the slope at all,
but rather on the depth of the flow itself. These submerged flows with fronds on
CHAPTER 2. LITERATURE REVIEW 33

Figure 2.4: Drag Coefficient vs Rep (Tanino and Nepf, 2008)

Figure 2.5: Vegetal Drag Coefficient vs Rec (Wu et al., 1999)

the vegetation are much more likely to create mixing layers between the surface and
canopy roof layer, although the layer at the bed thus finds little interaction in the
flow (Wilson et al., 2003). Shear layers that are formed in this sort of interaction
help lead to the transport of sediment in more than one dimensional direction (López
and Garcı́a, 1998). López and Garcı́a’s results discuss the need for studying sediment
transport at different buoyancy heights within the flow and would perhaps make use
of the theory surrounding Stokes’ number as employed by Ayaz and Pedley (1999).
Whilst flow density is a factor to be determined for the majority of the studies
described here, Ayaz and Pedley (1999), as previously mentioned, looks more at the
permeability of tightly spaced cylinders in modelling feeding by filtration. For this
CHAPTER 2. LITERATURE REVIEW 34

study the drag, whilst measured and noted to increase with the Reynolds number,
the decreasing permeability with the increasing Reynolds is the interesting result.
The study by Nepf (1999) in the comparison of ordered and random cylinder
arrays leads to the discovery that for high and low solid volume fractions in the
extremes the coefficient of drag behaves almost identically, in the random arrays the
drag coefficient is always higher than if the cylinders were ordered. For high Reynolds
numbered flows Järvelä (2002) discovered that the leaves on the plants made little
difference and were streamlined with the flow. This was further also observed by
Hickin (1984) who made note of the combing down of grass type plants in high speed
flows. Järvelä also noted that although a massively turbulent system (with Reynolds
numbers up to 187,000) the Froude number remained less than 0.25 in all cases
considered, therefore meaning there should be no jumps in the flow and it should
remain stable.
The numerical models of study showed favourable results when compared with
experimental models. Ghisalberti and Nepf (2004) carried out both laboratory and
numerical studies over the same conditions to compare results and found very close
solutions. One piece of research that stands out from all of the others is the CFD
model by Stoesser et al. (2003). As this model is being created for the specific job of
solving flood flow at the river Rhine the results that are presented are in actual times,
velocities and so forth. This makes it not only an interesting paper through difference
but for the interdisciplinary reader unfamiliar with the depths of fluid mechanics a
much more comfortable read.

2.4 Discussion

In conclusion to the presented literature, the problem of open channel flow past
vegetation is simple by design and yet complex by analysis. The laboratory modelling
of plant life can be done with cylinders, flexible rods, hair mattresses and much
to the experimenters imagination right up to actual vegetation itself. However the
more complex of these are not necessary in elementary channel flow and it is only
CHAPTER 2. LITERATURE REVIEW 35

where the experimenter starts to venture into turbulent submerged flow that the finer
points of the plant lifes’ biomechanisms come into serious contention. Field studies
themselves are a valuable asset to any piece of research and whilst throwing up more
complicated conditions these are the conditions that finalised models should be ready
to incorporate so early analysis is of great benefaction.
As is usual with any study in fluid mechanics the equations of motion and con-
tinuity are still the key building bricks to the solution. Many authors have tackled
many different problems regarding vegetated channel flow including sediment trans-
port to underwater currents and right up to the massive scale Rhine model. All the
models link together in one form or another meaning that future models will be all
the easier to construct by lending a priori knowledge.
Much like a workman never being satisfied with his tool, most of the research calls
for further research in different areas to where the current author is working. Hickin
(1984) starts this off from his overview of field study where he calls for experimental-
ists to undertake some of the problems that he outlines. Worcester (1995) suggests
the need for further slow stream research models which could be of help in finding
the ambient flow for less dense vegetation. Hickin also suggests a need for models for
flow of flood plains with sparse vegetated life and remaining large obstacles (such as
tree stumps) where forests have been subject to falling. This type of model appears
the CFD solution by Stoesser et al. (2003) although twenty years down the line.
The next chapter introduces the apparatus that was used to produce similar
experiments for this project; in these experiments the control parameters to be used
were inspired by the literature studied here.
Chapter 3

Experimental Procedure

3.1 Overview

When performing laboratory experiments it was important that the right apparatus
was available and that it was used safely and correctly. To gain results of the highest
standard in the time allowed it was also vital that the work was planned in advance
meaning as much laboratory time as possible. This chapter first describes the ap-
paratus that was available before describing the manner in which it was used for
gaining various results. The end of the chapter describes in detail the different series
of experiments that were planned and carried out during the course of the project.

3.2 Apparatus and Methods

3.2.1 The Flume

To perform the laboratory experiments a tilting flume was used, as seen in Fig. 3.1.
The flume was made up of two reservoirs connected by an open channel measuring
272 cm in length, 10.4 cm inside width and with walls 30 cm high. The channel floor
was flat and had glass walls to allow inspection of flow. Water was pumped into
Reservoir 1 where it settled before running into the channel; this allowed full flow
control by preventing any unwanted turbulent forces. The flow rate was controlled

36
CHAPTER 3. EXPERIMENTAL PROCEDURE 37

by a valve which took water from a supply circulating around the laboratory. The
second reservoir contained an adjustable weir to control the flow depth inside the
channel before the water exited the flume and back into the main circulation system.
A rail ran along the top of the channel which depth gauges could sit on and move

Channel Rail Reservoir 2


Flow in
272 cm

Flow 30 cm

Weir

Pivots
Reservoir 1
Screw Flow out

Figure 3.1: Diagram of the laboratory flume used for all experiments

across to measure depths; the rail had a rule attached so the position of the depth
gauge could also be recorded. The flume sat on two sets of legs, the first on a pivot
at a fixed one metre off the ground, between the first reservoir and the start of the
channel. At the other end of the channel the second pivot was attached to a large
vertical screw which could be adjusted to tilt the angle of the channel.

3.2.2 Test Specimen

The experiments investigated flow past two types of cylinders; rigid and flexible rods.
The rigid rods were metal and measured 6 mm in diameter, 200 mm in length. The
flexible rods were liquid filled plastic1 rods measuring 5 mm in diameter, 202.2 mm
in length.
For fixing the rods in place plastic boards were used measuring 300mm in length,
100mm in width, and 11mm in height. Fifteen rows of five 6mm holes were drilled
into the boards in a square array as shown in Fig. 3.2. As the flexible rods were
1
They were in fact used glow sticks that had a desired flexibility - probably stiffer than many
plants.
CHAPTER 3. EXPERIMENTAL PROCEDURE 38

a loose fit in the 6mm holes they were held in place using plastic connectors, this
meant a small length of the rods (≈ 6 mm) were slightly thicker (≈ 6 − 7 mm) at
the base. There were a total of six available boards with holes drilled into them for

10 mm 20 mm

10 mm

6 mm

100 mm
20 mm

300 mm

Figure 3.2: Plan view of plastic board (depth 11 mm)

performing experiments with, as these boards would not fill the entire length of the
channel further boards (without holes) were used to keep the bottom of the channel
level which ensured the boards did not affect the flow.

3.2.3 Taking Measurements

To find the depth of the flow two depth gauges were used, both had vernier scales
and allowed measurements to be taken to tenths of a millimetre. The advantage of
using two depth gauges was that they could be set to specific depths and placed in
separate positions whilst the flow rate valve, downstream weir, or screw were being
adjusted; this saved considerable time between certain tests.
In setting the angle of the flume the flow was stopped but with water kept in the
viewing channel so there was no flow movement. When the flume was required to be
horizontal a spirit level was used and the depth gauges could then be used to tweak
the final degree of tilt. When an angle was required the rule along the top rail was
used to place the two depth gauges at a set distance and their required difference in
depth calculated using the tan rule.
CHAPTER 3. EXPERIMENTAL PROCEDURE 39

3.2.4 Calculating the Flow

As described previously, the flow rate, Q, was adjusted by a valve connected to


Reservoir 1, there was no meter on the valve and so to calculate Q the volume of
flow was recorded over a time period and Q calculated using the formula

Change in volume
Q= .
Time elapsed
To do this, after leaving Reservoir 2 the flow was diverted into a separate tank under-
neath the lab rather than immediately into the circulating system. The tank had a
float on it connected to a vertical ruler which could be multiplied by a factor to find
the current tank volume in m3 s−1 . The volume could be measured to an accuracy2
of 2.76 × 10−4 m3 and time to one hundredths of a second giving an acceptable degree
of accuracy to this method. However, some accuracy was lost as the flow diversion
into the tank was not instantaneous and it could not be ensured (although strong
attempts were made) that the stopwatch started and stopped at the exact moment
of flow diversion.

3.3 Description of Experiments

The apparatus provided allowed for a range of different control parameters. The solid
volume fraction was altered by varying the number of rods per row on the boards.
Using the board dimensions that are described above equation (2.1) gives
2
15m πd4
φ= , (3.1)
0.3 × 0.104
where m is the number of rods per row and the rod diameter, d, is given in metres. As
the rigid and flexible rods had different diameters their solid volume fraction would
be different when they had the same rods per row. Defining mr as the number of
rigid rods per row and mf the number of flexible rods then solving (2.1) for equal φ
yields
mr = 1.44mf .
2
The vertical ruler had markings of one tenths of an inch which could be then read to tenth
accuracy between the markings. The conversion factor was 1” equivalent to 0.0276 m3 .
CHAPTER 3. EXPERIMENTAL PROCEDURE 40

Values of φ for different m’s are shown in Table 3.1. This table presents the value of
φ for 1–4 rigid and flexible rods per row and the corresponding number of rods per
row of the other rod type. The solid volume fraction is also calculated for 2.5 rods
per row which is used in Series D of the tests. The values of φ in Table 3.1 are clearly
distinct when read to three decimal places and so shall be referred to to this degree
of accuracy from now on. Two further columns, 15mr and 15mf , state the number of
rods per board. For instance if there are two rigid rods per row and equivalent solid
volume fraction could be measured in flexible rods by using 2.88 flexible rods per
row; the 15mf column gives a value of 43 rods per board, which is a better estimate
to work with. From looking at the table it can also be clearly seen that the solid
volume fraction calculated from mr = 4 cannot be recreated with flexible rods as this
requires mf > 5 when there were in fact only five holes on each row of the board.

Table 3.1: Investigated Solid Volume Fractions


mr 15mr mf 15mf φ (6 d.p.)
1 15 1.44 21.6 0.013593
2 30 2.88 43.2 0.027187
2.5 37.5 3.6 54 0.033984
3 45 4.32 64.8 0.040780
4 60 5.76 86.4 0.054374
0.69 10.42 1 15 0.009440
1.39 20.83 2 30 0.018880
1.74 26.04 2.5 37.5 0.023600
2.08 31.25 3 45 0.028320
2.78 41.67 4 60 0.037760

If there are m rods on each row of n holes then there are σ = nCm ways of placing
the rods into the holes. To ensure the rods were equally spread across the channel
length as well as width it was decided each arrangement should use all permutations
and then repeat the pattern every σ rows. This meant that for these particular
experiments we would have σ = 5 for m = 1, 4 and σ = 10 for m = 2, 3; thus the
arrangements for two and three rods per row would be repeated twice as many times
as for when there are one or four rods.
CHAPTER 3. EXPERIMENTAL PROCEDURE 41

3.3.1 Series A - Flow past different arrangements of equal φ

The aim of series A was to investigate the importance of the arrangement of the
rods and if, as suspected, the solid volume fraction could be used as a valid control
parameter. The channel was set to be horizontal and four boards were used to place
the rods in. The series investigated three separate values of φ with two separate
arrangements making six tests in total. Three flow rates were used through per
test and the depths of the fully developed flow were recorded at intervals along the
array. Table 3.2 shows the six tests and their corresponding solid volume fraction
and array arrangement3 . The number of times that the arrangement is repeated is
also presented alongside the test number.

Table 3.2: Test parameters for Series A


Test No. φ Array arrangement No. times repeated

Test A1 0.014 12

Test A2 0.014 12

Test A3 0.027 6

Test A4 0.027 6

Test A5 0.041 6

Test A6 0.041 6

To analyse the results the Mathematica Fit function calculated a linear fit of the
surface displacement, the gradient of this could then be compared with the flow rates
3
In the array arrangement diagrams the circles are holes in the board with the filled ones repre-
senting a rod, the flow is taken from left to right.
CHAPTER 3. EXPERIMENTAL PROCEDURE 42

to see if any relation existed.

3.3.2 Series B - Controlled depth changes past arrays of vary-

ing φ

Series B of experiments looked at flow past the four solid volume fractions corre-
sponding to mr = 1 − 4. These four tests had one arrangement of rods each, shown
in Table 3.3. As can be seen from the table all of the tests in Series B used different
arrangements to their Series A counterparts except Tests Test A6 and Test B3 which
are the same.
The downstream depth was controlled in Series B using the downstream weir to be
approximately the same for all of the tests. The flow rate was adjusted and recorded
to give three predecided upstream depths, hup . As well as measuring the flow rate to
give the required upstream flow depths the depth of the flow was measured at even
intervals (every five rows) along the arrays.

Table 3.3: Test parameters for Series B


Test No. φ Array arrangement No. times repeated

Test B1 0.014 18

Test B2 0.027 9

Test B3 0.041 Test A6 9

Test B4 0.054 18

3.3.3 Series C - Normal flow past arrays of varying φ

The third series of tests was to investigate normal flow through arrays of equal density.
The flume was tilted to angles of α = 1 ◦ and 2 ◦ and the downstream weir adjusted
CHAPTER 3. EXPERIMENTAL PROCEDURE 43

alongside the flow rate to acquire approximate constant depths of 3.17 cm and 5.17
cm. In order to achieve as near a constant depth as possible two depth gauges were
used, one upstream of the array of rods and the other downstream; both of these
were set to the same required depth. The downstream depth was controlled by the
weir in Reservoir 2 of Fig. 3.1 before the valve controlling the flow rate adjusted
the upstream depth; the upstream adjustments had some effect on the downstream
depth and so the weir needed to be altered once more after changing the flow rate.
Similarly, the weir adjustment affected the upstream flow depth and as a result to
achieve normal flow each test required alternate tweaking of both the valve and weir.
The adjustment of the valve had more effect on the downstream depth than the weir
had of the upstream and so the most efficient practise was to first use the weir before
using the valve to overestimate the upstream depth slightly then reducing this with
the weir. The set of tests for Series C is described in Table 3.4.

Table 3.4: Test parameters for Series C


Test No. α (degrees) φ Array arrangement No. times repeated

Test C1 1 0.014 Test B1 18


Test C2 1 0.027 Test B2 9

Test C3 1 0.041 9

Test C4 1 0.054 18

Test C5 2 0.014 Test B1 18


Test C6 2 0.027 Test B2 9
Test C7 2 0.041 Test B3 9
Test C8 2 0.054 Test B4 18

3.3.4 Series D - Ordered arrangements for equal φ

Series B and C rely on the results of Series A; the fact that different arrangements of
similar φ will not affect the results as long as all σ permutations of rods are used in the
CHAPTER 3. EXPERIMENTAL PROCEDURE 44

arrangement. Series D investigates the opposite of this by looking at three distinct


patterns of evenly spread out rows, P2, P3 and P2.5 (seen in Fig. 3.3), corresponding
to the number of rods per row. The third of these, as can be seen from the diagram,
has an average of 2.5 rods per row and is alternate rows of P2 and P3. As in Series
B, the horizontal flume cases of Series D were performed under three flow rates to
acquire specific up- and downstream depths. For the normal flow cases each test
looked at maintaining the two approximately constant depths previously described
at two angles. All of the tests of Series D are described in Table 3.5

P2 P3 P2.5
Figure 3.3: Repeated patterns for Series D

Table 3.5: Test parameters for Series D


Test No α (degrees) φ Array arrangement No. times repeated

Test D1 0 0.027 P2 90
Test D2 0 0.041 P3 90
Test D3 0 0.034 P2.5 45
Test D4 1 0.027 P2 90
Test D5 1 0.041 P3 90
Test D6 1 0.034 P2.5 45
Test D7 2 0.027 P2 90
Test D8 2 0.041 P3 90
Test D9 2 0.034 P2.5 45

3.3.5 Series E - Flow past flexible rods

Series E looked at tests similar to those seen in Series B and D but were completed
with 5mm flexible rods. In total four tests were carried out, all in a horizontal
channel, with four upstream depths for each. The parameters for the tests are shown
CHAPTER 3. EXPERIMENTAL PROCEDURE 45

in Table 3.6, Test E3 had no particular arrangement of rods but instead started out
with the design of Test B1 of one rod per row and seven extra rods were added
to make the solid volume fraction equal to Test B1’s as guided by the values in
Table 3.1, for fairness these rods were placed on alternate rows from the second
row of each board and spread evenly between the choice of five holes, however, no
particular arrangement was planned or recorded.

Table 3.6: Test parameters for Series B


Test No. φ Array arrangement No. times repeated

Test E1 0.009 Test B1 18


Test E2 0.014 22 rods per board N/A
Test E3 0.019 Test D1 90
Test E4 0.024 Test D3 45

3.4 Discussion

This chapter has described the apparatus that was available for experimentation and
the methods that would be used in the laboratory. The different series of tests to be
completed have been described and tabulated to provide a solid reference tool. The
following chapter describes the analysis of Series A before providing in depth working
on how the other series’ tests are analysed using the theory of Chapter 1.
Chapter 4

Analysing Results and Interpreting


the Data

4.1 Overview

Of the five series of tests described in the previous chapter, Series B, C and D would
be interpreted in similar fashions, using the theory described in the introduction.
Furthermore, Series E also employed these methods whilst also requiring more qual-
itative analysis. The tests in Series A, however, were designed as a basis to justify
assumptions made in the choice of φ as a control parameter in the various series. As
a result the results of Series A are described in detail at the beginning of this chapter.
The analysis of the latter four series is far from simple and so for completeness in
the latter part of this chapter the data obtained for Tests B3 and C4 are analysed.

4.2 The Results of Series A

Six tests were carried out with the channel set to be horizontal, each one investigated
three surface profiles for varying flow rate, Q. Each test used a different arrangement
of rods but were paired in terms of solid volume fraction so that Tests A1 and A2
had φ = 0.014, A3 and A4 used φ = 0.027, and A5 and A6 measured φ = 0.041.
Observations of Test A3 are seen in Fig. 4.1, where the flow direction is from left

46
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 47

to right. This photograph shows that in a horizontal channel the flow depth differs
from being higher upstream and reducing gradually through the array, towards the
end of the array the depth appears to decrease more suddenly; all tests in Series A
observed similar results to this. Increasing the flow rate also increased the depth at
all points along the channel for all tests carried out.

Figure 4.1: Flow past an array with φ = 0.027

Depths were recorded at various points throughout the channel so surface profiles
could be presented graphically, such as in Fig. 4.2. This figure shows the results of
three flow rates in both Tests A5 and A6. x = 0 is the depth measurement in front
of the first row of rods. The plots show a concrete representation of what is observed
in Fig. 4.1, it is also observed that the surface profiles steepen with increasing flow
rate. In order to investigate the change in gradient linear functions were fitted to the
data, however, as discussed above, the profile falls off suddenly downstream, and so
the end points of data were not used in the linear fitting. Furthermore, the upstream
values show a slight increase in depth which does not aid Series A’s aim of studying
flow inside the array and so some upstream values were discounted. As a result the
linear functions were generated using Mathematica’s Fit function and used the data
range of 0.2 < x < 1.
An example of the linear functions fitted calculated in the above manner is shown
in Fig. 4.3. The three flow rates and the depths observed in the laboratory are shown
as points with the relevant linear fit drawn in the central portion as a line in the same
colour. Similar linear functions were created for all other tests in Series A and their
absolute gradients could be compared against the flow rate as seen in Fig. 4.4. From
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 48

0.14
Test A5, Q = 9.9E-4
Test A5, Q = 1.8E-3
Test A5, Q = 2.7E-3
Test A6, Q = 7.0E-4
0.12 Test A6, Q = 1.7E-3
Test A6, Q = 2.6E-3

0.1

0.08
Depth (m)

0.06

0.04

0.02

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
x (m)

Figure 4.2: Surface depth for Tests A5 and A6 (φ = 0.041)

0.12
Q = 1.1E-3
Q = 1.7E-3
Q = 3.0E-3

0.1

0.08
z (m)

0.06

0.04

0.02

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
x (m)

Figure 4.3: Laboratory data (points) and the linear functions (lines) fitted to central
portions for Test A3

the graph we see that an increasing flow rate leads to an increase in the steepness for
all solid volume fractions that were investigated. More relevantly, the arrangement of
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 49

the rods does not appear to be a contributing factor as regardless of this the gradient
increases with the flow rate at an almost equal pace. Some errors are present as
the plots are not perfectly straight, noticeably for low values of Q with φ = 0.027.
This slight anomaly in the results is not any indication that our conclusion that the
arrangement pattern having little effect on the depth is incorrect. Certainly, looking
at the low flow rate results for all solid value fractions they are less in tune with the
other results; this could indicate that the flow rate calculations are less accurate for
lower flow rates. Bearing in mind the method of flow rate calculations (section 3.2.4)
it is quite possible for lower flow rates that there will be a greater loss of accuracy.
From Fig. 4.4 we can also see that it is not only the flow rate that is a working as
a control parameter here. Whilst the exact arrangement is unimportant as long as all
σ permutations are used it is clear that the solid volume fraction is significant. When
φ = 0.041 the gradient changes more dramatically over a smaller range of Q than
when φ = 0.014, the plot for φ = 0.027 lies between theses.
The objective of Series A was to investigate if tests could be performed for var-
ious solid volume fractions without having to consider the exact arrangement as a
parameter. The results above concluded this was the case and at the same time
gave an insight into the behaviour of flow inside the array. The observation that
the depth decreased quite significantly further downstream indicated that for more
accurate analysis within the array a longer control area was required; instead of the
four boards used in Series A all subsequent series used six boards. The final analysis
of Series A showed that higher values of φ led to an increased steepness of flow within
the array, meaning that there was a greater range of the up and downstream depths,
essentially indicating larger solid volume fractions gave more resistance to the flow;
something that was to be investigated in depth in Series B.

4.3 Calculating resistance to the flow – Test B3

To calculate the drag coefficient on a single rod in the array a series of calculations
are performed using the equations from section 1.2. Recall that in Series B depth
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 50

0.05
phi = 0.014
phi = 0.027
phi = 0.041

0.04

0.03
Abs(gradient)

0.02

0.01

0
0 0.001 0.002 0.003 0.004 0.005
Q (m^3 / s)

Figure 4.4: Absolute depth gradient against flow rate for varying φ.

measurements were taken every five rows of rods and that the flow was fully developed
and steady. As a result there are three pieces of information available; two sets of
(x, z) coordinates describing depth at a point along the channel, and the flow rate,
Q.
Looking at particular values obtained in the laboratory, Test B3, the desired
upstream depth was ≈8 cm and the desired downstream depth ≈0.4 cm; a flow rate
of Q = 1.17 × 10−3 m3 s−1 gave this result. This particular test was chosen as the
other tests were performed either side of Test B3’s solid volume fraction and this
particular upstream depth; by choosing a test sample in the middle of the control
parameters the results represent a mid point to work from. The recorded upstream
depth was actually h1 = 8.05 cm and five rows down the depth was h2 = 7.93 cm.
Fig. 4.5 shows this control volume, all values are in metres for use in the equations.

The channel width is b= 0.104 m and so using eqn (1.4) the upstream velocity
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 51

h1 = 0.0805
h2 = 0.0793

u1
u2

x1 = 0 x2 = 0.101

Figure 4.5: Control Volume of first five rows of rods of Test B3

can be calculated

Q
u1 =
bh1
= 0.140 m s−1 (3 d.p.),

and downstream
u2 = 0.142 m s−1 (3 d.p.).

The Froude numbers at each x-coordinate can then be calculated by equation (1.12)
(taking g= 9.81 m s−2 ),

u1
Fr1 = √
gh1
= 0.158 (3 d.p.)

Fr2 = 0.162 (3 d.p.).

The Froude number is low for both cases, increasing with the velocity as would be
expected for such a small change in the fluid depth. The Froude numbers are well
below the critical value (Fr= 1) and if the trend of the slight change from Fr1 to Fr2
continues down the flow it does not suggest that the critical value will be obtained
for the test; certainly no hydraulic jumps are expected but rather the flow depth will
continue to decrease and velocity increase as we move further downstream.
The open channel Reynolds number, Rec , requires the hydraulic diameter DH ,
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 52

and wetted perimeter, PW . The latter of these is found with equation (1.1),

PW1 = b + 2h1

= 0.2650 m,

PW2 = 0.2626 m.

The hydraulic diameter is then calculated with equation (1.2)


4bh1
DH1 =
PW1
= 0.1264 m (4 d.p.),

DH2 = 0.1256 m (4 d.p.).

Using (1.9) now gives


u1 DH1
Rec1 =
ν
= 17734 (5 s.f.),

Rec2 = 17896 (5 s.f.),

where the kinematic viscosity is taken to be 1.0 × 10−6 m2 s−1 . These values describe
a turbulent regime, indicating the Darcy-Weisbach formula (eqn. (1.19)) can be
used to help find the channel’s resistance if required. Indeed, the ratio between the
depth and width of the channel is of order one so the shear stress of the channel’s
boundaries are important. Table 1.1 suggests that for the channel used a value of
ks = 0.01 − 0.02 mm should be used; the average (ks = 0.015 mm) was taken to be

the sand roughness coefficient for all calculations. Equation (1.19) finds 1/ f , which
is also used explicitly inside the formula. To combat this, the equation is solved by
iterating i times, with i = 0 an initial guess, so the equation now takes the form
 (i)  (i−1) !
1 ks 2.51 1
√ = −2.0 log + √ . (4.1)
f 3.71DH Rec f

Owing to the logarithmic behaviour the formula converges quickly so any positive
√ (0)
value is suitable for the initial guess; for simplicity an initial guess of 1/ f =1
was used. Using the above values of DH1 and Rec1 the formula was iterated over the
initial guess as shown below:
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 53

√ (i)
Iteration (i) 1/ f
0 1
1 2.986018068
2 2.978453720
3 2.978482412
4 2.978482303
5 2.978482304
6 2.978482304

As can be seen the result converges to nine decimal places in only six iterations.
Squaring and inverting the result gives the friction factor as

f1 = 0.1114 (4 d.p.),

which can then find the shear stress, τ , using equation (1.18)

f1 2
τ1 = uρ
8 1
= 0.2496 N m−2 (4 d.p.),

where the water’s density has been taken as ρ = 1000 kg m−3 . In a similar fashion it
is found that
f2 = 0.1122 (4 d.p.),

and so
τ2 = 0.2649 N m−2 (4 d.p.).

The stress acts on the wetted perimeter along the length of the control volume,
so this gives

τ1 + τ2 PW1 + PW2
Shear Stress Force = × × (x1 − x2 )
2 2
= 7.5 × 10−3 N.

The other forces acting in the control volume are the net pressure force and the
drag, Fd , on the cylinders; resolving these forces in the direction of flow will give the
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 54

rate of change of momentum. The net pressure force is evaluated using equation (1.6)
1
Net Pressure Force = ρgb(h21 − h22 )
2
= 0.098 N (3 d.p.).

The rate of change of momentum uses equation (1.5)

Rate of Change of Momentum = ρQ(u2 − u1 )

= 2.495 × 10−3 N (3 d.p.).

Clearly the net pressure force is the larger of these three forces and so the drag on
the cylinders will be mainly affected by this. Resolving in the direction of the flow
the forces are balanced as

Rate of Change of Momentum = Net Pressure Force − Shear Stress − Drag,

which using the above values gives

Fd = 0.0878 N (4 d.p.).

This drag force is split between all of the rods in the control volume whilst each
rod experiences a different drag force because of the varying velocity and flow depth.
Rather than work these drag forces out individually, all rods in the control volume
are assumed to have equal drag and thus a drag coefficient is found for the entire
control volume. To do this the area, Ap , is found by averaging the depths at x1 and
x2 , hence
(h1 + h2 )
Ap = dmn,
2
where d is the rod diameter, m is the number of rods per row and n the number of
rows. Here d = 6 mm, m = 3 and n = 5, so

Ap = 7.19 × 10−3 m2 .

Using equation (1.13) the drag coefficient for each rod in the control volume is
calculated using the upstream velocity
Fd
Cd = 1/2ρu2 A
1 p

= 1.240 (3 d.p.).
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 55

Figure 4.6: Cd vs Rep for a single cylinder and the region behind it. (Taken from
Acheson, 1990, p. 150)

The cylindrical Reynolds number, Rep , is also evaluated by equation (1.10) and
so

Rep1 = 842 (3 s.f.),

Rep2 = 855 (3 s.f.).

Looking at Fig. 4.6, which shows typical drag coefficients for cylindrical Reynolds
numbers, then for the above values of Rep1 and Rep2 , the calculated value of Cd =
1.240 appears to be quite accurate. The figure also shows the region behind the
cylinder; for these values of Rep this region has no separation and there will be vortex
shedding behind the cylinders. However, as the value of the cylindrical Reynolds
increases (& 1100) there is a transitional period and a wake appears behind the
cylinder. The graph also shows typical ranges of the drag coefficient that can
be expected. As the experiments here use arrays of multiple cylinders the drag
coefficients are expected to differ from the shape indicated by the graph, however,
the range given here should be expected.
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 56

Owing to the number of calculations needed in order to find this one drag coeffi-
cient it is unrealistic to work out the drag coefficients further down the array and for
all other tests. To overcome this a program was written in Fortran 90 which would
read in data files and perform the above calculations, the program is presented and
explained in Appendix B.
Running the program over the laboratory data and outputting the forces gives
the graph shown in Fig. 4.7. The graph’s visualisation is not what one would expect,
certainly with no extra energy being added to the system we would expect monotonic
functions of the different forces. Studying the measured laboratory depths across the
array (Fig. 4.8) we can see that despite there being a good number of evenly spaced
and carefully taken measurements the gradients of the individual sections do not
follow a smooth pattern. When measuring the surface of turbulent flows it is hard

0.3 Lab Rate of change of momentum


Lab Net pressure force
Lab Shear
Lab Drag
0.25

0.2
Force (N)

0.15

0.1

0.05

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x (m)

Figure 4.7: Forces acting in the channel for Test B3 (hup ≈ 8 cm)

to take accurate readings to tenths of a millimetre because the surface never stays
quite still as seen in Fig. 4.1. For this reason it is appropriate to fit functions to the
depth measurements to help flatten out the surface. As with in Section 3.3.1 the
Mathematica Fit function was used to fit a linear line to the results, a quadratic fit
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 57

0.09
Lab Data
Linear Fit
Quadratic Fit
0.08

0.07

0.06

0.05
z (m)

0.04

0.03

0.02

0.01

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x (m)

Figure 4.8: Surface profiles of Test B3 (hup ≈ 8 cm)

was also generated using the same function. However, looking at the surface profile
in Fig. 4.8, we see, as with the analysis of Series A, that the downstream section
falls off suddenly which is not helpful in studying the behaviour in the middle of the
array. For this reason the depth measurements after 1.6m were neglected when fitting
lines to the data. The depth functions generated by Mathematica are also shown in
Fig. 4.8 with their calculated values and errors presented in Table 4.1. Clearly the
quadratic fitting function gives less errors to the calculated values, however, Tanino
and Nepf (2008) suggest that the surface displacement will be linear and so as a result
the quadratic will be inaccurate. As a result both linear and quadratic functions are
used to calculate the drag coefficients and the values are compared in the next step.
Although the employed method of using all σ permutations to ensure no errors
from pattern design in the results it is possible that measurements were taken at
points where there were mid pattern fluctuations, for instance, if a rod was directly
downstream of one measurement point it would be expected that the depth would
have a slight increase in this close proximity. To investigate how this affected the
results we first note that the pattern used in Test B3 repeated every ten rows and the
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 58

Table 4.1: Functions fitted to laboratory data and their errors for Test B3 (hup ≈ 8
cm)
x (m) z (m)
Lab data Linear Fit Error Quadratic Fit Error
0.000 0.0805 0.08251 0.00201 0.08017 0.00033
0.101 0.0793 0.08063 0.00133 0.07918 0.00012
0.199 0.0786 0.07880 0.00020 0.07809 0.00051
0.297 0.0764 0.07697 0.00057 0.07690 0.00050
0.396 0.0751 0.07512 0.00002 0.07557 0.00047
0.496 0.0737 0.07325 0.00045 0.07412 0.00042
0.599 0.0722 0.07132 0.00088 0.07250 0.00030
0.700 0.0705 0.06944 0.00106 0.07079 0.00029
0.798 0.0688 0.06761 0.00119 0.06902 0.00022
0.899 0.0674 0.06572 0.00168 0.06707 0.00033
1.003 0.0659 0.06378 0.00212 0.06494 0.00096
1.098 0.0628 0.06200 0.00080 0.06289 0.00009
1.199 0.0614 0.06012 0.00128 0.06058 0.00082
1.298 0.0580 0.05827 0.00027 0.05821 0.00021
1.397 0.0559 0.05642 0.00052 0.05572 0.00018
1.499 0.0523 0.05451 0.00221 0.05304 0.00074
1.598 0.0503 0.05266 0.00236 0.05032 0.00002
1.698 0.0438 0.05079 0.00699 0.04745 0.00365
1.803 0.0379 0.04883 0.01093 0.04432 0.00642
1.933 0.0384 0.04640 0.00800 0.04026 0.00186

measurements were taken every five rows. By using alternate rows of the laboratory
depths from Table 4.1 the possibility of these errors was reduced. Defining Lab A
as using all of the rows of data and Lab B as using the first row and alternate rows
thereafter this can be investigated further. The alternate rows method also gives
larger control volumes and as a result any slight errors would be less obvious.
There were now four sets of data available for calculating the drag coefficient; Lab
A, Lab B, Linear fit and Quadratic fit. Fig. 4.9 shows the drag coefficients calculated
from these data sets, plotted against the x-direction of the channel. It should also
be pointed out that the laboratory data’s x-coordinates are not all equal distances
apart. This was because the x scale on the flume in the laboratory was only marked
to tenths of an inch; despite this every care was taken in spacing the measurements
and it can be seen from the values in Table 4.1 that the measurements were taken to
within one decimal place of what should be every 10 cm. When plotting the linear
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 59

and quadratic functions the functions were evaluated with x increasing by 0.1 m each
time meaning all control volumes were of equal length. The spiking of the Cd using

3
Lab A
Lab B
Linear Fit
Quadratic Fit
2.5

2
Cd

1.5

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x (m)

Figure 4.9: Cd vs x (m) - Calculated from lab data and fitted functions of Test B3
(hup ≈ 8 cm)

the laboratory data is to be expected from looking at the forces in Fig. 4.7 and it
was hoped that the approach for Lab Data B might produce less violent looking
results. Indeed, the peaks and troughs of the spikes in this drag coefficient plot is less
extreme than those of Lab Data A’s. The calming of the peaks and troughs shows a
general shape that the drag coefficient should follow. Along the middle of the channel
all four methods give similar looking drag coefficients. A problem with calculating
the drag coefficient that arises with these methods is that it is averaged with the
previous row of data. Hence the drag coefficient is calculated for the second depth
measurement onwards. This meant that for Lab B the first drag coefficient was at
x=0.2 m rather than x=0.1 m, leaving less upstream data for comparison with the
linear and quadratic fitted functions.
Looking at Fig. 4.10, which shows the drag coefficient plotted against the cylin-
drical Reynolds number, the drag coefficient shown by Lab Data B appears to match
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 60

the results of the quadratic fit closer than that of the linear fit. Both lab data sets
start out with values of Cd < 1, this is mirrored by the quadratic fit whilst the lin-
ear function is around the value of two; disagreeing with the value range indicated in
Fig. 4.6. Lab Data A does have some high values which the linear fit appears to agree
more with than the quadratic function, however, Lab Data B’s objective was to calm
down these peaks (and troughs) and in doing so certainly brought the higher spikes
further down than the lower ones up. In respect of the arguments discussed here

2.5
Lab A
Lab B
Linear Fit
Quadratic Fit

1.5
Cd

0.5

0
800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800
Re p

Figure 4.10: Cd vs Rep - Calculated from lab data and fitted functions of Test B3
(hup ≈ 8 cm)

it seems that the range of values for the drag coefficient presented by the quadratic
fit are more realistic than those of the linear function. Furthermore the quadratic
function displays a drag coefficient plot which matches the rough shape and closer
values of the lab data.
The quadratic functions surface profile’s gradient steepens along the flow, this
will lead to an increase in net pressure force as the depth difference over the control
volume will also be increasing, as a result there will be an increase in the drag force.
Fig. 4.11 shows the velocity calculated from the laboratory data as well as generated
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 61

from the linear and quadratic fits. For all of these cases it can be seen that the
velocity increases gradually along the flow with the rate of change of this increasing
as we move further down the flow. As a result u2 , will increase dramatically. The

0.3
Lab Data
Linear Fit
Quadratic Fit

0.25

0.2
u (m/s)

0.15

0.1

0.05

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x (m)

Figure 4.11: u vs x (m) - Calculated from lab data and fitted functions of Test B3
(hup ≈ 8 cm)

drag coefficient (eqn (1.13)), being a balance between Fd and u2 , could be expected
to change direction as a result of the drag force starting out dominant force before
being overtaken by u2 further down the flow.
As a result of the arguments above it is clear that the quadratic is better than the
linear function in giving drag coefficient values which are not only a more accurate
representation of the data recorded in the laboratory but also more realistic in terms
of the range. For these reasons the quadratic functions methodology of calculating
drag coefficients was used for other tests; these results are presented in the next
chapter.
CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 62

4.4 Normal Flow

As Series C used (approximately) constant flow depths with the channel at an angle,
α, there was no net pressure force or rate of change of momentum. The normal flow
depths aimed to be 3.17 and 5.17 cm, flow interaction with rods meant that a flat
surface could not be achieved within the array but every effort was taken to ensure
up- and downstream values as close to the required depths as possible; in practise
these were at a maximum 2 mm from the desired depth which, in relation to the
effect of the slope, was acceptable. Of course this did mean that the above forces
could be present but the approximation was sufficient to neglect them for analysing
the natural flow phenomenon.
One specific test, Test C4 (with hup = 5.17 cm and α = 1 ◦ ), had a resulting
flow rate of Q = 1.062 × 10−3 m3 s−1 to satisfy normal flow conditions and so using
equation (1.4)
u = 0.198 m s−1 (3 d.p.).

The volume is the area of boards multiplied by the depth of flow which is

V = 9.67 × 10−3 m3 .

Using (1.21) the force from the fluid acting down the channel slope is

Fm = mg sin α

= 1.655 N (3 d.p.).

As with the previous section the calculation of shear stress requires the Reynolds
number and so the wetted perimeter (eqn. (1.1)) and hydraulic diameter (eqn. (1.2))
are required,

PW = 0.2074 m

DH = 0.1037 m (4 d.p.).

Using (1.9) the Reynolds number is now found as

Rec = 20486 (5 s.f.).


CHAPTER 4. ANALYSING RESULTS AND INTERPRETING THE DATA 63

The Darcy-Weisbach friction factor is found as in the previous section by iterating



over the initial guess of 1/ f = 1 and gives

f = 0.1266 (4 d.p.),

and then eqn (1.18) can be used with this and the velocity found above to give

τ = 0.6176 N m−2 (4 d.p.).

The shear acts on the boundary of the volume so

Shear Stress Force = τ PW L

= 0.231 N (3 d.p.)

where L is the length of the array, here 1.8 m.

4.5 Discussion

The results observed from the tests of Series A justified the use of φ as a control
parameter for further tests, these initial tests also gave an indication of problems
that may arise such as the anomaly in Fig. 4.4 which allowed the opportunity to
improve experimental techniques before pursuing the test stage further.
For analysing the results of Series B, several methods were brought forward to
calculate the drag coefficients. Of the arguments for and against it is clear that using
a quadratic fit of the surface displacement is the best representation. The results
then calculated from the quadratic function then give a range of drag coefficient
values that appears realistic and with solid reasoning. The linear function overshoots
the expected upstream drag coefficient by a substantial inaccurate margin and the
surface profile that it represents does not agree with physical observations of those
made against Fig. 4.1.
Proceeding with the quadratic fit function for all tests the following chapter
presents the results calculated using the above method and provides significant anal-
ysis in the resistance of flow.
Chapter 5

Results

5.1 Overview

The previous chapter showed how to calculate the drag coefficient accurately from
the laboratory data using the theory at the start of this project, the methodology
was applied to the other experiments and the results presented here. This chapter
first discusses how the flow rate is affected by solid volume fractions. The drag
coefficient within the horizontal arrays of Series B are then investigated using both
the solid volume fraction and upstream depth as control parameters. These results
are compared with the horizontal patterned tests (D1, D2, D3) to investigate the
particular arrangement as a parameter. The results of Series C, which concluded of
tests in normal flow conditions, are then discussed. These tests investigate how solid
volume fraction, angle of slope, and normal flow depth, effect the resistance. Once
more the results are compared with those using the patterned arrangement (Tests D4
- D9). Series E comprised of tests with flexible rods in a horizontal channel and these
results are then compared with those form Series B to establish how the flexibility
effects the resistance to flow.
Five upstream depths were investigated for horizontal channel flow, hup1 – hup5 ,
as well as two approximately constant depths for the normal flow cases, hN1 and
hN2 . The values for these are shown in Table 5.1. The results in this chapter are
presented graphically and selected in a manner that allows access to the results as

64
CHAPTER 5. RESULTS 65

required, the results calculated from the laboratory values using the Fortran programs
(Appendix B) are tabulated in Appendix C.

Table 5.1: Abbreviations of controlled depths used in results


Abbreviation Depth (cm)
hup1 6
hup2 8.5
hup3 10.4
hup4 13
hup5 16.5
hN1 3.17
hN2 5.17

5.2 Solid Volume Fraction and Flow Rate

In order to achieve the same upstream depths for different solid volume fractions the
flow rate was altered accordingly, Fig. 5.1 shows the results of these tests for three
upstream values, hup1 , hup2 , and hup3 . We see from the graph that a larger flow

0.0024
phi = 0.014
phi = 0.027
phi = 0.041
phi = 0.054
0.002

0.0016
Q (m^3 / s)

0.0012

0.0008

0.0004

0
h1 h2 h3
Upstream depth

Figure 5.1: Q vs φ for various hup in Series B

rate leads to a deeper upstream depth for all values of φ, and that larger values of φ
CHAPTER 5. RESULTS 66

require less flow to achieve similar flow depths of a less dense array. By comparing
these results side to side it is clear that a deeper upstream depth gives a larger range
of Q from one solid volume fraction to the next. Recalling that the hup values are
approximate, the actual achieved depths are shown in Table 5.2. The value of hup1

Table 5.2: Actual upstream flow depths for tests in horizontal channel
Test No. φ Upstream depths (cm)
hup1 hup2 hup3 hup4 hup4
Test B1 0.014 5.72 8.48 9.37
Test B2 0.027 5.95 8.09 9.72
Test B3 0.041 5.54 8.05 9.94
Test B4 0.054 5.99 8.34 10.27
Test D1 0.027 5.68 7.42 10.06
Test D2 0.041 5.64 7.81 10.22
Test D3 0.034 5.54 7.48 9.68
Test E1 0.009 8.47 10.40 13.04 16.87
Test E2 0.014 8.60 10.30 13.50 16.81
Test E3 0.019 8.30 10.68 13.30 16.79
Test E4 0.024 8.16 10.92 12.81 16.62

for Test B3 is smaller than the other surrounding values which gives an indication
of why the value of Q appears closer to the value for Test B4. Also, Test B1’s hup3
value is smaller than similar values for the other tests, if it were larger then we would
expect a larger value of Q and the range would be much more exaggerated.
Tests B2 and D1 had equal solid volume fractions but with D1’s rods being of
design P2 giving two lines of rods, Tests B3 and D2 also shared this similarity but with
three rods per row; Fig. 5.2 shows a comparison of the flow rates for various upstream
depths (as with Fig. 5.1) but this time using these two solid volume fractions and
their different arrangements. For the lower upstream values (hup1 ) the difference
in Q between the straight and scattered arrays is small; this increases for hup2 and
then even more so for hup3 . The arrangements of D1 and D2 ask for a larger flow
rate to match the upstream depths of their counterparts, again, looking as the actual
values of the upstream depths (Table 5.1) we see that they differ slightly from the
other hup values. However, it is the hup1 and hup2 values which have the greatest
difference; in these cases both Series D tests have lower upstream depths and so the
large flow rate seen for the hup3 depth is the most accurate explanation of what is
CHAPTER 5. RESULTS 67

0.0028
B2 phi = 0.027
D1 phi = 0.027
B3 phi = 0.041
D2 phi = 0.041
0.0024

0.002

0.0016
Q (m^3 / s)

0.0012

0.0008

0.0004

0
h1 h2 h3
Upstream depth

Figure 5.2: Comparison of similar φ’s for patterned and un-patterned arrangements

happening.
From the above results we can conclude that denser arrays lead to greater flow
depths for similar flow rates, which indicates these arrays have more resistance to the
flow in the channel (understandable as there are more rods to give such resistance).
It is also seen that when rods are placed behind each other in arrangement patterns
P2 and P3 then a greater flow rate is required to achieve a flow depth, indicating
this style of array has lower resistance. The rods being placed in lines mean that
parts of the channel have a clear slip stream for the flow to move through. From this
we understand that the flow must be turbulent as to create a wake region behind
the rods, as in Fig. 4.6 when placed in lines, whilst in the arrays of Series B there
is vortex shedding, partially across the y-direction of the channel as they avoid the
rods.
CHAPTER 5. RESULTS 68

h1
1.4 h2
h3

1.2

0.8
Cd

0.6

0.4

0.2

0
15000 20000 25000 30000 35000 40000
Re

Figure 5.3: Cd vs Rec for Test B2 with different upstream depths

5.3 Resistance by Solid Volume Fraction

Considering flow in a channel for a fixed upstream depth the above results indicated
that the denser arrays were more resistant and would therefore have more drag. The
drag coefficient is plotted against the x-coordinate of the channel for the different
upstream values for all four tests of Series B with upstream depth hup3 in Fig. 5.4.
Here the drag coefficient has quite a significant range at the start of the array for the
various solid volume fractions. For φ = 0.014 the upstream drag coefficient is Cd =
0.495 which is the lower end of the expected scale whilst the densest array with φ =
0.054 is almost double this with Cd = 0.966. Moving through the four investigated
solid volume fractions (which are all equal distances apart) from φ = 0.014 − 0.054
the upstream drag coefficient increases, although by a lesser amount each time. This
indicates that whilst denser arrays do increase upstream drag the relationship is
not linear and one would expect there to be a limiting value where increasing the
array density no longer causes significant changes to the drag coefficient. The drag
coefficient also has a maximum value mid way along the channel, for φ = 0.054 this
is at x = 0.9 m where Cd = 1.51. For φ = 0.027 and 0.041 the maximum is slightly
CHAPTER 5. RESULTS 69

2
phi = 0.014
phi = 0.027
phi = 0.041
phi = 0.054

1.5

Cd

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
x (m)

Figure 5.4: Cd vs x for Series B tests and upstream depth hup3

less (Cd = 1.40 and 1.38 respectively) but also at x = 0.9 m, whilst the lowest solid
volume fraction of φ = 0.014 experiences a maximum that is bounded between the
above values with Cd = 0.142 and further downstream at x = 1.0 m. The exact
location of the maximum is not as relevant as the fact that it is approximately in the
middle of the array; it is here that the inertial term of the fluid starts to dominate the
pressure (and hence drag) force. Fig. 5.5 shows the results of Series B still with the
same upstream depth of hup3 but now plots the drag coefficient against the Reynolds
number. The Reynolds number increases with the x direction along the channel and
so each plot can be read from left to right in terms of flow direction. However, Fig. 5.4
indicates that the maximum of Cd is at the midpoint of the array whilst in this new
graph they are toward the starting values of the range of Reynolds numbers. Also,
the lower solid volume fraction gives a greater range of larger Rec values. The two
sections before and after the maximum point of the Cd vs Rec curve represent the
switch of dominance of the drag and inertial terms that is mentioned above. So the
maximum gives the Reynolds number required to tip the dominance and Fig. 5.5 tells
us a lower value of Rec is required for this to occur in the larger solid volume fractions.
CHAPTER 5. RESULTS 70

2
phi = 0.014
phi = 0.027
phi = 0.041
phi = 0.054

1.5

Cd

0.5

0
15000 20000 25000 30000 35000 40000 45000
x (m)

Figure 5.5: Cd vs Rec for Series B tests and upstream depth hup3

The manner in which the Reynolds number approaches the maximum suggests that
in the front half of the array there is little change of inertial terms, but the change
significantly increases the drag coefficient. The second part of the curve demonstrates
the flow velocity still increasing but now that it is dominant it has a negative effect
on the drag coefficient. Comparing the second section for all plots tells us that flow
through denser arrays experience a greater velocity1 change over the same distance,
although the actual values are much smaller than those for the lower φ’s.
Once more looking at equal solid volume fractions of Series B against the patterns
P2 and P3 from Series D the drag coefficients can be compared with the Reynolds
number as in Fig. 5.6. In this graph the B2 and B3 plots are those from Fig. 5.5, the
D1 and D2 plots are of equal φ respectively and are the repeated pattern test arrays
P2 and P3 respectively. The new plots indicate that overall the drag coefficients are
much smaller when the rods are placed into lines, their values of Cd for the Series
B tests were described above and particularly noted to be quite similar, however, for
1
The kinematic viscosity is taken as constant and the DH will be approximately constant down-
stream for all of the tests as the downstream depth was controlled so velocity can be discussed here.
Actual velocity values are seen in the results in Appendix C
CHAPTER 5. RESULTS 71

the Series D tests the peaks are not a similar height; D1’s maximum is Cd = 0.637
whilst D2’s is Cd = 0.894. In an array with the rods using all σ permutations, such

B2 phi = 0.027
1.6 D1 phi = 0.027
B3 phi = 0.041
D2 phi = 0.041
D3 phi = 0.034
1.4

1.2

1
Cd

0.8

0.6

0.4

0.2

0
20000 25000 30000 35000 40000 45000 50000
Re

Figure 5.6: Comparison of Cd vs Rec for equal φ but repeated rows.

as those used in Series B, with equal upstream depths, the drag coefficients cover
similar ranges. Fig. 5.6, however, now shows us that when the chosen arrangement is
specifically designed to let flow past, the solid volume fraction has a substantial effect
on the drag coefficient. The values of the Reynolds numbers at the start of the Series
D plots are about the same as the downstream values for their Series B counterparts.
Fig. 5.6 also shows the plot of Test D3, for which φ = 0.034, and uses alternating
rows of Tests D1 and D2 (pattern P2.5). Immediately it is visible that the pattern
design P2.5 is dissimilar to the other pattern tests of Series D and is more suitable
for comparison with the tests of Series B. The alternate design of three and two
rods per row leads to a solid volume fraction of φ = 0.034 which means Tests B2
and B3 lie at equal distances either side of Test D3. This plot is dissimilar to the
other obtained plots and whilst the size of the drag coefficient’s range is suitable
the actual maximum values of this range are certainly higher than is predicted by
Fig. 4.6. Fig. 5.7 examines the behaviour of the drag coefficient against the Reynolds
CHAPTER 5. RESULTS 72

number for these three tests and three upstream depths. Test D3 is represented by

2.5
B2 h1 phi = 0.027
B2 h2 phi = 0.027
B2 h3 phi = 0.027
D3 h1 phi = 0.034
D3 h2 phi = 0.034
2 D3 h3 phi = 0.034
B3 h1 phi = 0.041
B3 h2 phi = 0.041
B3 h3 phi = 0.041

1.5
Cd

0.5

0
5000 10000 15000 20000 25000 30000 35000 40000
Re

Figure 5.7: Cd vs Rec for Tests B2, B3 and D3

the green plots whilst B2 and B3 are by red and blue respectively; for all tests hup1
is the furthest left and then hup2 in the middle with hup3 on the right. The Series
B plots mirror the discussion from above and it would be expected that a test with
solid volume fraction between these twos’ would lie intermittently between the results,
however, this is not the case. The results of Test D3 with hup3 , as discussed above,
yield similar Reynolds values (although not between the other tests’ results) and the
drag coefficient is a little (though not alarmingly) higher. As a direct comparison it
can be said that the curve for Test D3 is very similar to that of Test B3’s but with
an increase in drag. Moving down the scale, the results for hup2 show that Test D3
has now moved down the range of Reynolds numbers, not only this, but the drag
coefficients are again larger than the other tests, even more so than for hup3 . The
up and downstream drag coefficients for Test D3 with hup3 were Cd = 1.199 and
0.792 respectively with a range of 0.407; for hup2 they are Cd = 1.398 and 0.994 which
has approximately the same range. However, moving to the shallowest flow of hup1
there is no-mid curve maximum as the plot starts at Cd = 2.477 and then moves
CHAPTER 5. RESULTS 73

monotonically downstream to Cd = 1.625. This range does seem unrealistic and


the function itself does not conform to the shape that the other test results adhere
to. However, looking at the results for the other two upstream depths this result
does appear to fit with the trend visible in Fig. 5.7. It is also quite possible that
the quadratic graph that was applied to the data was not a good representation,
although the results that are shown in Table C.21 in Appendix C appear correct.
Nonetheless, from the hup2 and hup3 plots in Fig. 5.7 it is clear that the arrangement
P3 is responsible for producing higher drag coefficients than those from non-particular
arrangements with both higher and lower solid volume fractions.

5.4 Normal Flow

The tests for normal flow were those from Series C as well as Tests D4 - D9. These
tests investigated angles of one and two degrees and each of these angles looked at
two different flow depths, hN1 and hN2 . Constant flow depth and velocity meant for
each test the rods experienced the same amount of drag and so one drag coefficient
was produced for each test, these are shown in Fig. 5.8 for the tests of Series C. The
effect that the angle has on the drag coefficient can be seen by comparing the red and
green columns for normal depth hN1 and blue and pink for hN2 . These comparisons
tell us that an increase in angle increases the drag coefficient for all solid volume
fractions that were investigated. An increase in angle means that the fluid mass
is acting more so in the flow direction, although one degree is quite a conservative
change for φ = 0.014, 0.027, 0.041 where the drag coefficient has increased by a factor
of two. This is not the case for φ = 0054 although the increase from Cd = 0.588 to
Cd = 0.861 is still quite significant. Similarly, for depth hN2 , the change in angle
increases the drag coefficient by a quite noticeable difference.
Two normal flow depths allow for the inspection of drag as these are changed
and the slope is kept constant; comparing the red with blue, or green with pink,
columns allows for this comparison for α = 1 ◦ and 2 ◦ respectively. An increase in
the flow depth leads to an increase in the drag for both angles except in the specific
CHAPTER 5. RESULTS 74

1.6
h1 alpha = 1
h1 alpha = 2
h2 alpha = 1
1.4 h2 alpha = 2

1.2

Cd 1

0.8

0.6

0.4

0.2

0
0.014 0.027 0.041 0.054
phi

Figure 5.8: Drag coefficient for tests of Series C

case of φ = 0.041 and a slope of 2 ◦ . In this case the drag coefficient is Cd = 1.597,
which is significantly bigger than those for the other solid volume fractions under
similar conditions, although still in the expected range of values; to ensure that the
misfitting value was not a miscalculation by the program the values were checked by
hand. In fact, whilst this does look like an anomaly in the trend, looking closer at the
results in view of comparing against the solid volume fractions gives quite unexpected
results. For hN1 and α = 1 ◦ the drag coefficient is small and increases from 0.248
at φ = 0.014 to 0.345 at φ = 0.027, there is then a small decrease to Cd = 0.331 at
φ = 0.041 before increasing again to 0.588 at φ = 0.054. With hN1 and α = 2 ◦ the
drag coefficient again increases with φ but this time does so up to φ = 0.041 before
decreasing for the final measured solid volume fraction. This second trend is again
repeated for hN2 and α = 1 ◦ , but for hN2 and α = 2 ◦ the drag coefficient increases
and decreases before increasing again and finally decreasing as we move through the
four solid volume fractions.
These results in terms of comparing how the solid volume fraction affects normal
flow do not follow similar trends to what was seen in the analysis of Series B. The
CHAPTER 5. RESULTS 75

flow rates in Table C.22 are all quite small but to create normal flow they do increase
with the solid volume fraction as was seen in Series B. Being small could lead to
accuracy errors as was seen earlier, however, it is improbable that so many errors
were observed in so few tests. One reason the increase in solid volume fraction alone
is not sufficient to give an increase in the drag coefficient, which might be expected,
is that the solid volume fraction is providing more resistance but also a larger surface
for the drag to act from. Another factor is the volume, V , which is used to calculate
the fluid mass and assumed equal for all tests,

V = Lbh,

when in fact the volume of water does not include the volume occupied by the rods
and so the volume can more accurately be described as

πd2
 
V = h Lb − mn ,
4

where m is the number of rods per row in n rows as usual. Using values for a control
volume covering all six boards, L = 1.8m, n = 90, d = 0.006m and m controls the
value of φ the actual volumes used were V = 5.93 × 10−3 m3 and 9.67 × 10−3 m3 for
hN1 and hN2 respectively. Table 5.3 shows the error of the volume estimation in cubic
metres that is encountered by the increasing number of rods per row and hence solid
volume fraction (see Table 3.1). The errors are around 2 - 10% of the volumes that
are given above, increasing with m, and so could play quite a significant role in the
results.

Table 5.3: Error in volume for different rods per row (m)
m Error (m3 )
1 0.000132
2 0.000263
3 0.000395
4 0.000526

The drag coefficients can also be plotted against the Reynolds number that
achieves normal flow for the different tests. Fig. 5.9 shows the drag coefficient against
the Reynolds number with a channel slope of α = 1 ◦ and normal flow depth hN2 for
CHAPTER 5. RESULTS 76

the solid volume fractions used in Series B, the result of Test D7 (using pattern P2.5)
is also shown. Here we see that the densest array has a much larger drag coefficient

0.9
phi = 0.014
phi = 0.027
phi = 0.034
0.8 phi = 0.041
phi = 0.054

0.7

0.6

0.5
Cd

0.4

0.3

0.2

0.1

0
14000 16000 18000 20000 22000 24000 26000 28000 30000
Re

Figure 5.9: Cd vs Rec for α = 1 ◦ and hN1

and a lower Reynolds number to achieve this. The two central φ values of Series
B give similar drag coefficients as discussed at the start of this section and from the
plot we see that their Reynolds numbers are close (approximately 2000 apart). The
drag coefficient for φ = 0.014 is the smallest value observed and occurs at the highest
Reynolds number of 28434 which is almost twice that of the densest array (φ = 0.054
had Rec = 15626). Looking at the patterned array of Test D7 we once again see
results similar to those observed in analysing Series B, the drag coefficient is much
larger and this is achieved from a low Reynolds number.
As with the horizontal channel experiments, tests were carried out using equal
solid volume fractions but different repeated patterns of P2 and P3. Fig. 5.10 shows
the drag coefficient against the Reynolds number for φ = 0.027 and the four com-
binations of varying slope and flow depth; these tests are repeated for both the σ
permutations and patterned methods with similar parameters being shown in the
same colours. The drag coefficient is reduced significantly in all cases of using the
CHAPTER 5. RESULTS 77

1.6
alpha = 1 h1 C2
alpha = 1 h1 D4
alpha = 1 h2 C2
1.4 alpha = 1 h2 D4
alpha = 2 h1 C6
alpha = 2 h1 D7
alpha = 2 h2 C6
1.2 alpha = 2 h2 D7

Cd 1

0.8

0.6

0.4

0.2

0
22000 24000 26000 28000 30000 32000 34000 36000 38000 40000 42000
Re

Figure 5.10: Cd vs Rec for normal flow (φ = 0.027)

patterned arrangement, the Reynolds number also increases showing once more that
the repeated rows streamline the flow and create slip streams either side of the rods.
Similarly, Fig. 5.11 shows the same conditions as above but this time with a
solid volume fraction φ = 0.041 and hence pattern P3. The graph shows the same
observation as for its φ = 0.027 counterpart but shows that the Reynolds numbers
experienced are smaller when increasing the solid volume fraction. Certainly for the
patterned tests the Reynolds number increases by roughly 10,000 for all cases whilst
the arrangements of Series C vary in the range of 2,000. There is an anomaly to the
latter of these ranges for α = 2 ◦ and depth hN1 , which is Test C7, as this is the
same anomaly encountered earlier it seems increasingly likely that there was an error
when transcribing this particular set of laboratory data.

5.5 Flexible Rods

Series E looked at flow past flexible rods in a horizontal channel with four different
upstream depths, hup2 – hup5 (see Table 5.2). For Test E1 the solid volume fraction
CHAPTER 5. RESULTS 78

1.6
alpha = 1 h1 C3
alpha = 1 h1 D5
alpha = 1 h2 C3
1.4 alpha = 1 h2 D5
alpha = 2 h1 C7
alpha = 2 h1 D8
alpha = 2 h2 C7
1.2 alpha = 2 h2 D8

Cd 1

0.8

0.6

0.4

0.2

0
16000 18000 20000 22000 24000 26000 28000 30000 32000
Re

Figure 5.11: Cd vs Rec for normal flow (φ = 0.041)

was φ = 0.009 and the arrangement of rods the same as that of Test B1’s. Similarly
to previous experiments, increasing the flow rate led to an increase in the flow depth,
as well as measuring this points along the the array as previously at the behaviour of
the rods were also important. Fig. 5.12 shows photographs of Test E1 with upstream
flow depths hup3 and hup5 , these photos show the rods bending with the flow, as
might be expected, the bending increases noticeably with hup which indicates the
rods bending is effected by the flow rate.

(a) hup3 (b) hup5

Figure 5.12: Laboratory observations at the front of the array of Test E1 with different
upstream depths

Plots of the drag coefficient against the x-coordinate for Test E1 and the four
CHAPTER 5. RESULTS 79

upstream depths are shown in Fig. 5.13. Here we can see that much like the tests of
Series B the drag coefficient has a maximum within the array, however, for Series B
this maximum was located around the middle of the array, with this graph we see
the maximum is further downstream of the array’s midpoint. This tells us that

1
h2
h3
h4
h5

0.8

0.6
Cd

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
x (m)

Figure 5.13: Cd vs x (m) for Test E1

whilst the inertial term in eqn (1.13) does dominate the drag force the flexibility of
the rods appears to prolong this event. The drag coefficient is also plotted against
the Reynolds number for the above case as can be seen in Fig. 5.14. The maximum
drag coefficient is larger for deeper upstream values, as is the range and values of
Reynolds numbers as was found for rigid rods. However, the actual values observed
in these results is that Cd < 1 which differs from results from the results of Series
B. As well as the rods having some bend in them with the increase of flow rate there
was also very noticeable vibrations in the rods which varied at different points down
the array. For Test E1 and upstream depth hup2 there were slight vibrations in the
rods for the front few rows; the vibrations travelled in the y-direction of the channel
and so were perpendicular to the direction of flow. Increasing the flow depth led to
further vibrations; for hup3 there were bends in the front six rows of rods as seen
CHAPTER 5. RESULTS 80

1
h2
h3
h4
h5

0.8

0.6
Cd

0.4

0.2

0
30000 40000 50000 60000 70000 80000 90000
Re

Figure 5.14: Cd vs Rec for Test E1

in Fig. 5.12(a) whilst the next rows of rods vibrated heavily2 up until the middle of
the array when the rods started to vibrate less until the last few rows of rods in the
array showed no signs of movement.
As was previously discussed the diameter of the flexible rods were 5mm whilst the
rigid rods were 6mm, Test E1 used the same rod arrangement as Test B1 but because
of this difference in the diameters these tests had solid volume fractions φ = 0.009 and
φ = 0.014 respectively, placing extra rods into the arrangement as guided by Table 3.1
meant that a second test (E2) could be carried out with flexible rods and φ = 0.014.
Fig. 5.15 shows the drag coefficient against the Reynolds number for these three tests
and their similar upstream depths. From the graph it is clear that the tests with
flexible rods show that the slight increase in solid volume fraction leads to a slight
increase in drag coefficient as well as a decrease in the Reynolds number to achieve
this; as was concluded for rigid rods previously, however, in comparison to Test B1,
the drag coefficients and Reynolds numbers for Test E2 are dissimilar giving upstream
Reynolds numbers within 5000 of the rigid compatriots whilst E1’s are within 10,000;
2
As there was no equipment available to measure the vibrations they are described in relation to
the other vibrations; here heavy vibrations implies faster oscillations.
CHAPTER 5. RESULTS 81

h2 E1 phi = 0.009
1.4 h2 E2 phi = 0.014
h2 B1 phi = 0.014
h3 E1 phi = 0.009
h3 E2 phi = 0.014
1.2 h3 B1 phi = 0.014

0.8
Cd

0.6

0.4

0.2

0
25000 30000 35000 40000 45000 50000 55000 60000 65000
Re

Figure 5.15: Cd vs Rec for Tests E1, E2 and B1 with upstream depths hup2 and hup3

in both cases the shallower upstream depth gives closer results. More noticeable is the
difference in drag coefficient which has dramatically decreased by the introduction of
flexible rods.
Test E3 examined an array of pattern P2 with the objective to examine the
streamlining effect that was present in the results of Test D1 and so no extra rods
were used to equal the solid volume fraction of these two tests. Recalling the results of
Tests E1 and E2 in Fig. 5.15 (which had a solid volume fraction difference of 0.005),
as the drag coefficients are similar for both tests, although the Reynolds numbers
do change, the difference in φ appears negligible when considering the dramatic
difference caused by the flexibility of the new test subject. Test E3 had a solid
volume fraction of 0.019 and D1’s 0.027 meaning a difference of 0.008, so this could
be neglected and rather the comparison would be for flexible vs rigid rods. The above
analysis indicates this slight difference in φ is negligible when comparing the flexible
and rigid rods. Fig. 5.16 shows a comparison of these two tests, plotting their drag
coefficient against the Reynolds number for upstream depths hup2 and hup3 . In the
first instance when D1 was compared against B2 it was seen that the drag coefficient
CHAPTER 5. RESULTS 82

0.7
h2 E3 phi = 0.019
h2 D1 phi = 0.027
h3 E3 phi = 0.019
h3 D1 phi = 0.027
0.6

0.5

0.4
Cd

0.3

0.2

0.1

0
25000 30000 35000 40000 45000 50000 55000 60000 65000
Re

Figure 5.16: Cd vs Rec for Tests E3 and D1 (pattern P2)

decreased substantially for an equal solid volume fraction when pattern P2 down
the entire channel, now examining the results in Fig. 5.16 the drag coefficient has
again been reduced. The Reynolds number has also increased as a result by around
15,000 for hup2 and as much as 20,000 with hup3 . In both cases examined here the
downstream drag coefficient is very small (Cd = 0.062 and 0.065 for hup2 and hup3
respectively) implying that the inertial term has incredible dominance over the rods’
drag.
Previous analysis in this chapter indicated that arrays using pattern P2.5 increase
the drag coefficient significantly, more so than increasing the solid volume fraction.
Flexible rods were tested with this design in Test E4, the plot of the drag coefficient
against the Reynolds number is shown for the four considered upstream depths in
Fig. 5.17. Once more the drag coefficient is seen to increase with the increasing
value of hup and in comparison to the other tests the values in the Reynolds number
range has decreased as can be seen by comparing this with Fig. 5.14. The comparison
of two tests using pattern P2.5 is seen in Fig. 5.18 which shows the drag coefficients
against the Reynolds numbers for Tests E4 and D3 which have solid volume fractions
CHAPTER 5. RESULTS 83

1.2
h2
h3
h4
h5
1

0.8
Cd

0.6

0.4

0.2

0
25000 30000 35000 40000 45000 50000 55000 60000 65000
Re

Figure 5.17: Cd vs Rec for Test E4 (φ = 0.024)

φ = 0.024 and 0.034 respectively. The flexibility of the rods here appears to be

1.8
E4 phi = 0.024
D3 phi = 0.034
E4 phi = 0.024
1.6 D3 phi = 0.034

1.4

1.2

1
Cd

0.8

0.6

0.4

0.2

0
15000 20000 25000 30000 35000 40000 45000
Re

Figure 5.18: Cd vs Rec for Tests E4 and D3

counteracting pattern 2.5’s usual resilience to the flow and has decreased the drag
and increased the Reynolds number indicating more dominance from the inertial
CHAPTER 5. RESULTS 84

term. When studying the results from Series B it was concluded that there was
vortex shedding in parts of the arrays, the frequency of this shedding increased with
Rep until a stable wake region had formed at a certain point of turbulence. On
a single rod these vortices occur alternately either side and with them an area of
low pressure is formed in their vicinity; when such vortices are formed in the case
of flexible rods it appears that the rods are dragged into the alternate areas of low
pressure and gives the vibrating effect. This phenomenon, known as vortex induced
vibration (VIV), can be analysed using the non-dimensional Strouhal number

γL
St = , (5.1)
U

where γ is the frequency of vortex shedding, and U and L the usual characteristic
velocity and length scales respectively; here it is suitable to use the rod diameter as
the length scale and the fluid velocity as the characteristic velocity. The Strouhal
number can be related to the cylindrical Reynolds number by the formula
 
19.7
St = 0.198 1 − , (5.2)
Rep

when 300 < Rep < 2×105 (Blevins, 1977), it is quite apparent that in this given range
of Rep the Strouhal number will be close to 0.2. As discussed earlier the frequency
of vibrations in the rods was unable to be recorded and so the Strouhal number can
be used as an important tool to find this. Gabbai and Benaroya (2005) state that
at St ≈ 0.2 the vortex shedding frequency matches the natural frequency of a rod
that is free to move, but as it strays from this critical value the frequencies do not
match and more frequent vibrations occur. Using equation (5.2) with the cylindrical
Reynolds number for tests in Series E can give values for St as seen in Table 5.4.
The frequency of the vibrations, γ, are then calculated using equation (5.1). The
laboratory observations for this particular test was that the first twenty five rows of
rods showed clear bending in the direction flow whilst the rest of the rods had strong
fluctuations in the y-direction of flow. The data in the table does not have the ability
to predict such bending but the frequencies have been predicted to increase further
downstream which agrees with the laboratory observations.
CHAPTER 5. RESULTS 85

Table 5.4: Strouhal number and predicted frequency of vibrations for Test E2 (hup4 )
x (m) u( m s−1 ) Rep St γ (s−1 )
0.0 0.3160 1579.9 0.19553 12.358
0.1 0.3193 1596.7 0.19556 12.488
0.2 0.3233 1616.4 0.19559 12.647
0.3 0.3278 1639.2 0.19562 12.825
0.4 0.3331 1665.3 0.19566 13.035
0.5 0.3390 1695.1 0.19570 13.268
0.6 0.3458 1728.9 0.19574 13.538
0.7 0.3534 1767.1 0.19579 13.839
0.8 0.3621 1810.3 0.19585 14.183
0.9 0.3718 1858.9 0.19590 14.567
1.0 0.3828 1913.8 0.19596 15.003
1.1 0.3952 1975.9 0.19603 15.494
1.2 0.4092 2046.2 0.19609 16.048
1.3 0.4252 2126.1 0.19617 16.682
1.4 0.4435 2217.3 0.19624 17.407
1.5 0.4644 2321.9 0.19632 18.234
1.6 0.4885 2442.6 0.19640 19.189
1.7 0.5166 2583.2 0.19649 20.301
1.8 0.5496 2748.2 0.19658 21.608

5.6 Comparison of results with other authors

The results in this chapter compared the drag coefficient against the open channel
Reynolds number, Rec . By direct comparison of these plots against plots of the
drag coefficient with the x-coordinate along the channel these Cd vs Rec graphs can
now be interpreted to represent the change in drag in terms of pressure and inertia
dominance. The results presented in Chapter 2 all use different parameters for such
analysis; Tanino and Nepf (2008) use the cylindrical Reynolds number, Rep , which
also increases with the flow and so could be used in the same manner to describe
flow. The advantage of such a method is that it is focussed wholly on the flow
interaction with the rods whilst the use of Rec understands the model is performed
in a channel. As described in Section 4.3 the ratio of the channel depth and width in
this project meant that it was important to consider the actual channel at all times
which signifies the hydraulic diameter as the length scale in the Reynolds number,
however, Tanino and Nepf’s study was performed in a channel approximately four
times the width (40 cm) and was more concerned with denser arrays, so favoured the
CHAPTER 5. RESULTS 86

cylindrical Reynolds number. This was also calculated for the results of this chapter,
as seen in Appendix C, and so the drag coefficient can be compared between both
studies as seen in Fig. 5.19. What is initially obvious from the comparison is the

Figure 5.19: Comparison of Cd vs Rep for Series B (h1 ) and results from Tanino and
Nepf (2008)

shape of the curve is not similar between the studies, certainly the drag coefficients
for Tanino and Nepf’s results are very large and approach the value of 10. These
values, however, are for solid volume fractions that were much larger than the arrays
considered in this project. Furthermore, the solid volume fraction of φ = 0.091, which
is the closest value to those used in this project, yields similar drag coefficients. The
curve shape is still dissimilar between the two studies but the Reynolds numbers are
of the same order with this those of this study’s being at the top end of the scale.
With the main difference being the shape of the curve it is possible that the analysis
of depth measurements differs between the two studies.
CHAPTER 5. RESULTS 87

5.7 Discussion

This chapter has presented the results of laboratory experiments that were performed
to investigate the resistance to flow that can be presented from vegetation. The results
have been presented graphically with actual values referred to where appropriate; the
results of all tests discussed here are presented in Appendix C. Some assumptions
were made in gathering the results; the depth profile was assumed to be of the form of
a smooth quadratic function and the channel’s roughness was assumed to be equal on
the walls and floor. When performing experiments, the holes in the boards that were
unoccupied by rods were left uncovered which could have created extra roughness;
this was also not taken into account. The final chapter presents a conclusion to the
results and describes further research that could be pursued from the results found
in this study.
Chapter 6

Conclusions

This research project has presented an insight into flow through emergent vegetation
and in particular has examined laboratory findings for flow past rigid rods in a channel
which was both horizontal and sloped, further experiments examined flexible rods in
a horizontal channel in an attempt to model the vegetation to a more realistic degree.
The solid volume fraction of the rods was used as a control parameter, as was using
specific arrangements; in all tests the flow depth was varied. This extensive set of
experiments gave a broad range of results which were analysed together leading to
some very interesting results.
The drag coefficient was found to be directly linked to the density of rods in an
array, and by placing the rods in lines rather than scattered arrays the resistance
decreased as the fluid’s inertia rocketed. Whilst these results were quite elementary,
the specific pattern of P2.5 led to quite an incredible increase in the drag which
outweighed the solid volume fraction’s significance; this was something that had not
been encountered in the literature study so presented new ideas on resistance in
arrays. Test D3 used pattern P2.5 and with upstream depth hup2 the upstream drag
coefficient is Cd = 1.398. This is greater than Test B2’s (Cd = 0.908) at the same
upstream depth as expected, however, Test B3 has a higher solid volume fraction
than Test D3 but Cd = 0.970 which is lower than D3’s.
The arrangement was also found to decrease the drag in the array if using one of
the patterned tests (P2 or P3). For instance, with upstream depth hup3 Tests B2 and

88
CHAPTER 6. CONCLUSIONS 89

D1 both had equal solid volume fraction with D1 using pattern P2 but the upstream
drag coefficient was Cd = 0.7853 for B2 but Cd = 0.608 for D1. Furthermore the
maximum drag coefficient was found to be at the point where the inertial terms took
dominance from the net pressure force; in the Series B tests this maximum was mid
array but for patterned arrangements was further upstream. In contrast the use of
flexible rods moved the maximum drag coefficient further downstream from the mid
point of the array for all types of arrangements.
The introduction of flexible rods led to a decrease in the overall drag coefficient,
Test E2 and B1 had equal solid volume fraction yet with upstream depth hup2 B1 had
upstream drag coefficient Cd = 0.540 and a maximum of 1.425; E2 yielded Cd = 0.427
upstream which is lower than its Series B counterpart by a slight amount although
E2’s maximum was 0.657 which is significantly lower.
Tests with the flexible rods were exciting as they took the project in a direction
that is relatively unexplored in the area of the research. The choice of glow sticks may
seem surprising but recreating the biomechanical features of plant life is extremely
difficult in a laboratory situation; Järvelä (2002) used real plants in the laboratory
which encountered problems of dying and breaking and Wu et al. (1999) used a
rubberised horsehair mattress which had no real scientific basis to its use. Indeed,
Wu stated that as the mattress was likely to be stiffer than vegetation then its
introduction can only be an advantage as it would surely be a better model than
the rigid rods, a idea followed here in justification of using glow sticks. The only
study found to use any realistic plant type material was by Wilson et al. (2003) who
manufactured rubber type plants; certainly beyond the scope of this research.
The literature review at the start of the project gave a strong overview of the
subject and methods that could be brought into the laboratory, however, the exper-
imental process still required original work to be carried out, such as control tests
examining the effects of the boards on the flow and the accuracy of flow rate calcu-
lations. The exact tests to be performed then needed to be well planned to fit in
with the tight time frame. Given the opportunity of further research the normal flow
experiments would be extended as the results clearly held anomalies which could not
CHAPTER 6. CONCLUSIONS 90

be properly identified owing to the smaller range of results as opposed to the hor-
izontal channel tests. Further tests on the roughness of the channel could also be
very relevant to the results, including the uncovered board holes and channel walls
discussed a the end of the previous chapter. Also, the apparatus used by other au-
thors (Ghisalberti and Nepf, 2005; Järvelä, 2002) was of much larger dimensions, this
meant their results were not as influenced by the up- and downstream interactions
which played a significant role in this research; it would be beneficial to repeat some
of the experiments from this project on larger scales. The boards to hold the rods
used in this project also allowed a range of solid volume fractions that gave a solid
base of comparison, however, these solid volume fractions were all relatively low; the
maximum was φ = 0.054, although using the maximum five rods per row would have
given φ = 0.068. Tanino and Nepf (2008) went up to φ = 0.35 and Koch and Ladd
(1996) used φ = 0.40. It would be very beneficial to the research to investigate solid
volume fractions over a range up to similar values.
This project gives a solid framework for further research in directions following
rigid or flexible rods as well as for normal flow. Starting with initial theory of open
channel flow, this project has then covered literature relevant to the problem to
be investigated and shown how this can be used to plan and execute a series of
experiments. The theory is then used to analyse the results and the findings have
been compared with other studies as well as used to find new results.
Bibliography

Acheson, D. J. (1990). Elementary Fluid Dynamics. Oxford University Press, 2006


edition.

Ayaz, F. and Pedley, T. J. (1999). Flow through and particle interception by an


infinite array of closely-spaced circular cylinders. European Journal of Mechanics
- B/Fluids, 18(2):173 – 196.

Bixio, A. C. (2007). Experimental observations on flow resistance through emergent


vegetation. In Proceedings of the congress- International Association for Hydraulic
Research, volume 1, page 166. CONF.

Blevins, R. D. (1977). Flow-induced Vibration. Van Nostrand Reinhold Company,


second edition.

Buat, P. L. G. d. (1779). Principes d’Hydraulique. (Hydraulic Principles). 3rd edition.

Chanson, H. C. (1999). The Hydraulics of Open Channel Flow: An Introduction.


Elsevier, second edition.

Cheng, N. S. (2007). Power-law index for velocity profiles in open channel flows.
Advances in Water Resources, 30:1775–1784.

Colebrook, C. F. (1939). Turbulent flow in pipes, with particular reference to the


transition region between the smooth and rough pipe laws. Journal Institute Civil
Engineering, 1938 - 1939(4):133–566.

Fathi-Maghadam, M. and Kouwen, N. (1997). Nonrigid, nonsubmerged, vegetative


roughness on floodplains. Journal of Hydraulic Engineering, 123(1):51–57.

91
BIBLIOGRAPHY 92

Finnigan, J. (2000). Turbulence in plant canopies. Annual Review of Fluid Mechanics,


32(1):519–571.

Gabbai, R. D. and Benaroya, H. (2005). An overview of modeling and experiments


of vortex-induced vibration of circular cylinders. Journal of Sound and Vibration,
282:575–616.

Ghisalberti, M. and Nepf, H. M. (2004). The limited growth of vegetated shear layers.
Water Resources Research, 40:W07502.

Ghisalberti, M. and Nepf, H. M. (2005). Mass transport in vegetated shear flows.


Environmental Fluid Mechanics, 5:527 – 551.

Graf, W. H. and Altinakar, M. S. (1998). Fluvial Hydraulics: Flow and transport


processes in channels of simple geometry. Wiley, 3rd edition.

Hickin, E. J. (1984). Vegetation and river channel dynamics. The Canadian Geogra-
pher, 28(2):111–126.

Jadhav, R. S. and Buchberger, S. G. (1995). Effects of vegetation on flow through


free water surface wetlands. Ecological Engineering, 5:481–496.

Järvelä, J. (2002). Flow resistance of flexible and stiff vegetation: a flume study with
natural plants. Journal of Hydrology, 269(1-2):44 – 54.

Klopstra, D., Barneveld, H. J., van Noortwijk, J. M., and van Velzen, E. H. (1997).
Analytical model for hydraulic roughness of submerged vegetation. In The 27th
Congress of the International Association for Hydraulic Research, volume 3, pages
775 – 780. ASCE.

Koch, D. L. and Ladd, A. J. C. (1996). Moderate reynolds number flows through


periodic and random arrays of aligned cylinders. Journal of Fluid Mechanics,
349:31–66.
BIBLIOGRAPHY 93

Lee, J. K., Roig, L. C. ad Jenter, H. L., and Visser, H. M. (2004). Drag coefficients
for modeling through periodic and random arrays of aligned cylinders. Ecological
Engineering, 22(4):237 – 248.

López, F. and Garcı́a, M. (1998). Open-channel flow through simulated vegetation:


Suspended sediment transport modeling. Water Resources Research, 34:2341–2352.

Moody, L. F. (1944). Friction factors for pipe flow. Transactions of the ASME,
66:671–684.

Naot, D., Nezu, I., and Nakagawa, H. (1996). Hydrodynamic behavior of partly
vegetated open channels. Journal of Hydraulic Engineering, 122(11):625–633.

Neary, V. S. (2003). Numerical solution of fully developed flow with vegetative resis-
tance. Journal of Engineering Mechanics, 129(5):558–563.

Nepf, H. M. (1999). Drag, turbulence, and diffusion in flow through emergent vege-
tation. Water Resources Research, 35:479–490.

Prandtl, L. (1935). The mechanics of viscous flows, page 142. Springer-Verlag, Berlin,
3rd edition.

Reynolds, O. (1883). An experimental investigation of the circumstances which de-


termine whether the motion of water shall be direct or sinuous, and of the law of
resistance in parallel channels. Philosophical Transactions of the Royal Society of
London, 174:935–982.

Stoesser, T., Wilson, C. A. M. E., Bates, P. D., and Dittrich, A. (2003). Applica-
tion of a 3d numerical model to a river with vegetated floodplains. Journal of
Hydroinformatics, 5(2):99–112.

Tanino, Y. and Nepf, H. M. (2008). Laboratory investigation of mean drag in a


random array of rigid, emergent cylinders. Journal of Hydraulic Engineering,
134(1):34–41.
BIBLIOGRAPHY 94

Velasco, D., Bateman, A., Redondo, J. M., and Demedina, V. (2003). An open
channel flow experimental and theoretical study of resistance and turbulent char-
acterization over flexible vegetated linings. Flow, Turbulence and Combustion,
70:69–88.

Wilson, C. A. M. E., Stoesser, T., Bates, P. D., and Batemann Pinzen, A. (2003).
Open channel flow through different forms of submerged flexible vegetation. Jour-
nal of Hydraulic Engineering, 129(11):847–853.

Worcester, S. E. (1995). Effects of eelgrass beds on advection and turbulent mixing in


low current and low shoot density environments. Marine Ecology Progress Series,
126:223–232.

Wu, F.-C., Shen, H. W., and Chou, Y.-J. (1999). Variation of roughness coefficients
for unsubmerged and submerged vegetation. Journal of Hydraulic Engineering,
125(9):934–942.
Appendix A

Nomenclature

Symbol Description Units Notes


α Angle of channel slope
γ Vortex shedding frequency s−1
λ Vegetal area coefficient m−1
ν Kinematic viscosity m2 s−1
ρ Water density kg m−3
σ No. of permutations of rows
τ Shear stress N m−2

A Channel cross sectional area m2 Subscripts c and p denote


channel cross section and rods
facial area
b Channel width m
Ć Chézy coefficient m1/2 s−1
Cd Drag coefficient
Cd 0 Alternate drag coefficient m−1
Cf Skin friction coefficient
d Rod diameter m
DH Hydraulic diameter m
f Darcy-Wesibach friction factor

95
APPENDIX A. NOMENCLATURE 96

Fd Drag Force N
Fm Fluid mass force N
Fp Pressure force N
Fr Froude number
h Flow depth m
hup Upstream flow depth m
hN Normal flow depth m
ks Roughness mm
L Length of control volume m
L Characteristic length m
m Rods per row Subscripts r and f denote rigid
and flexible rods
n No. rows
nm Manning’s roughness coefficient s m−1/3
PW Wetted perimeter m
p Pressure Pa patm denotes atmospheric
pressure
Q Flow rate m3 s−1
Re Reynolds number
Rec Open channel Reynolds number
Rep Cylindrical Reynolds number
St Strouhal number
u Flow velocity m s−1
U Characteristic velocity m s−1
V Fluid volume m3
xi Position along channel length m
Appendix B

Fortran 90 Program

B.1 Description

Several programs were written using Fortran 90 all following the same form but
adapted for each specific series of tests. For this reason only one code is included,
the commonest used program for calculating the velocity, u; the Froude number,
Fr; both Reynolds numbers, Rec and Rep ; and the drag coefficient, Cd . The steps of
calculations are the same as in chapter 4 after the program reads a data file containing;
the Start and End measurements of the volume tank and time elapsed which are used
for calculating the flow rate; the a, b and c parts of the quadratic equation

z = ax2 + bx + c,

which was calculated from the Mathematica Fit function; the number of rods per
row (m); and the name of an output file to save the results to.

B.2 Code

MODULE Global_data
!Symbolic names for kind types of single- and double-precision reals:
INTEGER, PARAMETER :: SP = KIND(1.0)
INTEGER, PARAMETER :: DP = KIND(1.0D0)

97
APPENDIX B. FORTRAN 90 PROGRAM 98

!Constants used:
REAL(dp), PARAMETER :: pi=3.141592653589793238462643383279502884197_dp,&
density=1000_dp, viscosity=1.0E-06_dp, constB=0.104_dp, constG=9.81_dp,&
diameter=0.006_dp
INTEGER, PARAMETER :: lines = 18

REAL(dp) :: flowQ, lineA, lineB, lineC, rods_per_row


CHARACTER(LEN=50) :: output_file

CONTAINS
! subroutine gives choice of inputting via file or typing in values
! assumes file is in correct format
SUBROUTINE input_parameters

IMPLICIT NONE

INTEGER :: i
REAL(dp) :: flowStart, flowEnd, flowTime,parameters(6)
CHARACTER(LEN=1) :: choice
CHARACTER(LEN=50) :: input_file

! Calculating flow rate


WRITE(6,*)’TYPE f FOR FILE m FOR MANUALLY::’
READ*,choice
IF (choice .EQ. ’f’) THEN
WRITE(6,*)’ENTER FILENAME::’
READ*,input_file
OPEN(UNIT=9,file=input_file)

DO i=1,6
APPENDIX B. FORTRAN 90 PROGRAM 99

READ(9,*)
READ(9,*)parameters(i)
ENDDO
READ(9,*)
READ(9,*)rods_per_row
READ(9,*)
READ(9,*)output_file

flowStart = parameters(1)
flowEnd = parameters(2)
flowTime = parameters(3)
lineA = parameters(4)
lineB = parameters(5)
lineC = parameters(6)
CLOSE(UNIT=11)

ELSEIF (choice .EQ. ’m’) THEN


WRITE(6,*)’Start measurement: ’
READ*,flowStart
WRITE(6,*)’End measurement: ’
READ*,flowEnd
WRITE(6,*)’Time elapsed: ’
READ*,flowTime

WRITE(6,*)’Flow rate = ’,flowQ


! Other stuff
WRITE(6,*)’No rods per row: ’
READ*,rods_per_row
WRITE(6,*)’The "a" of the line z = ax^2 + bx + c: ’
READ*,lineA
APPENDIX B. FORTRAN 90 PROGRAM 100

WRITE(6,*)’The "b" of the line z = ax^2 + bx + c: ’


READ*,lineB
WRITE(6,*)’The "c" of the line z = ax^2 + bx + c: ’
READ*,lineC
WRITE(6,*)’Output filename: ’
READ*,output_file

ELSE
WRITE(6,*)’Error’
ENDIF
flowQ = (flowEnd - flowStart) * 0.0276 / flowTime

END SUBROUTINE input_parameters


END MODULE Global_data

PROGRAM seriesB
USE global_data, wp=>dp
IMPLICIT NONE
INTEGER :: i

REAL(wp) :: x(0:lines),z(0:lines), u(0:lines), wetp(0:lines), &


& hyd(0:lines), froude(0:lines), reynolds(0:lines), &
& reynoldsp(0:lines), tau(0:lines), dragC(1:lines), &
& tol, oneOverSqrtDarcy, oneOverSqrtDarcyTemp, darcy, &
& rateChangeMom, netPressureForce, shearStressForce, area, dragForce
REAL(wp), EXTERNAL :: funcDarcy
! Read parameters from file
CALL input_parameters
! Put data into the arrays...
DO i=0, lines
APPENDIX B. FORTRAN 90 PROGRAM 101

x(i) = i * 0.1
z(i) = lineC + lineB * x(i) + lineA * x(i) * x(i)
u(i) = flowQ / (constB * z(i))
wetp(i) = constB + 2. * z(i)
hyd(i) = constB * z(i) * 4. / wetp(i)
froude(i) = u(i) / sqrt(constG * z(i))
reynolds(i) = u(i) * hyd(i) / viscosity
reynoldsp(i) = u(i) * diameter / viscosity
tol = 1.
oneOverSqrtDarcy = 1.

DO WHILE (tol > 1.0E-06)


oneOverSqrtDarcyTemp = funcDarcy(oneOverSqrtDarcy, hyd(i), reynolds(i))
tol = abs( oneOverSqrtDarcy - oneOverSqrtDarcyTemp )
oneOverSqrtDarcy = oneOverSqrtDarcyTemp
ENDDO
darcy = 1./(oneOverSqrtDarcy * oneOverSqrtDarcy)
tau(i) = 0.125 * density * u(i) * u(i) * darcy
IF(i>0) THEN ! As forces use previous line’s data
rateChangeMom = density * flowQ * ( u(i) - u(i-1) )
netPressureForce = 0.5 * density * constG * constB * &
& ( z(i-1) * z(i-1) - z(i) * z(i) )
shearStressForce = ( tau(i) + tau(i-1) ) * &
& ( wetp(i) + wetp(i-1) ) * ( x(i) - x(i-1) ) / 4.
dragForce = netPressureForce - rateChangeMom - shearStressForce
area = ( z(i) + z(i-1) ) * rods_per_row * 5. * diameter / 2.
dragC(i) = 2. * dragForce / ( density * u(i) * u(i) * area )
ENDIF
ENDDO
OPEN(UNIT=11,file=TRIM(output_file)//’_quad.out’)
APPENDIX B. FORTRAN 90 PROGRAM 102

OPEN(UNIT=12,file=TRIM(output_file)//’_quad.tex’)
WRITE(11,’(9(E12.5,1X))’)x(0),z(0),u(0), &
& froude(0),reynolds(0),reynoldsp(0)
DO i=1, lines
WRITE(11,’(10(E12.5,1X))’)x(i),z(i),u(i), &
& froude(i),reynolds(i),reynoldsp(i),dragC(i)
ENDDO
CLOSE(UNIT=11)
CLOSE(UNIT=12)
END PROGRAM seriesB

FUNCTION funcDarcy(oneOverSqrtDarcy,inHyd,inRe)
USE global_data, wp=>dp
REAL(wp),INTENT(IN) :: oneOverSqrtDarcy, inHyd, inRe
REAL(wp) :: funcDarcy
funcDarcy = -2.0 * log10( (0.015 / (3.71 * inHyd)) + &
& (2.51 * oneOverSqrtDarcy / inRe) )
END FUNCTION funcDarcy
Appendix C

Results

Table C.1: Test B1 hup1


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0569 0.1636 0.2189 17788.8 981.8
0.1 0.0563 0.1656 0.2228 17898.1 993.4 1.1804
0.2 0.0556 0.1676 0.2270 18013.9 1005.8 1.2071
0.3 0.0548 0.1699 0.2316 18136.6 1019.2 1.2282
0.4 0.0541 0.1722 0.2365 18266.3 1033.5 1.2435
0.5 0.0533 0.1748 0.2417 18403.3 1048.8 1.2529
0.6 0.0525 0.1775 0.2474 18547.9 1065.2 1.2564
0.7 0.0516 0.1805 0.2536 18700.5 1082.8 1.2538
0.8 0.0507 0.1836 0.2602 18861.2 1101.6 1.2450
0.9 0.0498 0.1870 0.2674 19030.6 1121.8 1.2301
1.0 0.0489 0.1906 0.2752 19209.0 1143.5 1.2088
1.1 0.0479 0.1945 0.2837 19396.8 1166.9 1.1813
1.2 0.0469 0.1987 0.2929 19594.5 1191.9 1.1475
1.3 0.0459 0.2032 0.3029 19802.6 1218.9 1.1073
1.4 0.0448 0.2080 0.3138 20021.6 1248.1 1.0607
1.5 0.0437 0.2133 0.3257 20252.0 1279.5 1.0078
1.6 0.0426 0.2189 0.3388 20494.6 1313.6 0.9486
1.7 0.0414 0.2251 0.3532 20750.0 1350.5 0.8831
1.8 0.0402 0.2318 0.3691 21018.9 1390.7 0.8112

103
APPENDIX C. RESULTS 104

Table C.2: Test B1 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0846 0.2247 0.2467 28937.3 1348.0
0.1 0.0839 0.2265 0.2496 29080.5 1358.8 0.5396
0.2 0.0831 0.2287 0.2534 29258.1 1372.3 0.7058
0.3 0.0821 0.2314 0.2579 29471.1 1388.6 0.8591
0.4 0.0810 0.2347 0.2633 29721.3 1408.0 0.9976
0.5 0.0797 0.2384 0.2697 30010.3 1430.6 1.1194
0.6 0.0783 0.2428 0.2771 30340.3 1456.8 1.2229
0.7 0.0767 0.2478 0.2857 30714.0 1486.9 1.3065
0.8 0.0749 0.2536 0.2957 31134.1 1521.3 1.3688
0.9 0.0730 0.2601 0.3073 31604.3 1560.7 1.4088
1.0 0.0710 0.2676 0.3206 32128.6 1605.5 1.4254
1.1 0.0688 0.2761 0.3361 32711.6 1656.7 1.4178
1.2 0.0665 0.2858 0.3540 33358.9 1715.1 1.3855
1.3 0.0640 0.2970 0.3749 34077.0 1782.0 1.3282
1.4 0.0613 0.3098 0.3995 34873.4 1859.0 1.2456
1.5 0.0585 0.3247 0.4285 35757.0 1947.9 1.1379
1.6 0.0556 0.3419 0.4631 36738.3 2051.4 1.0053
1.7 0.0525 0.3621 0.5048 37830.0 2172.8 0.8480
1.8 0.0492 0.3861 0.5557 39047.1 2316.6 0.6666

Table C.3: Test B1 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0934 0.2403 0.2510 32105.9 1441.6
0.1 0.0927 0.2421 0.2539 32264.3 1452.7 0.4947
0.2 0.0918 0.2445 0.2576 32464.6 1466.8 0.6679
0.3 0.0907 0.2474 0.2622 32708.3 1484.2 0.8281
0.4 0.0895 0.2508 0.2677 32997.3 1504.9 0.9730
0.5 0.0881 0.2549 0.2742 33334.1 1529.3 1.1006
0.6 0.0865 0.2596 0.2819 33721.3 1557.8 1.2092
0.7 0.0847 0.2651 0.2909 34162.2 1590.6 1.2969
0.8 0.0827 0.2714 0.3013 34660.8 1628.5 1.3622
0.9 0.0806 0.2786 0.3134 35221.6 1671.8 1.4040
1.0 0.0782 0.2869 0.3275 35850.1 1721.5 1.4211
1.1 0.0757 0.2964 0.3439 36552.5 1778.4 1.4126
1.2 0.0731 0.3073 0.3630 37336.4 1843.6 1.3781
1.3 0.0702 0.3198 0.3854 38210.4 1918.8 1.3170
1.4 0.0672 0.3343 0.4118 39185.1 2005.6 1.2294
1.5 0.0639 0.3511 0.4433 40272.9 2106.6 1.1152
1.6 0.0605 0.3708 0.4812 41488.6 2224.8 0.9749
1.7 0.0570 0.3941 0.5272 42850.2 2364.5 0.8088
1.8 0.0532 0.4219 0.5840 44379.6 2531.3 0.6173
APPENDIX C. RESULTS 105

Table C.4: Test B2 hup1


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0597 0.1397 0.1826 15532.2 838.3
0.1 0.0590 0.1413 0.1857 15625.4 847.8 0.9075
0.2 0.0583 0.1430 0.1891 15727.0 858.2 0.9613
0.3 0.0575 0.1450 0.1929 15837.5 869.7 1.0088
0.4 0.0567 0.1471 0.1971 15957.2 882.3 1.0496
0.5 0.0558 0.1494 0.2018 16086.4 896.1 1.0834
0.6 0.0549 0.1519 0.2069 16225.6 911.2 1.1099
0.7 0.0539 0.1546 0.2126 16375.3 927.7 1.1289
0.8 0.0529 0.1576 0.2188 16535.9 945.8 1.1404
0.9 0.0518 0.1609 0.2257 16708.0 965.5 1.1440
1.0 0.0507 0.1645 0.2333 16892.3 987.0 1.1397
1.1 0.0495 0.1684 0.2417 17089.4 1010.7 1.1274
1.2 0.0483 0.1728 0.2510 17300.1 1036.5 1.1071
1.3 0.0470 0.1775 0.2614 17525.2 1065.0 1.0787
1.4 0.0457 0.1827 0.2730 17765.7 1096.2 1.0423
1.5 0.0443 0.1884 0.2860 18022.6 1130.7 0.9979
1.6 0.0428 0.1948 0.3006 18297.1 1168.8 0.9456
1.7 0.0413 0.2019 0.3171 18590.4 1211.1 0.8855
1.8 0.0398 0.2097 0.3357 18903.9 1258.3 0.8177

Table C.5: Test B2 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0801 0.1699 0.1916 21431.4 1019.4
0.1 0.0793 0.1717 0.1947 21569.4 1030.2 0.7853
0.2 0.0783 0.1739 0.1984 21731.5 1043.1 0.9050
0.3 0.0772 0.1764 0.2027 21918.8 1058.2 1.0128
0.4 0.0759 0.1793 0.2077 22132.5 1075.6 1.1077
0.5 0.0746 0.1826 0.2135 22374.0 1095.5 1.1883
0.6 0.0731 0.1864 0.2201 22645.1 1118.2 1.2538
0.7 0.0714 0.1907 0.2278 22947.6 1144.0 1.3033
0.8 0.0696 0.1956 0.2366 23284.0 1173.3 1.3359
0.9 0.0677 0.2011 0.2467 23656.8 1206.5 1.3512
1.0 0.0657 0.2074 0.2584 24069.0 1244.2 1.3488
1.1 0.0635 0.2145 0.2718 24524.3 1287.0 1.3282
1.2 0.0612 0.2226 0.2874 25026.8 1335.8 1.2895
1.3 0.0587 0.2319 0.3056 25581.1 1391.6 1.2328
1.4 0.0561 0.2426 0.3270 26192.9 1455.7 1.1582
1.5 0.0534 0.2550 0.3523 26868.8 1529.9 1.0663
1.6 0.0505 0.2694 0.3826 27616.6 1616.3 0.9575
1.7 0.0476 0.2863 0.4192 28445.3 1717.8 0.8326
1.8 0.0444 0.3064 0.4641 29366.0 1838.5 0.6923
APPENDIX C. RESULTS 106

Table C.6: Test B2 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0961 0.1904 0.1961 25696.0 1142.3
0.1 0.0951 0.1923 0.1991 25866.6 1154.0 0.7248
0.2 0.0940 0.1947 0.2028 26073.7 1168.4 0.8674
0.3 0.0926 0.1976 0.2073 26319.1 1185.5 0.9965
0.4 0.0910 0.2010 0.2126 26604.9 1205.8 1.1102
0.5 0.0893 0.2049 0.2189 26933.6 1229.3 1.2069
0.6 0.0874 0.2094 0.2262 27308.1 1256.6 1.2854
0.7 0.0852 0.2147 0.2348 27732.0 1288.0 1.3442
0.8 0.0829 0.2207 0.2447 28209.3 1324.1 1.3825
0.9 0.0804 0.2276 0.2563 28745.1 1365.5 1.3993
1.0 0.0777 0.2355 0.2698 29345.0 1413.0 1.3941
1.1 0.0748 0.2446 0.2856 30015.9 1467.8 1.3667
1.2 0.0717 0.2552 0.3043 30766.0 1531.1 1.3170
1.3 0.0684 0.2674 0.3264 31604.8 1604.6 1.2452
1.4 0.0649 0.2817 0.3530 32544.2 1690.4 1.1517
1.5 0.0613 0.2986 0.3851 33597.9 1791.6 1.0373
1.6 0.0574 0.3187 0.4246 34783.2 1912.1 0.9031
1.7 0.0534 0.3429 0.4739 36121.0 2057.3 0.7500
1.8 0.0491 0.3725 0.5366 37637.1 2235.1 0.5792

Table C.7: Test B3 hup1


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0552 0.1005 0.1365 10770.7 603.1
0.1 0.0547 0.1016 0.1387 10829.8 609.6 1.1271
0.2 0.0540 0.1028 0.1412 10894.2 616.7 1.1924
0.3 0.0534 0.1041 0.1439 10964.2 624.5 1.2509
0.4 0.0526 0.1055 0.1468 11039.8 633.1 1.3022
0.5 0.0519 0.1071 0.1501 11121.4 642.5 1.3461
0.6 0.0511 0.1088 0.1537 11209.2 652.7 1.3823
0.7 0.0502 0.1106 0.1577 11303.4 663.9 1.4106
0.8 0.0493 0.1127 0.1620 11404.4 676.0 1.4307
0.9 0.0483 0.1149 0.1668 11512.4 689.3 1.4425
1.0 0.0473 0.1173 0.1721 11627.8 703.8 1.4459
1.1 0.0463 0.1199 0.1780 11751.0 719.7 1.4407
1.2 0.0452 0.1228 0.1844 11882.5 737.0 1.4271
1.3 0.0441 0.1260 0.1916 12022.6 755.9 1.4048
1.4 0.0429 0.1294 0.1995 12171.9 776.7 1.3741
1.5 0.0417 0.1333 0.2084 12331.0 799.5 1.3349
1.6 0.0404 0.1374 0.2183 12500.4 824.6 1.2874
1.7 0.0391 0.1421 0.2294 12681.0 852.4 1.2317
1.8 0.0377 0.1472 0.2419 12873.4 883.1 1.1681
APPENDIX C. RESULTS 107

Table C.8: Test B3 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0802 0.1409 0.1589 17778.1 845.4
0.1 0.0792 0.1427 0.1619 17912.0 856.0 0.9701
0.2 0.0781 0.1447 0.1653 18064.3 868.1 1.0624
0.3 0.0769 0.1470 0.1693 18235.8 881.9 1.1438
0.4 0.0755 0.1496 0.1738 18427.5 897.6 1.2135
0.5 0.0741 0.1525 0.1790 18640.7 915.3 1.2707
0.6 0.0725 0.1559 0.1848 18876.6 935.2 1.3149
0.7 0.0708 0.1596 0.1915 19137.0 957.5 1.3455
0.8 0.0690 0.1638 0.1991 19423.4 982.7 1.3621
0.9 0.0670 0.1685 0.2077 19738.0 1010.9 1.3645
1.0 0.0650 0.1738 0.2176 20083.1 1042.7 1.3524
1.1 0.0628 0.1798 0.2290 20461.4 1078.6 1.3258
1.2 0.0606 0.1865 0.2420 20876.0 1119.3 1.2848
1.3 0.0582 0.1942 0.2571 21330.5 1165.4 1.2297
1.4 0.0556 0.2030 0.2748 21829.0 1218.1 1.1608
1.5 0.0530 0.2131 0.2955 22376.4 1278.6 1.0786
1.6 0.0503 0.2248 0.3201 22978.3 1348.6 0.9837
1.7 0.0474 0.2384 0.3496 23641.3 1430.2 0.8768
1.8 0.0444 0.2544 0.3854 24373.5 1526.4 0.7588

Table C.9: Test B3 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0987 0.1599 0.1625 21777.5 959.2
0.1 0.0975 0.1618 0.1655 21950.7 970.9 0.9110
0.2 0.0961 0.1641 0.1690 22151.6 984.6 1.0186
0.3 0.0946 0.1667 0.1731 22381.7 1000.4 1.1141
0.4 0.0929 0.1698 0.1778 22642.8 1018.6 1.1964
0.5 0.0911 0.1732 0.1833 22936.8 1039.4 1.2644
0.6 0.0890 0.1772 0.1896 23266.1 1063.1 1.3175
0.7 0.0869 0.1816 0.1968 23633.4 1089.9 1.3548
0.8 0.0845 0.1867 0.2051 24041.9 1120.3 1.3759
0.9 0.0820 0.1925 0.2146 24495.3 1154.8 1.3803
1.0 0.0793 0.1990 0.2257 24997.9 1194.1 1.3678
1.1 0.0764 0.2065 0.2385 25554.8 1238.8 1.3384
1.2 0.0734 0.2150 0.2534 26172.0 1289.9 1.2922
1.3 0.0702 0.2248 0.2709 26856.4 1348.6 1.2295
1.4 0.0668 0.2361 0.2916 27616.3 1416.5 1.1509
1.5 0.0633 0.2492 0.3163 28461.8 1495.5 1.0571
1.6 0.0596 0.2647 0.3462 29404.6 1588.2 0.9489
1.7 0.0557 0.2831 0.3828 30459.2 1698.3 0.8274
1.8 0.0517 0.3051 0.4284 31643.3 1830.8 0.6936
APPENDIX C. RESULTS 108

Table C.10: Test B4 hup1


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0595 0.0899 0.1177 9987.1 539.7
0.1 0.0587 0.0913 0.1204 10067.0 547.8 1.6844
0.2 0.0577 0.0928 0.1233 10151.6 556.6 1.7010
0.3 0.0568 0.0943 0.1264 10241.3 566.0 1.7111
0.4 0.0558 0.0960 0.1298 10336.3 576.2 1.7146
0.5 0.0547 0.0978 0.1335 10436.7 587.1 1.7115
0.6 0.0537 0.0998 0.1376 10543.0 598.8 1.7018
0.7 0.0525 0.1019 0.1420 10655.3 611.5 1.6854
0.8 0.0514 0.1042 0.1468 10774.0 625.2 1.6625
0.9 0.0502 0.1067 0.1520 10899.4 640.1 1.6331
1.0 0.0490 0.1094 0.1578 11032.0 656.1 1.5973
1.1 0.0477 0.1123 0.1641 11172.1 673.5 1.5552
1.2 0.0464 0.1154 0.1710 11320.2 692.5 1.5069
1.3 0.0451 0.1188 0.1788 11476.8 713.1 1.4526
1.4 0.0437 0.1226 0.1873 11642.4 735.6 1.3925
1.5 0.0423 0.1267 0.1968 11817.7 760.3 1.3269
1.6 0.0408 0.1312 0.2074 12003.3 787.5 1.2559
1.7 0.0393 0.1362 0.2194 12199.8 817.4 1.1799
1.8 0.0378 0.1418 0.2329 12408.2 850.6 1.0991

Table C.11: Test B4 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0829 0.1244 0.1379 15898.3 746.3
0.1 0.0820 0.1257 0.1402 16004.8 754.4 0.8709
0.2 0.0809 0.1274 0.1429 16131.6 764.3 1.0060
0.3 0.0797 0.1293 0.1462 16279.7 775.9 1.1280
0.4 0.0784 0.1316 0.1500 16450.2 789.4 1.2352
0.5 0.0769 0.1342 0.1545 16644.5 805.0 1.3265
0.6 0.0752 0.1372 0.1597 16864.1 823.0 1.4005
0.7 0.0733 0.1406 0.1657 17110.8 843.5 1.4564
0.8 0.0713 0.1445 0.1727 17386.9 867.1 1.4933
0.9 0.0692 0.1490 0.1808 17694.7 894.0 1.5106
1.0 0.0669 0.1541 0.1902 18037.3 924.7 1.5079
1.1 0.0644 0.1600 0.2012 18418.1 960.0 1.4851
1.2 0.0618 0.1668 0.2141 18841.1 1000.6 1.4423
1.3 0.0591 0.1746 0.2294 19311.1 1047.5 1.3799
1.4 0.0561 0.1837 0.2475 19833.7 1102.1 1.2985
1.5 0.0530 0.1944 0.2694 20415.7 1166.2 1.1990
1.6 0.0498 0.2070 0.2961 21065.2 1242.0 1.0827
1.7 0.0464 0.2221 0.3292 21791.8 1332.8 0.9509
1.8 0.0429 0.2405 0.3709 22607.4 1443.2 0.8052
APPENDIX C. RESULTS 109

Table C.12: Test B4 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1018 0.1408 0.1409 19377.3 844.6
0.1 0.1005 0.1425 0.1436 19537.3 855.2 0.9657
0.2 0.0990 0.1446 0.1467 19725.9 867.8 1.0930
0.3 0.0974 0.1471 0.1505 19944.7 882.6 1.2057
0.4 0.0955 0.1499 0.1549 20195.6 899.7 1.3024
0.5 0.0935 0.1532 0.1600 20480.8 919.5 1.3819
0.6 0.0912 0.1570 0.1660 20803.0 942.2 1.4430
0.7 0.0888 0.1614 0.1729 21165.4 968.2 1.4849
0.8 0.0861 0.1663 0.1810 21571.6 998.0 1.5069
0.9 0.0833 0.1720 0.1903 22026.1 1032.2 1.5085
1.0 0.0802 0.1786 0.2013 22534.0 1071.4 1.4896
1.1 0.0770 0.1861 0.2142 23101.4 1116.6 1.4503
1.2 0.0735 0.1948 0.2294 23735.6 1169.0 1.3909
1.3 0.0699 0.2050 0.2476 24445.5 1229.9 1.3121
1.4 0.0660 0.2169 0.2695 25241.5 1301.5 1.2147
1.5 0.0620 0.2311 0.2963 26136.7 1386.4 1.1001
1.6 0.0577 0.2480 0.3295 27146.8 1488.2 0.9697
1.7 0.0533 0.2687 0.3716 28291.4 1612.2 0.8252
1.8 0.0487 0.2943 0.4259 29595.1 1765.8 0.6686

Table C.13: Test D1 hup1


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0565 0.1495 0.2007 16195.1 896.9
0.1 0.0557 0.1517 0.2051 16317.2 909.9 0.9548
0.2 0.0549 0.1539 0.2097 16442.6 923.5 0.9308
0.3 0.0541 0.1563 0.2146 16571.5 937.8 0.9058
0.4 0.0532 0.1588 0.2197 16704.1 952.6 0.8800
0.5 0.0524 0.1613 0.2251 16840.5 968.1 0.8532
0.6 0.0515 0.1640 0.2308 16980.8 984.3 0.8256
0.7 0.0506 0.1669 0.2368 17125.2 1001.3 0.7971
0.8 0.0498 0.1698 0.2431 17273.8 1019.0 0.7677
0.9 0.0489 0.1729 0.2498 17426.7 1037.6 0.7374
1.0 0.0480 0.1762 0.2569 17584.2 1057.2 0.7064
1.1 0.0470 0.1796 0.2644 17746.3 1077.7 0.6745
1.2 0.0461 0.1832 0.2724 17913.3 1099.3 0.6418
1.3 0.0452 0.1870 0.2809 18085.4 1122.0 0.6083
1.4 0.0442 0.1910 0.2899 18262.8 1145.9 0.5741
1.5 0.0433 0.1952 0.2995 18445.6 1171.2 0.5390
1.6 0.0423 0.1996 0.3098 18634.2 1197.9 0.5032
1.7 0.0414 0.2043 0.3208 18828.7 1226.1 0.4667
1.8 0.0404 0.2093 0.3326 19029.4 1256.0 0.4294
APPENDIX C. RESULTS 110

Table C.14: Test D1 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0740 0.2032 0.2385 24811.1 1219.0
0.1 0.0729 0.2060 0.2435 25013.6 1236.0 0.6081
0.2 0.0718 0.2091 0.2491 25232.8 1254.7 0.6303
0.3 0.0707 0.2125 0.2552 25469.6 1275.1 0.6479
0.4 0.0695 0.2162 0.2619 25725.0 1297.5 0.6607
0.5 0.0682 0.2203 0.2694 25999.8 1321.9 0.6685
0.6 0.0668 0.2248 0.2776 26295.2 1348.6 0.6713
0.7 0.0654 0.2296 0.2866 26612.4 1377.8 0.6689
0.8 0.0639 0.2350 0.2967 26952.7 1409.7 0.6614
0.9 0.0624 0.2408 0.3078 27317.6 1444.7 0.6485
1.0 0.0608 0.2472 0.3201 27708.8 1483.1 0.6303
1.1 0.0591 0.2542 0.3339 28128.1 1525.3 0.6068
1.2 0.0574 0.2620 0.3492 28577.6 1571.8 0.5780
1.3 0.0555 0.2705 0.3665 29059.5 1623.1 0.5439
1.4 0.0537 0.2800 0.3859 29576.5 1680.0 0.5045
1.5 0.0517 0.2905 0.4079 30131.5 1743.2 0.4599
1.6 0.0497 0.3023 0.4329 30727.6 1813.8 0.4101
1.7 0.0476 0.3155 0.4616 31368.7 1892.9 0.3552
1.8 0.0455 0.3304 0.4946 32058.8 1982.2 0.2952

Table C.15: Test D1 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1000 0.2513 0.2537 34394.3 1507.9
0.1 0.0987 0.2547 0.2588 34695.2 1528.0 0.5114
0.2 0.0973 0.2584 0.2646 35029.7 1550.6 0.5471
0.3 0.0957 0.2626 0.2710 35399.4 1575.9 0.5772
0.4 0.0940 0.2673 0.2783 35806.5 1604.0 0.6014
0.5 0.0922 0.2726 0.2865 36253.3 1635.3 0.6196
0.6 0.0903 0.2783 0.2957 36742.4 1670.1 0.6314
0.7 0.0883 0.2848 0.3060 37276.8 1708.7 0.6367
0.8 0.0861 0.2919 0.3176 37860.0 1751.6 0.6352
0.9 0.0838 0.2999 0.3307 38495.7 1799.2 0.6269
1.0 0.0814 0.3087 0.3454 39188.5 1852.3 0.6118
1.1 0.0789 0.3186 0.3621 39943.3 1911.4 0.5896
1.2 0.0763 0.3296 0.3810 40765.9 1977.6 0.5605
1.3 0.0735 0.3420 0.4027 41663.0 2052.0 0.5244
1.4 0.0706 0.3559 0.4276 42642.3 2135.7 0.4815
1.5 0.0676 0.3718 0.4564 43712.7 2230.5 0.4318
1.6 0.0645 0.3898 0.4900 44884.4 2338.5 0.3753
1.7 0.0613 0.4104 0.5294 46169.7 2462.3 0.3123
1.8 0.0579 0.4342 0.5762 47582.8 2605.4 0.2427
APPENDIX C. RESULTS 111

Table C.16: Test D2 hup1


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0563 0.1091 0.1469 11799.1 654.8
0.1 0.0556 0.1105 0.1497 11877.1 663.2 1.1589
0.2 0.0548 0.1120 0.1527 11958.7 672.0 1.1635
0.3 0.0541 0.1136 0.1559 12044.1 681.4 1.1651
0.4 0.0533 0.1152 0.1594 12133.3 691.4 1.1636
0.5 0.0525 0.1170 0.1630 12226.6 702.0 1.1591
0.6 0.0517 0.1189 0.1670 12324.0 713.2 1.1515
0.7 0.0508 0.1209 0.1712 12425.9 725.2 1.1408
0.8 0.0500 0.1230 0.1757 12532.2 737.8 1.1271
0.9 0.0491 0.1252 0.1805 12643.3 751.3 1.1103
1.0 0.0481 0.1276 0.1857 12759.4 765.6 1.0905
1.1 0.0472 0.1302 0.1913 12880.6 780.9 1.0677
1.2 0.0462 0.1329 0.1973 13007.1 797.2 1.0420
1.3 0.0452 0.1358 0.2038 13139.3 814.6 1.0133
1.4 0.0442 0.1389 0.2108 13277.4 833.3 0.9817
1.5 0.0432 0.1422 0.2185 13421.8 853.2 0.9473
1.6 0.0421 0.1458 0.2267 13572.6 874.7 0.9101
1.7 0.0411 0.1496 0.2357 13730.2 897.7 0.8702
1.8 0.0400 0.1537 0.2456 13895.1 922.5 0.8276

Table C.17: Test D2 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0777 0.1630 0.1868 20312.8 978.3
0.1 0.0767 0.1652 0.1905 20471.3 991.1 0.7001
0.2 0.0756 0.1676 0.1947 20648.5 1005.6 0.7530
0.3 0.0743 0.1703 0.1994 20845.3 1021.9 0.7987
0.4 0.0730 0.1734 0.2048 21062.8 1040.1 0.8369
0.5 0.0716 0.1768 0.2108 21302.1 1060.5 0.8672
0.6 0.0701 0.1805 0.2177 21564.5 1083.3 0.8892
0.7 0.0685 0.1848 0.2254 21851.7 1108.6 0.9027
0.8 0.0668 0.1895 0.2340 22165.2 1136.9 0.9074
0.9 0.0650 0.1947 0.2438 22507.2 1168.5 0.9033
1.0 0.0631 0.2006 0.2550 22879.7 1203.8 0.8902
1.1 0.0611 0.2072 0.2676 23285.4 1243.3 0.8681
1.2 0.0590 0.2146 0.2821 23727.1 1287.6 0.8370
1.3 0.0568 0.2229 0.2987 24208.2 1337.7 0.7972
1.4 0.0545 0.2324 0.3178 24732.5 1394.3 0.7486
1.5 0.0521 0.2431 0.3401 25304.2 1458.7 0.6916
1.6 0.0496 0.2554 0.3662 25928.6 1532.4 0.6264
1.7 0.0470 0.2696 0.3971 26611.5 1617.5 0.5534
1.8 0.0443 0.2861 0.4341 27359.8 1716.4 0.4729
APPENDIX C. RESULTS 112

Table C.18: Test D2 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1012 0.1944 0.1950 26707.1 1166.1
0.1 0.1000 0.1968 0.1987 26928.0 1180.8 0.6181
0.2 0.0985 0.1996 0.2030 27181.8 1197.8 0.6847
0.3 0.0970 0.2029 0.2080 27470.1 1217.3 0.7432
0.4 0.0952 0.2066 0.2138 27795.1 1239.6 0.7930
0.5 0.0933 0.2108 0.2203 28159.2 1264.9 0.8334
0.6 0.0913 0.2156 0.2278 28565.0 1293.5 0.8640
0.7 0.0890 0.2210 0.2364 29016.0 1325.9 0.8843
0.8 0.0866 0.2271 0.2463 29515.7 1362.4 0.8939
0.9 0.0841 0.2339 0.2576 30068.6 1403.7 0.8928
1.0 0.0814 0.2417 0.2705 30679.7 1450.4 0.8806
1.1 0.0785 0.2506 0.2855 31355.0 1503.5 0.8574
1.2 0.0755 0.2607 0.3029 32101.2 1563.9 0.8232
1.3 0.0723 0.2722 0.3232 32926.6 1633.1 0.7782
1.4 0.0689 0.2854 0.3471 33840.6 1712.6 0.7226
1.5 0.0654 0.3008 0.3755 34854.8 1804.8 0.6568
1.6 0.0617 0.3187 0.4096 35982.6 1912.3 0.5813
1.7 0.0579 0.3399 0.4510 37240.5 2039.3 0.4965
1.8 0.0539 0.3651 0.5022 38648.5 2190.8 0.4030

Table C.19: Test D3 hup1


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0556 0.0866 0.1172 9308.1 519.5
0.1 0.0549 0.0878 0.1196 9374.0 526.6 2.4765
0.2 0.0541 0.0890 0.1222 9442.4 534.2 2.4566
0.3 0.0533 0.0903 0.1249 9513.3 542.1 2.4326
0.4 0.0525 0.0917 0.1278 9586.8 550.4 2.4046
0.5 0.0517 0.0932 0.1309 9663.0 559.2 2.3727
0.6 0.0508 0.0947 0.1342 9742.1 568.4 2.3368
0.7 0.0500 0.0964 0.1376 9824.1 578.2 2.2971
0.8 0.0491 0.0981 0.1413 9909.1 588.5 2.2535
0.9 0.0482 0.0999 0.1453 9997.3 599.4 2.2061
1.0 0.0473 0.1018 0.1495 10088.8 610.9 2.1551
1.1 0.0464 0.1039 0.1540 10183.7 623.1 2.1005
1.2 0.0454 0.1060 0.1588 10282.3 636.0 2.0424
1.3 0.0445 0.1083 0.1639 10384.5 649.8 1.9808
1.4 0.0435 0.1107 0.1695 10490.7 664.3 1.9159
1.5 0.0425 0.1133 0.1755 10601.0 679.9 1.8478
1.6 0.0415 0.1161 0.1819 10715.5 696.4 1.7766
1.7 0.0405 0.1190 0.1889 10834.5 714.1 1.7023
1.8 0.0394 0.1222 0.1964 10958.2 733.0 1.6252
APPENDIX C. RESULTS 113

Table C.20: Test D3 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0744 0.1304 0.1526 15962.7 782.3
0.1 0.0734 0.1322 0.1558 16090.6 793.1 1.3984
0.2 0.0723 0.1342 0.1593 16232.1 805.0 1.4810
0.3 0.0711 0.1364 0.1633 16387.9 818.4 1.5516
0.4 0.0698 0.1389 0.1678 16558.7 833.3 1.6097
0.5 0.0685 0.1416 0.1728 16745.4 849.8 1.6549
0.6 0.0670 0.1447 0.1784 16949.0 868.2 1.6865
0.7 0.0655 0.1481 0.1847 17170.4 888.5 1.7045
0.8 0.0639 0.1519 0.1918 17411.0 911.1 1.7083
0.9 0.0622 0.1560 0.1998 17672.2 936.2 1.6981
1.0 0.0604 0.1607 0.2088 17955.4 964.1 1.6736
1.1 0.0585 0.1659 0.2190 18262.4 995.3 1.6349
1.2 0.0565 0.1717 0.2306 18595.3 1030.1 1.5822
1.3 0.0544 0.1782 0.2439 18956.2 1069.2 1.5158
1.4 0.0523 0.1856 0.2591 19347.9 1113.3 1.4361
1.5 0.0500 0.1939 0.2767 19773.2 1163.2 1.3434
1.6 0.0477 0.2033 0.2973 20235.5 1220.1 1.2384
1.7 0.0453 0.2142 0.3214 20738.8 1285.3 1.1217
1.8 0.0428 0.2268 0.3501 21287.7 1360.6 0.9941

Table C.21: Test D3 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0963 0.1588 0.1633 21447.0 952.6
0.1 0.0950 0.1609 0.1666 21633.8 965.5 1.1987
0.2 0.0936 0.1634 0.1705 21845.9 980.2 1.3035
0.3 0.0920 0.1662 0.1749 22084.9 997.0 1.3946
0.4 0.0903 0.1693 0.1799 22352.2 1016.0 1.4709
0.5 0.0885 0.1729 0.1856 22649.9 1037.5 1.5317
0.6 0.0864 0.1769 0.1922 22980.1 1061.7 1.5761
0.7 0.0843 0.1815 0.1996 23345.3 1089.0 1.6037
0.8 0.0820 0.1866 0.2081 23748.4 1119.7 1.6141
0.9 0.0795 0.1924 0.2179 24192.8 1154.4 1.6069
1.0 0.0769 0.1989 0.2290 24682.4 1193.5 1.5821
1.1 0.0741 0.2063 0.2419 25221.6 1237.9 1.5397
1.2 0.0712 0.2147 0.2569 25815.9 1288.4 1.4799
1.3 0.0682 0.2243 0.2743 26471.2 1346.0 1.4032
1.4 0.0650 0.2354 0.2948 27194.9 1412.3 1.3101
1.5 0.0616 0.2482 0.3191 27995.6 1488.9 1.2015
1.6 0.0581 0.2631 0.3483 28883.5 1578.4 1.0781
1.7 0.0545 0.2806 0.3838 29870.9 1683.8 0.9411
1.8 0.0507 0.3016 0.4276 30972.6 1809.6 0.7915
Table C.22: Series C Results
hN1 hN2
3 −1 3 −1
Test No. φ Q( m s ) Fr Rec Rep Cd Q( m s ) Fr Rec Rep Cd

α=1 C1 0.014 0.001190 0.64725 28434 2166 0.2478 0.001954 0.51034 37689 2181 0.4765
C2 0.027 0.000976 0.53091 23323 1776 0.3452 0.001305 0.34087 25173 1457 0.7958
C3 0.041 0.000882 0.47974 21075 1605 0.3315 0.001084 0.28295 20896 1209 0.8334
C4 0.054 0.000654 0.35570 15626 1190 0.5880 0.001062 0.27740 20486 1185 0.6546
α = 2◦ C5 0.014 0.001379 0.75020 32956 2510 0.6932 0.002063 0.53883 39793 2302 1.1899
C6 0.027 0.000971 0.52799 23195 1767 1.0371 0.001454 0.37967 28039 1622 1.4121
APPENDIX C. RESULTS

C7 0.041 0.000688 0.37409 16434 1252 1.5965 0.001332 0.34773 25680 1486 1.1493
C8 0.054 0.000792 0.43105 18936 1442 0.8609 0.001088 0.28407 20979 1214 1.3442

Table C.23: Series D Results


hN1 hN2
Test No. φ Q( m3 s−1 ) Fr Rec Rep Cd Q( m3 s−1 ) Fr Rec Rep Cd
alpha 1 D4 0.027 0.001236 0.67207 29524 2249 0.0909 0.001973 0.51517 38045 2201 0.2299
D5 0.041 0.000987 0.53694 23588 1797 0.2201 0.001448 0.37809 27923 1616 0.4049
D6 0.034 0.000675 0.36699 16122 1228 0.8677 0.001175 0.30684 22660 1311 0.8252
alpha 2 D7 0.027 0.001359 0.73944 32484 2474 0.3664 0.002082 0.54366 40150 2323 0.5807
D8 0.041 0.000995 0.54130 23779 1811 0.6471 0.001610 0.42036 31044 1796 0.7421
D9 0.034 0.000834 0.45360 19927 1518 1.2183 0.001384 0.36135 26686 1544 1.2647
114
APPENDIX C. RESULTS 115

Table C.24: Test E1 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0845 0.2921 0.3207 37616.3 1460.4
0.1 0.0839 0.2942 0.3242 37783.9 1470.9 0.1549
0.2 0.0832 0.2968 0.3285 37987.3 1483.8 0.2447
0.3 0.0824 0.2998 0.3336 38227.7 1499.1 0.3277
0.4 0.0814 0.3034 0.3395 38506.3 1517.0 0.4030
0.5 0.0803 0.3075 0.3465 38824.9 1537.7 0.4698
0.6 0.0791 0.3123 0.3545 39185.3 1561.4 0.5270
0.7 0.0777 0.3177 0.3637 39589.7 1588.3 0.5741
0.8 0.0763 0.3237 0.3742 40040.7 1618.7 0.6101
0.9 0.0747 0.3306 0.3862 40541.1 1653.0 0.6346
1.0 0.0730 0.3383 0.3998 41094.4 1691.6 0.6468
1.1 0.0712 0.3470 0.4153 41704.2 1735.1 0.6463
1.2 0.0692 0.3568 0.4330 42375.2 1784.0 0.6326
1.3 0.0671 0.3678 0.4532 43112.2 1839.0 0.6052
1.4 0.0649 0.3802 0.4764 43921.2 1901.2 0.5639
1.5 0.0626 0.3943 0.5030 44808.9 1971.5 0.5084
1.6 0.0602 0.4103 0.5339 45783.0 2051.4 0.4384
1.7 0.0576 0.4285 0.5699 46852.6 2142.5 0.3536
1.8 0.0549 0.4494 0.6122 48028.2 2247.2 0.2540

Table C.25: Test E1 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1042 0.3235 0.3200 44897.5 1617.7
0.1 0.1034 0.3260 0.3236 45122.8 1629.9 0.2200
0.2 0.1025 0.3290 0.3281 45400.4 1645.1 0.3226
0.3 0.1014 0.3327 0.3336 45732.3 1663.3 0.4172
0.4 0.1001 0.3369 0.3401 46120.6 1684.7 0.5025
0.5 0.0986 0.3419 0.3476 46568.2 1709.7 0.5775
0.6 0.0970 0.3477 0.3565 47078.4 1738.5 0.6410
0.7 0.0952 0.3543 0.3666 47655.0 1771.4 0.6922
0.8 0.0932 0.3618 0.3783 48302.6 1808.9 0.7300
0.9 0.0911 0.3703 0.3918 49026.3 1851.5 0.7537
1.0 0.0888 0.3799 0.4072 49832.3 1899.7 0.7626
1.1 0.0863 0.3909 0.4249 50727.7 1954.4 0.7560
1.2 0.0836 0.4033 0.4453 51720.7 2016.5 0.7336
1.3 0.0808 0.4174 0.4689 52821.2 2087.0 0.6948
1.4 0.0778 0.4335 0.4962 54040.3 2167.4 0.6393
1.5 0.0746 0.4519 0.5281 55391.5 2259.3 0.5670
1.6 0.0713 0.4730 0.5656 56890.6 2365.1 0.4776
1.7 0.0678 0.4975 0.6101 58556.3 2487.5 0.3709
1.8 0.0641 0.5260 0.6633 60410.9 2630.2 0.2468
APPENDIX C. RESULTS 116

Table C.26: Test E1 hup4


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1305 0.3538 0.3127 52616.8 1768.8
0.1 0.1294 0.3566 0.3165 52922.7 1783.2 0.3257
0.2 0.1282 0.3602 0.3212 53296.5 1800.9 0.4343
0.3 0.1267 0.3644 0.3269 53741.0 1822.1 0.5338
0.4 0.1250 0.3694 0.3336 54259.5 1847.0 0.6231
0.5 0.1231 0.3752 0.3415 54856.1 1875.9 0.7008
0.6 0.1209 0.3818 0.3506 55535.9 1909.1 0.7660
0.7 0.1186 0.3894 0.3611 56304.3 1947.1 0.8176
0.8 0.1160 0.3981 0.3732 57168.3 1990.4 0.8546
0.9 0.1132 0.4079 0.3871 58135.7 2039.6 0.8763
1.0 0.1102 0.4191 0.4031 59215.6 2095.3 0.8820
1.1 0.1069 0.4317 0.4215 60419.0 2158.6 0.8710
1.2 0.1035 0.4461 0.4427 61758.7 2230.5 0.8430
1.3 0.0998 0.4625 0.4674 63249.5 2312.4 0.7976
1.4 0.0959 0.4812 0.4960 64909.4 2406.0 0.7344
1.5 0.0918 0.5027 0.5296 66759.5 2513.4 0.6535
1.6 0.0875 0.5275 0.5693 68825.3 2637.4 0.5547
1.7 0.0830 0.5563 0.6165 71137.5 2781.5 0.4379
1.8 0.0782 0.5901 0.6736 73733.3 2950.5 0.3032

Table C.27: Test E1 hup5


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1682 0.3934 0.3063 62497.8 1966.9
0.1 0.1671 0.3959 0.3092 62800.4 1979.4 0.2434
0.2 0.1658 0.3991 0.3130 63190.8 1995.5 0.3693
0.3 0.1641 0.4031 0.3177 63672.0 2015.6 0.4869
0.4 0.1622 0.4079 0.3234 64248.1 2039.6 0.5948
0.5 0.1599 0.4136 0.3302 64924.3 2068.1 0.6916
0.6 0.1574 0.4202 0.3382 65706.3 2101.2 0.7759
0.7 0.1546 0.4279 0.3475 66601.6 2139.5 0.8465
0.8 0.1515 0.4367 0.3582 67618.6 2183.4 0.9023
0.9 0.1481 0.4467 0.3706 68767.4 2233.5 0.9421
1.0 0.1444 0.4581 0.3849 70060.0 2290.6 0.9653
1.1 0.1404 0.4711 0.4014 71510.7 2355.6 0.9709
1.2 0.1361 0.4859 0.4205 73136.1 2429.6 0.9584
1.3 0.1316 0.5028 0.4426 74956.4 2513.9 0.9273
1.4 0.1267 0.5221 0.4683 76995.2 2610.4 0.8773
1.5 0.1216 0.5442 0.4984 79281.4 2721.1 0.8081
1.6 0.1161 0.5697 0.5338 81849.6 2848.7 0.7198
1.7 0.1104 0.5994 0.5760 84742.1 2996.8 0.6123
1.8 0.1043 0.6340 0.6266 88010.6 3170.0 0.4857
APPENDIX C. RESULTS 117

Table C.28: Test E2 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0860 0.2560 0.2786 33186.9 1279.9
0.1 0.0852 0.2586 0.2829 33397.5 1292.9 0.4270
0.2 0.0842 0.2615 0.2878 33633.4 1307.7 0.4772
0.3 0.0832 0.2649 0.2932 33895.6 1324.3 0.5220
0.4 0.0820 0.2686 0.2994 34185.2 1342.8 0.5611
0.5 0.0808 0.2727 0.3063 34503.6 1363.3 0.5939
0.6 0.0795 0.2772 0.3140 34852.1 1386.1 0.6203
0.7 0.0780 0.2823 0.3226 35232.6 1411.3 0.6398
0.8 0.0765 0.2878 0.3322 35646.8 1439.2 0.6523
0.9 0.0749 0.2940 0.3429 36096.8 1470.0 0.6574
1.0 0.0732 0.3008 0.3549 36585.0 1504.0 0.6551
1.1 0.0714 0.3083 0.3683 37114.2 1541.6 0.6450
1.2 0.0696 0.3166 0.3833 37687.1 1583.2 0.6271
1.3 0.0676 0.3259 0.4002 38307.4 1629.3 0.6012
1.4 0.0655 0.3361 0.4192 38978.7 1680.5 0.5674
1.5 0.0634 0.3475 0.4407 39705.3 1737.5 0.5254
1.6 0.0611 0.3602 0.4652 40492.3 1801.2 0.4753
1.7 0.0588 0.3745 0.4931 41345.3 1872.7 0.4171
1.8 0.0564 0.3907 0.5253 42270.5 1953.3 0.3508

Table C.29: Test E2 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1024 0.2757 0.2751 38042.1 1378.7
0.1 0.1014 0.2786 0.2793 38300.5 1392.9 0.4717
0.2 0.1002 0.2819 0.2843 38598.1 1409.3 0.5439
0.3 0.0989 0.2856 0.2900 38936.8 1428.2 0.6087
0.4 0.0974 0.2899 0.2966 39318.6 1449.7 0.6653
0.5 0.0958 0.2948 0.3041 39745.9 1474.0 0.7131
0.6 0.0941 0.3003 0.3126 40221.4 1501.3 0.7515
0.7 0.0922 0.3064 0.3222 40748.3 1532.1 0.7801
0.8 0.0901 0.3133 0.3332 41330.2 1566.6 0.7983
0.9 0.0880 0.3211 0.3456 41971.3 1605.3 0.8057
1.0 0.0857 0.3297 0.3597 42676.2 1648.6 0.8021
1.1 0.0832 0.3394 0.3757 43450.6 1697.2 0.7871
1.2 0.0806 0.3504 0.3940 44300.7 1751.8 0.7606
1.3 0.0779 0.3627 0.4149 45233.9 1813.4 0.7224
1.4 0.0750 0.3766 0.4390 46258.6 1882.9 0.6725
1.5 0.0720 0.3924 0.4669 47384.6 1961.9 0.6109
1.6 0.0688 0.4104 0.4994 48623.6 2051.9 0.5376
1.7 0.0655 0.4311 0.5376 49989.1 2155.3 0.4527
1.8 0.0621 0.4550 0.5830 51497.2 2274.8 0.3563
APPENDIX C. RESULTS 118

Table C.30: Test E2 hup4


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1349 0.3160 0.2747 47440.4 1579.9
0.1 0.1335 0.3193 0.2790 47802.9 1596.7 0.5234
0.2 0.1319 0.3233 0.2842 48225.4 1616.4 0.6055
0.3 0.1300 0.3278 0.2903 48711.2 1639.2 0.6789
0.4 0.1280 0.3331 0.2972 49263.7 1665.3 0.7428
0.5 0.1257 0.3390 0.3052 49887.2 1695.1 0.7963
0.6 0.1233 0.3458 0.3144 50586.7 1728.9 0.8388
0.7 0.1206 0.3534 0.3249 51367.9 1767.1 0.8696
0.8 0.1178 0.3621 0.3369 52237.7 1810.3 0.8881
0.9 0.1147 0.3718 0.3505 53203.7 1858.9 0.8939
1.0 0.1114 0.3828 0.3662 54275.2 1913.8 0.8867
1.1 0.1079 0.3952 0.3841 55462.9 1975.9 0.8661
1.2 0.1042 0.4092 0.4048 56779.2 2046.2 0.8320
1.3 0.1003 0.4252 0.4288 58239.1 2126.1 0.7843
1.4 0.0961 0.4435 0.4566 59860.0 2217.3 0.7231
1.5 0.0918 0.4644 0.4893 61662.8 2321.9 0.6483
1.6 0.0873 0.4885 0.5280 63672.6 2442.6 0.5602
1.7 0.0825 0.5166 0.5742 65919.7 2583.2 0.4590
1.8 0.0776 0.5496 0.6301 68440.9 2748.2 0.3448

Table C.31: Test E2 hup5


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1678 0.3467 0.2702 55047.7 1733.3
0.1 0.1659 0.3507 0.2749 55537.2 1753.5 0.6369
0.2 0.1637 0.3554 0.2805 56103.3 1777.1 0.7186
0.3 0.1612 0.3608 0.2869 56750.3 1804.2 0.7907
0.4 0.1585 0.3670 0.2943 57483.5 1835.2 0.8522
0.5 0.1555 0.3741 0.3029 58309.1 1870.3 0.9023
0.6 0.1523 0.3820 0.3126 59234.3 1910.1 0.9405
0.7 0.1488 0.3910 0.3237 60267.5 1955.1 0.9661
0.8 0.1450 0.4012 0.3363 61418.5 2005.8 0.9786
0.9 0.1410 0.4126 0.3508 62698.9 2063.1 0.9775
1.0 0.1367 0.4255 0.3675 64122.2 2127.7 0.9626
1.1 0.1322 0.4402 0.3866 65704.7 2200.9 0.9337
1.2 0.1274 0.4568 0.4087 67465.1 2283.9 0.8907
1.3 0.1223 0.4757 0.4343 69426.1 2378.6 0.8335
1.4 0.1170 0.4974 0.4643 71614.9 2486.9 0.7624
1.5 0.1114 0.5223 0.4997 74064.1 2611.7 0.6775
1.6 0.1055 0.5513 0.5418 76813.8 2756.4 0.5791
1.7 0.0994 0.5852 0.5925 79912.9 2925.8 0.4674
1.8 0.0930 0.6252 0.6544 83422.2 3126.0 0.3427
APPENDIX C. RESULTS 119

Table C.32: Test E3 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0820 0.2920 0.3256 37172.5 1460.2
0.1 0.0812 0.2950 0.3305 37401.7 1474.9 0.1622
0.2 0.0803 0.2984 0.3362 37661.7 1491.8 0.1923
0.3 0.0792 0.3022 0.3427 37953.8 1511.0 0.2189
0.4 0.0781 0.3065 0.3501 38279.3 1532.6 0.2418
0.5 0.0769 0.3114 0.3584 38639.9 1556.8 0.2608
0.6 0.0756 0.3168 0.3678 39037.4 1583.8 0.2756
0.7 0.0742 0.3228 0.3783 39473.9 1613.9 0.2860
0.8 0.0727 0.3295 0.3902 39951.8 1647.4 0.2917
0.9 0.0711 0.3369 0.4035 40473.7 1684.7 0.2926
1.0 0.0694 0.3452 0.4185 41042.6 1726.1 0.2885
1.1 0.0676 0.3544 0.4354 41662.0 1772.2 0.2793
1.2 0.0657 0.3647 0.4544 42335.7 1823.6 0.2648
1.3 0.0637 0.3762 0.4760 43068.1 1880.9 0.2450
1.4 0.0616 0.3890 0.5006 43864.2 1945.0 0.2197
1.5 0.0594 0.4034 0.5286 44729.9 2017.0 0.1888
1.6 0.0571 0.4196 0.5608 45671.4 2098.2 0.1523
1.7 0.0547 0.4380 0.5981 46696.5 2190.1 0.1101
1.8 0.0522 0.4589 0.6414 47813.8 2294.7 0.0620

Table C.33: Test E3 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1058 0.3316 0.3255 46235.2 1657.8
0.1 0.1047 0.3351 0.3307 46563.3 1675.4 0.2036
0.2 0.1034 0.3392 0.3367 46939.5 1695.8 0.2382
0.3 0.1020 0.3438 0.3437 47366.1 1719.0 0.2689
0.4 0.1005 0.3491 0.3516 47845.8 1745.5 0.2951
0.5 0.0988 0.3551 0.3607 48381.4 1775.3 0.3166
0.6 0.0969 0.3618 0.3710 48976.5 1808.8 0.3330
0.7 0.0950 0.3693 0.3826 49635.0 1846.5 0.3441
0.8 0.0928 0.3777 0.3958 50361.5 1888.6 0.3497
0.9 0.0906 0.3872 0.4107 51161.3 1935.8 0.3495
1.0 0.0882 0.3977 0.4277 52040.2 1988.7 0.3433
1.1 0.0856 0.4096 0.4469 53005.4 2048.0 0.3309
1.2 0.0829 0.4229 0.4689 54064.6 2114.6 0.3123
1.3 0.0801 0.4379 0.4941 55227.3 2189.6 0.2874
1.4 0.0771 0.4549 0.5230 56504.0 2274.3 0.2560
1.5 0.0740 0.4741 0.5566 57907.3 2370.5 0.2181
1.6 0.0707 0.4961 0.5956 59451.8 2480.3 0.1737
1.7 0.0673 0.5212 0.6416 61154.8 2606.2 0.1226
1.8 0.0637 0.5504 0.6961 63036.5 2751.9 0.0648
APPENDIX C. RESULTS 120

Table C.34: Test E3 hup4


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1310 0.3612 0.3187 53780.1 1806.0
0.1 0.1294 0.3655 0.3244 54236.5 1827.5 0.2802
0.2 0.1277 0.3704 0.3309 54753.7 1852.0 0.3130
0.3 0.1258 0.3760 0.3384 55335.0 1879.8 0.3414
0.4 0.1238 0.3822 0.3469 55984.4 1911.2 0.3650
0.5 0.1215 0.3893 0.3565 56706.1 1946.3 0.3835
0.6 0.1191 0.3971 0.3674 57505.2 1985.7 0.3967
0.7 0.1165 0.4060 0.3797 58387.5 2029.8 0.4044
0.8 0.1138 0.4158 0.3936 59359.7 2079.0 0.4062
0.9 0.1108 0.4268 0.4093 60429.5 2134.1 0.4020
1.0 0.1077 0.4391 0.4272 61605.7 2195.7 0.3917
1.1 0.1045 0.4529 0.4475 62898.6 2264.7 0.3752
1.2 0.1010 0.4684 0.4706 64320.3 2342.2 0.3524
1.3 0.0974 0.4859 0.4972 65884.5 2429.6 0.3231
1.4 0.0936 0.5057 0.5279 67607.6 2528.5 0.2875
1.5 0.0896 0.5282 0.5634 69508.8 2640.9 0.2454
1.6 0.0854 0.5539 0.6051 71610.7 2769.3 0.1969
1.7 0.0811 0.5834 0.6542 73940.6 2917.2 0.1419
1.8 0.0766 0.6178 0.7127 76530.8 3088.8 0.0803

Table C.35: Test E3 hup5


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1658 0.3988 0.3127 63148.6 1994.0
0.1 0.1635 0.4044 0.3193 63820.8 2022.0 0.3871
0.2 0.1611 0.4106 0.3267 64564.3 2053.1 0.4092
0.3 0.1584 0.4175 0.3350 65383.3 2087.7 0.4270
0.4 0.1555 0.4252 0.3442 66283.2 2126.1 0.4402
0.5 0.1525 0.4337 0.3546 67269.7 2168.5 0.4489
0.6 0.1493 0.4431 0.3662 68349.6 2215.4 0.4527
0.7 0.1458 0.4535 0.3791 69530.4 2267.3 0.4515
0.8 0.1422 0.4650 0.3936 70820.9 2324.8 0.4453
0.9 0.1384 0.4777 0.4099 72231.2 2388.5 0.4339
1.0 0.1345 0.4918 0.4282 73772.8 2459.2 0.4173
1.1 0.1303 0.5076 0.4489 75459.1 2537.8 0.3955
1.2 0.1259 0.5251 0.4724 77305.5 2625.6 0.3684
1.3 0.1214 0.5448 0.4992 79330.1 2723.8 0.3360
1.4 0.1167 0.5668 0.5298 81554.0 2834.2 0.2984
1.5 0.1118 0.5918 0.5652 84002.2 2958.8 0.2556
1.6 0.1067 0.6201 0.6062 86704.0 3100.4 0.2075
1.7 0.1014 0.6524 0.6543 89694.9 3262.2 0.1544
1.8 0.0959 0.6897 0.7111 93017.0 3448.6 0.0960
APPENDIX C. RESULTS 121

Table C.36: Test E4 hup2


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.0815 0.1989 0.2224 25262.5 994.5
0.1 0.0807 0.2010 0.2259 25425.2 1005.0 0.5100
0.2 0.0797 0.2034 0.2300 25609.6 1017.1 0.5682
0.3 0.0787 0.2061 0.2346 25816.7 1030.7 0.6204
0.4 0.0775 0.2092 0.2399 26047.5 1046.1 0.6661
0.5 0.0763 0.2127 0.2459 26303.1 1063.4 0.7050
0.6 0.0749 0.2165 0.2526 26584.9 1082.7 0.7367
0.7 0.0734 0.2208 0.2602 26894.5 1104.2 0.7607
0.8 0.0719 0.2256 0.2687 27233.6 1128.2 0.7768
0.9 0.0702 0.2310 0.2783 27604.2 1154.9 0.7847
1.0 0.0685 0.2369 0.2891 28008.5 1184.7 0.7843
1.1 0.0666 0.2436 0.3014 28449.0 1217.9 0.7753
1.2 0.0646 0.2510 0.3152 28928.6 1254.9 0.7577
1.3 0.0626 0.2593 0.3310 29450.6 1296.4 0.7314
1.4 0.0604 0.2686 0.3490 30018.8 1343.0 0.6964
1.5 0.0581 0.2791 0.3696 30637.5 1395.4 0.6529
1.6 0.0557 0.2909 0.3934 31311.5 1454.7 0.6008
1.7 0.0533 0.3044 0.4211 32046.5 1522.2 0.5403
1.8 0.0507 0.3199 0.4536 32849.2 1599.5 0.4716

Table C.37: Test E4 hup3


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1086 0.2233 0.2164 31410.8 1116.6
0.1 0.1073 0.2261 0.2204 31675.4 1130.6 0.6920
0.2 0.1058 0.2292 0.2250 31967.8 1146.1 0.7353
0.3 0.1042 0.2327 0.2301 32289.1 1163.4 0.7728
0.4 0.1025 0.2365 0.2358 32641.2 1182.5 0.8041
0.5 0.1007 0.2407 0.2421 33025.8 1203.6 0.8289
0.6 0.0988 0.2454 0.2492 33445.0 1227.0 0.8470
0.7 0.0968 0.2505 0.2571 33900.9 1252.7 0.8582
0.8 0.0947 0.2562 0.2659 34396.3 1281.0 0.8624
0.9 0.0924 0.2625 0.2757 34934.1 1312.3 0.8594
1.0 0.0900 0.2694 0.2866 35517.6 1346.9 0.8491
1.1 0.0875 0.2770 0.2989 36150.5 1385.2 0.8315
1.2 0.0849 0.2855 0.3128 36837.2 1427.6 0.8067
1.3 0.0822 0.2949 0.3284 37582.5 1474.7 0.7745
1.4 0.0794 0.3055 0.3461 38392.0 1527.3 0.7352
1.5 0.0765 0.3172 0.3663 39272.1 1586.1 0.6888
1.6 0.0734 0.3304 0.3894 40230.2 1652.2 0.6355
1.7 0.0702 0.3454 0.4161 41274.9 1726.9 0.5755
1.8 0.0669 0.3624 0.4472 42416.1 1811.8 0.5089
APPENDIX C. RESULTS 122

Table C.38: Test E4 hup4


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1271 0.2387 0.2137 35230.8 1193.3
0.1 0.1257 0.2414 0.2174 35520.9 1207.2 0.6664
0.2 0.1240 0.2447 0.2218 35857.5 1223.4 0.7502
0.3 0.1221 0.2484 0.2270 36243.0 1242.2 0.8246
0.4 0.1201 0.2527 0.2329 36680.4 1263.6 0.8889
0.5 0.1178 0.2576 0.2397 37173.3 1288.1 0.9423
0.6 0.1153 0.2632 0.2475 37725.7 1315.9 0.9841
0.7 0.1126 0.2695 0.2564 38342.3 1347.3 1.0137
0.8 0.1097 0.2766 0.2666 39028.7 1382.9 1.0307
0.9 0.1066 0.2846 0.2783 39791.4 1423.1 1.0348
1.0 0.1033 0.2937 0.2918 40638.0 1468.6 1.0256
1.1 0.0998 0.3040 0.3073 41577.4 1520.2 1.0030
1.2 0.0961 0.3158 0.3253 42619.9 1579.0 0.9670
1.3 0.0922 0.3292 0.3462 43778.2 1646.1 0.9178
1.4 0.0880 0.3446 0.3708 45066.9 1723.2 0.8555
1.5 0.0837 0.3625 0.4000 46503.8 1812.3 0.7806
1.6 0.0792 0.3832 0.4348 48110.2 1916.0 0.6934
1.7 0.0744 0.4076 0.4769 49912.1 2037.9 0.5945
1.8 0.0695 0.4365 0.5287 51941.7 2182.7 0.4845

Table C.39: Test E4 hup5


x (m) z (m) u( m s−1 ) Fr Rec Rep Cd
0.0 0.1646 0.2648 0.2084 41853.3 1324.0
0.1 0.1624 0.2683 0.2126 42278.1 1341.7 0.8396
0.2 0.1600 0.2724 0.2174 42759.7 1362.0 0.9099
0.3 0.1573 0.2770 0.2230 43301.8 1385.0 0.9706
0.4 0.1544 0.2822 0.2293 43908.6 1410.9 1.0211
0.5 0.1513 0.2880 0.2364 44585.1 1440.1 1.0608
0.6 0.1479 0.2946 0.2445 45336.9 1472.9 1.0892
0.7 0.1443 0.3020 0.2538 46170.6 1509.8 1.1060
0.8 0.1405 0.3102 0.2643 47094.0 1551.1 1.1108
0.9 0.1364 0.3195 0.2762 48116.1 1597.6 1.1035
1.0 0.1321 0.3300 0.2899 49247.4 1650.0 1.0839
1.1 0.1275 0.3418 0.3057 50500.4 1709.1 1.0521
1.2 0.1227 0.3552 0.3238 51889.7 1776.1 1.0081
1.3 0.1176 0.3704 0.3448 53432.9 1852.2 0.9521
1.4 0.1124 0.3879 0.3695 55150.7 1939.3 0.8846
1.5 0.1068 0.4079 0.3985 57068.4 2039.6 0.8058
1.6 0.1011 0.4312 0.4330 59216.7 2155.9 0.7163
1.7 0.0951 0.4584 0.4747 61633.1 2292.0 0.6167
1.8 0.0888 0.4906 0.5256 64364.1 2453.0 0.5076

You might also like