THE IMPACT OF ICE CONDITIONS ON LOCAL SCOUR AROUND BRIDGE
PIERS
by

Faye Hirshfield
B.Sc., University of Northern British Columbia, 2008
M.Sc., University of Northern British Columbia, 2010

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR
THE DEGREE OF
DOCTOR OF PHILOSOPHY
IN
NATURAL RESOURCES AND ENVIRONMENTAL STUDIES

THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA
April 2015

© Faye Hirshfield

APPROVAL
Name:

Faye Hirshfield

Degree:

Doctor of Philosophy in Natural Resources and Environmental Studies

Title:

THE IMPACT OF ICE CONDITIONS ON LOCAL SCOUR AROUND BRIDGE PIERS

Examining Committee:

Chair: Dr. Kevin Smith
University of Northern British Columbia

/3^

Supervisor: Dr. Jueyi Sui
University of Northern British Columbia

Committee Member. Dr. Jianbing Li
University of Northern British Columbia

Committee Member Dr. Youmin ng
University of Northern British Columbia

University of Northern British Columbia

em|^r: Dr. Bryan Kamey
Commi
University of Tordrito

External Examiner: DrkSamuel Li
Concordia University

Date Approved:

April 24, 2015

ABSTRACT

Ice cover imposes an additional boundary layer on a river channel and can influence sediment
transport around bridge structures. The main objective of this study was to gain a better
understanding of how ice cover is related to local scour around bridge piers. A set of flume

experiments were completed investigating local scour around bridge piers under open channel,

smooth ice and rough ice cover conditions. Three different non-uniform sediments were used
with Dso's of 0.47, 0.50 and 0.58 mm. The location of the average maximum velocity under
rough ice cover ranged from 0.36-0.43H and for smooth cover was 0.41 H, with depth (H)

measured from the channel bed to water surface. Turbulent intensity was greater under ice
cover than open channel conditions. Local pier scour under rough and smooth ice cover was

on average 37 and 20 percent greater than open channel scour depth respectively. The
maximum scour depth always occurred at the pier face. Greater pier scour under ice cover is
related to larger streamwise and downward flow velocities at the pier face under rough ice

cover. The scour hole velocity for rough and smooth ice cover was 33 and 1 5 percent greater
than open channel scour hole velocity respectively. Under all channel covers an armour layer
formed in the scour hole. For all experiments, as the armour layer size increased, the maximum
local scour depth decreased.

A critical assessment of current bridge research and construction was also conducted in order
to address how this study contributes to modem day bridge design. It was found that this study

addresses current knowledge gaps in bridge manual used in North America. Specifically, study
results concerning scour hole depth under ice cover and scour hole flow fields under ice cover

are important contributions to the field of bridge hydraulics.

ii

Table of Contents

ABSTRACT

”

LIST OF TABLES

vi

LIST OF FIGURES

viii

GLOSSARY

xii

ACKNOWLEDGEMENT

xiv

1.0 INTRODUCTION

1

1. 1 Background

1

1.2 Statement of the problem and research objectives

2

1.3 Significance of research

4

2.0 LITERATURE REVIEW

6

2. 1 Types of scour

6

2.2 Sediment transport around bridge piers

7

2.3 Incipient motion of sediment particles

11

2.4 Incipient motion of sediment under ice conditions

15

2.5 Summary and conclusions

17

3.0 EXPERIMENTAL SETUP: FLUME CONSTRUCTION

18

3. 1 Background

18

3.2 Flume Construction

18

3.3 Experimental runs

23

3.4 Experimental simulation time

26

3.5 Flume sediment

29

3.6 Flume scaling and pier selection

34
iii

3.7 Channel cover

37

3.8 Measuring flow velocity

38

3.9 Scour hole measurements

41

3.10 Armour layer sampling

41

3. 11 Flume operation procedures

44

3.12 Flume data and error analysis

44

3. 13 Summary and conclusions

46

4.0 RESULTS & DISCUSSION: VELOCITY AND TURBULENT FLOW FIELDS
47
UNDER OPEN AND ICE COVERED CHANNEL

4. 1 Experimental conditions

47

4.2 Flume velocity profiles

49

4.3 Flume turbulent intensity

54

4.4 Reynolds stresses

58

4.5 Summary and conclusions

62

5.0 RESULTS & DISCUSSION: SCOUR CHARACTERISTICS UNDER ICE COVER
64

5. 1 Scour depth versus sediment grain size

64

5.2 Scour depth versus pier size

68

5.3 Scour depth versus channel cover

71

5.4 Scour area and volume

78

5.5 Scour patterns under open, smooth and rough ice cover

82

5.6 Scour profiles

90

5. 7 Armour layer analysis

94

5.8 Scour hole velocity profiles

99

iv

I
5.9 Dimensional Analysis

1 06

5.10 Correlation of scour depth versus the Froude number

109

5.11 Correlation of scour depth versus shear stress - incipient motion

115

5.12 Multiple Regression Analysis

1 19

5.13 Summary and conclusions

133

6.0 CRITICAL ASSESSMENT OF BRIDGE RESEARCH AND DESIGN

135

6.0 Introduction

135

6.1 The use of flume experiments and their relation to field conditions

1 36

6.1. 1 Similitude theory and scaling

136

6.1.2 Scour equation development and field data

143

6.2 The use of CFD as it relates to bridge construction

148

6.2.1 Critical assessment of FLUENT for use in this thesis

150

6.2.2 CFD programs used for hydraulic design of bridges

1 53

6.2.3 Success and limitations of CFD

154

6.3 Bridges in Canada

1 57

6.3.1 Governing bodies and financial management

157

6.3.2 Construction standards

160

6.3.3 Bridge maintenance

166

6.4 Bridges in the United States

167

6.4.1 Governing bodies

167

6.4.2 Construction standards

170

6.4.3 Bridge maintenance

173

6.5 In situ scour measurement technology and countermeasures

174

6.6 Evaluation of this study in the context of current day bridge design

177

6.7 Conclusion and recommendations

182

v

i

7.0 CONCLUSIONS AND FUTURE RESEARCH

185

7.1 Thesis conclusions

185

7.2 Study limitations

1 88

7.3 Strengths of study and contributions to science

190

7.4 Future work

192

8.0 REFERENCES

194

APPENDIX A - FLUME DATA

206

APPENDIX B - SCOUR HOLE DATA

214

APPENDIX C - ARMOUR LAYER DATA

21 5

APPENDIX D - REGRESSION ANALYSIS

217

vi

LIST OF TABLES
Table 1. Measured velocity and stage resulting from various valve and tailgate
configurations. Configuration one and four were chosen for experimental conditions. Note:
testing was also undertaken using three tailgates (0.30 m) and velocity was not sufficient for
incipient motion to occur. In addition under the two tailgate configuration the small pier was
always placed in the upstream sandbox since under such channel conditions in the
25
downstream window no scour/incipient motion occurred
25
schedule
for
one
sediment
type
flume
experimental
Example
of
Table 2.
Table 3. Summary of literature standards regarding the definition of equilibrium scour
27
depth. Channel width is indicated by B
31
Table 4. Particle size distribution of sands used during flume experiments
uniformity
and
standard
deviation
(ag)
mean
(Dg»,
geometric
of
Summary
5.
Table
33
coefficient (Cu) for the three sands used in flume experiments
Table 6. Summary of literature values pertaining to geometric standard deviation of
33
sediments used in flume experiments
Table 7. Summary of flume scaling guidelines and comparison to scaling ratios calculated
from current study flume dimensions. Present flume ratios are on the left side of table with
36
literature guidelines on right side of table
48
Table 8. Summary of experimental conditions
52
Table 9. Maximum velocity and z/H values represented in Figure 23
Table 10. Roughness coefficients for channel material and ice cover. Associated equations
52
used in calculating the roughness coefficient can be found in Chapter 5, Section 5.12
75
Table 1 1. Scour depth values associated with Figure 35
76
Table 12. Scour depth values associated with Figure 36
77
Table 13. Scour depth values associated with Figure 37
93
Table 14. Slope of scour hole at upstream pier face for each run
Table 15. Average size of scour hole armour layer for each bed material and channel cover.
99
Table 16. Average velocity measured for each profile presented in Figure 56 and Figure 57.
In addition, the percent the velocity values are greater or less than open channel velocity are
presented. Data are missing due to ADV error for D5<r=0.47 mm under smooth cover
101
conditions
Table 17. Parameters associated with local pier scour
106
Table 18. Similarity parameters. Water Survey Canada data for March 4, 2014 was used to
calculate Froude and Reynolds numbers, WSC Station 08JC002, Isle Pierre
142
Table 19. Pier scour equations with associated author and study
144
Table 20. CSU calculated and measured maximum scour depths for open channel flow.
Measured scour depths are from open channel flow conditions for this study
148
Table 21. Bridge design flood frequencies as stipulated by the various Canadian Bridge
Design manuals. Provincial guidelines were taking from the TAC manual, Guide to Bridge
Hydraulics, with reference to Watt et al. (1989)
165
Table 22. Bridge design flood frequencies as stipulated in FHWA HEC-1 8 manual
(Zevenbergen et al., 2012)
172
Table 23. Measured experimental data for 54 flume runs
206

vi

Table 24. Measured experimental data for 54 flume runs. Note: for ADV Q data ‘nan’
208
indicates there was an error in the data measurement
Table 25. Measured maximum scour depth and various calculated parameters associated
210
experimental flume runs
with 54
for
212
54 flume experiments
Table 26. Calculated hydraulic parameters
Table 27. Scour depth values associated with Figure 29, and calculated percentages for open
214
channel flow
Table 28. Scour depth values associated with Figure 30, and calculated percentages for
2 14
smooth ice cover conditions
Table 29. Scour depth values associated with Figure 3 1 , and calculated percentages for rough
21 4
ice cover conditions
215
Table 30. Median size of armour layer in scour hole along with D50 of channel bed

vii

LIST OF FIGURES
9
Figure 1. Flow pattern around circular pier (taken from Melville & Coleman, 2000)
12
Figure 2. Overview of forces acting on a sediment particle
17
Hequ
cross
of.
of
section
reach
a
at
incurred
Changes
3.
Figure
Figure 4. (a) Construction of flume roof, (b) Viewing window and frame, (c) Concrete wall
removal and frame for viewing window, (d) Enclosed viewing platform, (e) Flow diffuser,
21
and (f) Flume head tank
22
Figure 5. Schematic overhead view of flume with exact construction dimensions
22
Figure 6. Side view of flume with exact dimensions
Figure 7. View of flume showing second valve configuration with white pipe entering flume
24
roof just upstream of weir, white arrow indicates white pipe
Figure 8. Aug 26, 2012, 2:32pm flume testing. Water ripples did not allow for down facing
28
photographs of the scour hole
Figure 9. August 27, 2012, 3:35 pm, side angle photograph taken during flume testing,
28
photograph inside scour hole was not practical due to flume design
30
Figure 10. Landscape sand sample that was used in flume
#
30
from
sandbox
1
being
shoveled
.
Sand
Figure 1 1
31
Figure 12. Particle size distribution of sediments used for flume experiments
35
Figure 13. 22 cm PVC pier installed in sandbox
Figure 14. A: Styrofoam floating around the pier inside flume. B: Styrofoam cubes attached
38
to create a rough ice cover
39
Figure 15. Diagram of 2D flow meter Sontek IQ (Sontek, 2014)
39
Figure 16. Sontek IQ mounted to flume floor
2014)
(Sontek,
40
Figure 17. Sontek 10 MHz 3-dimensional ADV
40
Figure 18. Use of Sontek 1 0-Mhz ADV to measure scour hole velocity under ice cover
hole
scour
40
velocity
inside
measuring
Figure 19. Sensor head of Sontek ADV
Figure 20. Numbered measuring points drawn on the outside of the small bridge pier (10 cm
width) and large bridge pier (22 cm width) for scour hole contour points of reference
41
Figure 21 . Differences in annour-layer thickness determined for the same deposit using
various prediction criteria
43
Figure 22. Example of armor layer in scour hole after experimental run # 1 , September 22,
2012
43
mm,
open
Figure 23. Velocity profiles for D50 = 0.58 mm, 0.50
0.47 mm under
channel,
smooth cover and rough cover channel conditions. Velocity profiles were measured with the
10 MHz ADV and presented under the two tailgate configuration (21-26 cm flow depth). .. 53
Figure 24. Spatially-averaged profiles of turbulent intensity for the streamwise (x) velocity
component, normalized by the shear velocity (U.), for open water and rough ice cover
conditions
55
Figure 25. Spatially-averaged profiles of turbulent intensity for the vertical (z) velocity
component, normalized by the shear velocity (U,), for open water and rough ice cover
56
Figure 26. Spatially-averaged profiles of turbulent intensity for the vertical (z) and
streamwise (x) velocity component, normalized by the shear velocity (U,), for open water
and smooth ice cover conditions
57
viii

Figure 27. Reynolds shear stress, normalized by the shear velocity, for open and rough cover
60
channel conditions
smooth
and
open
velocity,
for
stress,
by
normalized
shear
the
shear
Reynolds
Figure 28.
61
cover channel conditions
Figure 29. Maximum scour depth under open channel conditions for Dso=O.47 mm,
Dso=O.5O mm, Dso=O.58 mm. Numbers 1-6 indicate experimental run number. Associated
65
data values refer to Appendix B
mm,
Figure 30. Maximum scour depth under smooth ice channel conditions for D5o=O.47
D5o=O.5O mm, D5o=O.58 mm. Numbers 1-6 indicate experimental run number. Associated
66
data values refer to Appendix B
Figure 3 1 . Maximum scour depth under rough cover channel conditions for Dso=O.47 mm,
Dso=O.5O mm, D5o=O.58 mm. Numbers 1-6 indicate experimental run number. Associated
67
data values refer to Appendix B
width
cm
1
channel
for
and
conditions
1cm
open
22
Figure 32. Maximum scour depth under
69
pier. Numbers 1-6 indicate experimental run number
cm
22
channel
conditions
for
and
1cm
1
under
smooth
Figure 33. Maximum scour depth
70
width pier. Numbers 1-6 indicate experimental run number
width
cm
and
1cm
for
1
22
rough
channel
conditions
under
depth
scour
Figure 34. Maximum
70
pier. Numbers 1-6 indicate experimental run number
Figure 35. Maximum scour depth under open, smooth and rough channel condition for
75
D5o=O 58 mm. Numbers 1-6 indicate experimental run number
Figure 36. Maximum scour depth under open, smooth and rough channel condition for
76
Dso=0.50 mm. Numbers 1-6 indicate experimental run number
Figure 37. Maximum scour depth under open, smooth and rough channel condition for
77
Dso=O.47mm. Numbers 1-6 indicate experimental run number
Figure 38. Variation of scour volume and area around bridge pier under open channel
80
condition
Figure 39. Variation of scour volume and area around bridge pier under smooth ice cover
80
condition
Figure 40. Variation of scour volume and area around bridge pier under rough channel
condition
81
Figure 41. Maximum scour depth as related to the scour area and scour volume
81
Figure 42. Scour pattern for 11 cm and 22 cm pier in units of centimeters, D50 = 0.47 mm,
under open channel conditions
84
Figure 43. Scour patterns for 1 1 cm and 22 cm pier, D50 = 0.47 mm, under open channel
condition
85
Figure 44. Scour pattern for 1 1cm and 22 cm pier, D50 = 0.47 mm, under smooth channel
cover
86
Figure 45. Scour pattern for 11 cm and 22 cm pier, Dso = 0.47 mm, under smooth ice cover
channel condition
87
Figure 46. Scour pattern for 11 cm and 22 cm pier, D50 = 0.47 mm, under rough ice cover
channel condition
88
Figure 47. Scour pattern for 11 cm and 22 cm pier, D50 = 0.47 mm, under rough ice cover
condition
89

ix

Figure 48. Scour profile for 1 1 (left) and 22 (right) cm pier under open, smooth and rough
92
channel conditions for Dso=O.47 mm
rough
and
smooth
open,
(left)
under
pier
(right)
cm
for
and
profile
1
1
22
Figure 49. Scour
92
channel conditions for Dso=O.5O mm
rough
and
smooth
open,
(left)
and
under
(right)
pier
cm
for
profile
1
1
22
Figure 50. Scour
93
ice conditions for Dso= 0.58 mm
97
mm
0.58
layer
curve
for
and
related
distribution
armour
Example
of
Dso
.
Figure 5 1
=
97
Figure 52. Example of armour layer and related distribution curve for Dso = 0.50 mm
98
Figure 53. Example of armour layer and related distribution curve for Dso = 0.47 mm
98
Figure 54. Variation of scour hole armour layer size with median bed grain size
Figure 55. Variation of maximum scour depth and average scour hole armour layer size
99
under 1 1 and 22 cm pier width
Figure 56. Scour hole velocity profiles for the streamwise (Ux), lateral (Uy) and vertical
(Uz) velocity components under open, smooth and rough ice cover for 22 cm pier. All
profiles were measured under the two tailgate flume position with water depths from 22-24
104
cm
Figure 57. Scour hole velocity profiles for the streamwise (Ux), lateral (Uy) and vertical
(Uz) velocity components for 1 1 cm pier. All profiles were measured under the two tailgate
flume position with water depths from 22-24 cm. Smooth cover velocity profiles for
105
Dso=O.47 mm are missing due to ADV file error
Figure 58. Variation of maximum scour depth with Froude number for all experimental runs.
110
Figure 59. Variation of maximum scour depth with the Froude number under open and
111
rough cover for all three sediment sizes
Figure 60. Variation of maximum scour depth with Froude number for smooth channel
112
cover under various sediment sizes
Froude
Figure 61. Variation of dimensionless maximum scour depth with densimetric
1 14
number for all experiments
1 17
Figure 62. The variation of shear Reynolds number with dimensionless shear stress
Figure 63. Variation of maximum scour depth with dimensionless shear stress around 1 1 and
118
22 cm pier
Figure 64. Variation of maximum scour depth with dimensionless shear stress under rough
ice cover and open channel conditions for 11 cm pier
11 8
Figure 65. Variation of maximum scour depth with dimensionless shear stress under rough
ice cover and open channel conditions for 22 cm pier
119
Figure 66. Variation of scour depth for 11 cm pier under open channel conditions
124
Figure 67. Variation of scour depth for 22 cm pier under open channel conditions
124
Figure 68. Variation of scour depth in relation to Froude number, median bed sediment size
and pier width under open channel conditions
125
Figure 69. Variation of scour depth for 11 cm pier under ice covered conditions
129
Figure 70. Variation of scour depth for 22 cm pier under ice covered conditions
1 29
Figure 71. Variation of scour depth in relation to Froude number, median bed sediment size,
ice cover roughness and pier width
130
Figure 72. Variation of maximum scour depth in relation to Froude number, median
sediment size, pier size and armour layer under open channel conditions
132

x

Figure 73. Variation of maximum scour depth in relation to Froude number, median
1 33
sediment size, ice cover roughness, armour layer and pier size
depth,
d,
scour
is
where
piers
aligned
relationship
or
for
circular
scour
local
Figure 74. Basic
b is pier width and y is approach flow depth. Data represented in the above figure were
measured from uniform sediments. Data are also independent from sediment size effects as
163
b/D5o>50, (Melville, 1997; TAC, 2004)
176
Figure 75. SONAR scour monitoring equipment (NORTEK, 2014)
Figure 76. Variation of maximum scour depth for the 1 1 cm pier under open channel
21 7
conditions
Figure 77. Variation of maximum scour depth for the 22 cm pier under open channel
218
conditions
Figure 78. Variation of maximum scour depth for the 1 1 cm pier under ice covered
21 8
conditions
Figure 79. Variation of maximum scour depth for the 22 cm pier under open channel
219
conditions
219
Figure 80. Variation of maximum scour depth under ice covered conditions
220
Figure 81. Variation of maximum scour depth under open channel conditions
ice
covered
under
pier
1
1
depth
cm
for
the
Figure 82. Variation of maximum scour
220
conditions
Figure 83. Variation of maximum scour depth for the 1 1 cm pier under ice covered
conditions. Note ... the regression coefficient for D50 armour does not reflect results from this
221
study
Figure 84. Variation of maximum scour depth for the 1 1 cm pier under ice covered
22 1
conditions
Figure 85. Variation of maximum scour depth for the 22 cm pier under ice covered
conditions. Note the D50/H regression coefficient does not reflect results from this study. 222
Figure 86. Variation of maximum scour depth for the 22 cm pier under ice covered
conditions. Note the D50 armour regression coefficient does not reflect results from this
study
222
Figure 87. Variation of maximum scour depth for the 22 cm pier under ice covered
conditions. Note the D50/H regression coefficient does not reflect results from this study.. 223
Figure 88. Variation of maximum scour depth under ice covered conditions. Note the D50/H
and b/B regression coefficients do not reflect results from this study
223
Figure 89. Variation of maximum scour depth under ice covered conditions. Note the D50
armour regression coefficient does not reflect results of this study
224
Figure 90. Variation of maximum scour depth under ice covered conditions. Note the b/B
regression coefficient does not reflect results from this study
224
Figure 91. Variation of maximum scour depth under ice covered conditions. Note the D50
armour regression coefficient does not reflect results from this study
225

xi

GLOSSARY
Letters
b

pier width (cm)

B

channel width (cm)

Cu

coefficient of uniformity

dmu
D

maximum scour depth

diameter of sediment particle (m)

Dio

10th percentile particle diameter (mm)

d16

1 6th percentile particle diameter (mm)

D50
D50.

50th percentile particle diameter (mm)

D60

D90

60th percentile particle diameter (mm)
84th percentile particle diameter (mm)
90th percentile particle diameter (mm)

Dg

geometric mean

F, Fr

Froude number

Frc
Fo

critical Froude number

Fl
Fw

buoyant force

downward force

Fd

drag force

Fr

resistant force

g

gravitational acceleration (m/s2)

H, h

approach flow depth (cm)

nb

Manning’s roughness coefficient for channel bed

ni

Manning’s roughness coefficient for ice

Q

volumetric discharge (m3/s)

Ps
p

sediment density (kg/m3)

Dm

median sediment size of the armour layer (mm)

densimetric Froude number

fluid density (kg/m3)

xii

I

R«

Reynolds number

Rc*

shear Reynolds number

Rh
RMSx

hydraulic radius

streamwise turbulent intensity

RMSy

lateral turbulent intensity

RMSZ

vertical turbulent intensity

Tc,

r

dimensionless shear stress, critical Shields value

Tbc

critical bed shear stress

TKE

turbulent kinetic energy

U», u.

shear velocity (m/s)

U*c

critical shear velocity (m/s)

u

streamwise turbulent intensity

U

average approach velocity (m/s)

Uc
Ux

critical velocity (m/s)

Uy

lateral velocity (m/s)

Uz

vertical velocity (m/s)

V

volume

Vx

velocity in horizontal direction (m/s)

V

vertical turbulent intensity

X

horizontal longitudinal direction

y

horizontal transverse direction

z

vertical direction, vertical distance from bed

steamwise velocity (m/s)

Symbols
geometric standard deviation
V

kinetic viscosity of fluid

X

Einstein multiplication factor

a

surface tension

xiii

ACKNOWLEDGEMENT
I am grateful to my supervisor, Dr. Jueyi Sui and committee members, Phil Owens, Jianbing
Li, Bryan Kamey and Youmin Tang for support and direction throughout my PhD.
Thanks goes to fellow PhD Candidate Peng Wu for help and support throughout the entire
research process from flume building, to travelling, securing equipment, sieving sand to
finding places to live in Likely.
Thanks to my parents: Dad for original flume design drawings and going over material
specifications for flume construction. Mom thanks for your encouragement and support with
technology in a small town ©
Construction of the flume at the QRRC would not have been possible without enormous help
from both staff at the QRRC and community members in Likely. Thanks to Howard Fenton
for flume construction. I am grateful to Alex Koiter and Ben Anderson for help with moving
sand in and out of flume. Special thanks goes to Anja Forster for living in Likely for 5 months
and helping us build the flume.

I would also like to thank Jean Wang for help with installing CFD software and meetings
regarding FLUENT.
Appreciation is extended to the Prince George MOT for lending their bridge manuals to me.

Enormous thanks is extended to the National Sciences and Research Council of Canada for
providing me with a PhD scholarship. In addition, I would like to thank UNBC for providing
me with numerous graduate research and tuition awards throughout my PhD which greatly
assisted with securing equipment, travel costs along with tuition fees.
I am also grateful to the UNBC geography program faculty and staff for providing me with
teaching contracts throughout my PhD which helped with financing my life.
Lastly, thanks to my friends and special thanks to Steve for all your help and support which
allowed me to focus on writing my thesis.

xiv

1 .0 INTRODUCTION

1.1 Background
Scour is defined as the removal of material by running water (Chang, 1992). River
bed materials around stream banks, bridge piers and bridge abutments can be eroded and
carried away downstream. Scour that occurs around bridge piers can threaten the structural

integrity and cause bridge failure. Even though considerable research has been conducted
around bridge design and construction, bridges still fail due to scour during times of flooding,
extreme debris build-up and winter ice events (Melville and Coleman, 2000). Between 1966

and 2005 there were 1 502 bridge collapses in the United States (Briaud et al., 2007). According
to the New York State Department Bridge Failure Database river bed scour was responsible

for 878 of the failures (Briaud et al., 2007). For instance, in April 1987, the Schoharie Creek
Bridge in New York collapsed during spring freshet. Spring flows were higher than normal

and produced an estimated 50 year flood. Ten people perished in the accident. The United

States National Transportation safety board ruled the bridge collapsed due to sediment scour
around the bridge pier (Hains and Zabilansky, 2007).
Countries that lead the way in bridge research are New Zealand through the University

of Auckland in conjunction with Transit New Zealand and the United States (US) through the
Federal Highway Administration (FHWA) in conjunction with University Transportation
Centres. Studies are typically conducted in hydraulic laboratory flumes under open channel

conditions and considerable research has focused on sediment scour processes around bridge

structures (Alabi, 2006; Guo, 2012; Khwairakpam et al., 2012; Melville, 2008; Oliveto et al.,
2002; Raikar and Dey, 2009). The specific topic of pier scour has been widely studied over

the past three decades and summarized in reviews (Beltaos, 2000; Deng and Cai, 2010). A
1

r
number of master’s and doctoral theses have been produced examining pier, abutment and spur
dyke scour (Aliba, 2006; Hoque, 2009; Miranda, 2004; Zhang, 2005; Yang, 2005).

Very few studies, however, have looked at how winter ice cover conditions influence
sediment transport around bridge structures. A bridge over the White River in Vermont

experienced a number of ice jam breakups over several years; in the winter of 1990 the White
River Bridge failed. Examination into the failure revealed that the bridge foundation gradually
became deteriorated due to riverbed scour around the piers (Zabilansky, 1996). The difficulty
in studying pier scour under ice cover lies in the dangerous field conditions surrounding river
ice and maintenance of instrumentation around bridge structures. This research focuses on
measuring sediment transport processes under ice covered conditions around bridge piers.

1.2 Statement of the problem and research objectives
Since ice cover imposes an additional boundary layer on a channel, causing the

maximum flow velocity to migrate closer to the channel bed (Sui et al., 2010; Wang et al.,
2008) local pier scour processes should be different under ice cover than scour under open

channel conditions. Investigating pier scour development under ice conditions is important in
engineering and design of bridges to ensure public safety.

The main objective of this thesis is to gain a better understanding of how ice cover is related
to local scour around bridge piers. The following are the research objectives of this thesis:

(1)

Conduct experimental flume investigation for local pier scour. Examine (a) scour
depth and volume, (b) scour hole particle size distribution and (c) scour hole velocity

flow field under open channel, smooth ice and rough ice cover conditions.

2

(2)

Investigate relationships between the following variables:

a. ice cover roughness versus flow velocity distribution
b. channel cover versus turbulent intensity and Reynolds stress

c. scour depth versus particle size (D$o) and pier width
d. scour depth and scour profile versus channel cover

e. scour area and volume versus channel cover
f. scour depth and scour hole velocity profiles according to channel cover

g. channel cover and armour layer size

In fillfilling objective number 2, the following parameters will be used for dimensional
analysis:

a. pier width

b. scour depth

c. approach velocity
d. water depth

e. viscosity

f. particle size (D50)
g. channel roughness and ice cover roughness
(3)

Conduct a critical analysis of research practices as they relate to bridge design.

Specifically, empirical equations, scaling practices and computational fluid dynamics
pertaining to bridge scour studies and assess how components of this research work

fit into current bridge research.

3

J

1.3 Significance of research
In the field of hydraulic engineering it is standard to conduct flume experiments to
investigate relationships between fluid flow and hydraulic structures. To date, three studies

have investigated the influence of ice cover on local pier scour. Batuca and Dargahi (1986)
compared local scour around a pier under open water and floating cover conditions. Through

visual observations they found that the general size of the scour hole was greater under ice
covered conditions. Olsson (2000) compared pier scour depths under open, smooth and rough
ice cover and found that the rough ice cover resulted in a greater scour depth than the smooth

cover. Uniform sand was used. Olsson found that ice cover may increase the local scour depth
by 25-35 percent under live bed conditions. Hains (2004) studied the impacts of a fixed cover

on local pier scour. A fixed ice cover is not free to change position in accordance to river flow
or water level. A fixed ice cover can occur when ice attaches itself to the pier or channel sides.
The author examined pier scour under both floating and fixed covers along with open channel

conditions. Scour depths were similar under fixed and floating ice cover conditions.
As elements of this thesis are similar to the above three studies, it is important to highlight

how this research is original and differs from that found in the current literature.
(1) Numerous studies exist that investigate flow turbulence and scour depth around
bridge piers under open channels. It is the open channel literature in which pier scour

equations have been developed. This thesis examines pier scour under smooth and
rough ice cover. While it is noted that the three previous studies have investigated

pier scour under ice, there is still not a strong body of academic literature addressing

4

this topic. This thesis will expand on the three previous studies as explained in the
following points.

(2) The majority of existing open channel scour experiments along with the ice cover

experiments by Hains (2004), Olsson (2000), Batuca and Dargahi ( 1 986) use uniform
sediment.

The current research uses non-uniform sediment which allows for

examination of armour layer development within the scour hole. Using non-uniform
sediment is also more aligned to representing natural river systems.
(3) Scour hole velocity profiles under smooth and rough ice cover are measured in this

thesis. To the author’s knowledge there are no studies that measure scour hole flow

fields under ice cover.
As there is not a strong body of literature addressing pier scour under ice cover, it is anticipated

that findings from this study will be written up for potential publications in relevant
engineering journals. A summary of publication plans can be found in the conclusion, Chapter

7.

5

2.0 LITERATURE REVIEW

2.1 Types of scour
Chang (1992) defined scour as the removal of material by running water. Similarly,

Breusers et al. (1977) stated that scour is a natural process resulting from water flow in rivers.
A bridge can constrict river flow by introducing additional structures such as abutments and

piers. A constricted channel may have increased flow velocities and therefore and increased
capacity for sediment transport. Riverbed scour can be broadly classified as general scour,

local scour and contraction scour. Aliba (2006) provides the following definitions for scour.

General scour: can occur regardless of whether a river obstruction is present; involves the
overall lowering of the channel bed along the longitudinal profile. General scour occurs when

the river hydrology undergoes a change which leads to degradation of the channel bed. Human
induced general scour can be caused by dam construction and streambed mining. General

scour can also occur over short time periods as a result of flooding or seasonal freshet events.

Local scour: scour that is directly related to the presence of a river obstruction and involves
the removal of sediment from the base of bridge structures such as piers and abutments. When
water flow encounters a bridge pier or abutment, the flow is directed downward towards the

river bed and along the sides of the obstruction (Diab et al., 2010). The down flow causes
sediment transport away from the base of the bridge structure. Local scour will be further
discussed in Section 2.2.

6

Contraction scour: occurs when the river channel is constricted either by natural (landslide) or
human means (a bridge). When the flow area is reduced, the average flow velocity and bed
shear stress increase which then increase the erosive forces acting upon the channel bed (Ozalp,
2013). Increased erosion will move bed material from the contracted section until equilibrium

is reached. Bridges that cause a narrowing of the channel can be expected to have higher flow
velocities and larger scour depths (Zevenbergen et al., 2012). In circumstances where the
estimated depths from contraction scour are too large, the bridge crossing length must be
increased in order to reduce scour.

Live-bed scour: occurs when a stream is carrying a substantial sediment load and sediment is
transported into the scour depression all the while local flow fields are also removing sediment

from the scour hole. Live-bed scour commonly occurs in natural river systems and also in re¬

circulating flumes. Flow through flumes typically do not have upstream sediment being
deposited in the scour hole so live-bed scour does not occur.

Clear water scour: occurs when there is no sediment transport into the scour hole but rather
only sediment transport out of the scour hole. This occurs when there is no upstream supply

of sediment. This rarely occurs in active river channels.

2.2 Sediment transport around bridge piers
The flow pattern around a bridge pier is typically divided into 4 components: (1) the

down-flow at the pier face (2) the horseshoe vortex (3) the bow wave and (4) the wake vortices.
The following is a description of each component.

7

(1) Down-flow: the flow velocity decreases as it approaches a bridge pier and is reduced to

zero upon meeting the pier face, forming a stagnation point. When the flow hits the pier face
the overall channel depth also increases. The increase in flow depth depends on the approach
velocity and the pier shape (Yanmaz, 2002). As flow velocity decreases from the water surface

towards the riverbed, the resulting pressure gradient at the pier also decreases from the water

surface towards the riverbed. As flow velocity decreases towards the channel bed so does the
pressure. This pressure gradient creates down-flow at the pier face.

(2) Horseshoe vortex: the downward flow also interacts with the incoming river flow and
results in a vortex system as shown in Figure 1 . The vortex flow then moves along the side of

the pier downstream. Due to the shape of the vortex, flow around a pier is often referred to in

the literature as the ‘horseshoe vortex’ (Zhao et al., 2010). The horseshoe vortex is typically
formed after a scour hole is generated (Ozalp, 2013). The intensity of the horseshoe vortex
depends on the pier geometry and degree of turbulent flow (Yanmaz, 2002).

(3) Bow wave: the bow wave occurs at the upstream side of the pier at the water surface. The

bow wave originates from the increase in water depth as the approach flow hits the pier face
and a stagnation point is created. Richardson and Davis (2001) found that for low flow depths

the bow wave causes the horseshow vortex to become weaker and leads a reduction in scour
depth.

(4) Wake vortex: as flow passes the sides of a pier, flow separation occurs and wake vortices

form (Figure 1). These wake vortices also contribute to erosion of sediment around the base
8

of the pier. Wake vortices are transferred downstream by the approach flow and work to
transport sediment that is already entrained by the down-flow and horseshoe vortex (Melville

and Coleman, 2000). The strength of the horseshoe vortex is much larger than the wake

vortices hence the maximum scour depth typically occurs upstream of the pier face.

Figure 1. Flow pattern around circular pier (taken from Melville & Coleman, 2000).

The main mechanism that causes scour at bridge piers is the down-flow and subsequent

horseshoe vortices that form at the base of the scour hole (Muzzammile et al., 2004). The
down-flow reaches the channel bed and transports sediment away from the pier base creating
a scour hole. The strength of the down-flow reaches a maximum just below the natural bed

level (Alabi, 2006). Once the down-flow reaches the channel bed it interacts with the
oncoming flow and a complex vortex system develops. As indicated in Figure 1 , flow past the

sides of the cylinder separates and wake vortices are formed. Both the horseshoe and wake

9

vortices erode sediment from the pier base. The strength of the wake vortices however is
reduced with distance downstream and sediment deposition is common downstream of the pier
(Richardson and Davies, 1995).

Numerous studies have examined the factors which affect the magnitude of local pier

scour (Lagasse et al., 2001; Melville and Coleman, 2000; Raudkivi and Ettema, 1983;
Richardson and Davies, 1995). The following parameters have been recognized as affecting
local scour depth and can be grouped by the following headings according to Breusers et al.
(1997):

• Bed sediment parameters: cohesiveness of sediment, grains size distribution, particle
mass density, particle shape, bed roughness and angle of repose.

• Pier parameters: orientation (angle of attack), size, shape, spacing and number of piers.
• Flow parameters: flow velocity, flow depth, shear velocity and velocity distribution.

• Fluid parameters: kinematic viscosity, mass density and acceleration due to gravity.
The following is a brief summary of the effects of the above parameters on pier scour as
outlined by Melville and Coleman (2000).
Local scour depth under clear water conditions will increase with velocity until a
maximum threshold velocity is reached. The maximum scour depth occurs when u«/u»c = 1.
(where u» is the shear velocity and u»c is the critical shear velocity). Flow depth is usually

referred to as flow shallowness and is examined by relating flow depth (H) to the pier width
(b). At ratios of b/H < 0.7, local scour is dependent only on pier width, and at ratios of 0.7<

b/H<5 scour depth is dependent on both pier width and flow depth. Pier diameter and scour
depth are related as the greater the pier diameter the stronger the vortex system. The larger the

10

pier width, the stronger the flow vortices and the larger the scour hole volume and scour depth.

Pier shape is only important if complete axial flow can be guaranteed. In other words, even a
small angle of attack will eliminate the effectiveness of a streamlined pier shape. A study by

Mostafa (1994) compared all pier shapes and concluded that a circular shape minimizes scour
depth to the greatest extent. Sediment coarseness also affects local scour depth. Sediment

coarseness as defined by Melville and Coleman (2000) is the ratio of pier width to mean grain
size of bed material, b/Dso. Local scour depth is affected by sediment coarseness at ratios of

b/Dso < 50. Sediment nonuniformity also effects local scour depth. For non-uniform
sediments, close to the threshold velocity, armouring of the scour hole will occur. This will

reduce the local scour depth in comparison to uniform sediments.

2.3 Incipient motion of sediment particles
Incipient motion of sediment occurs when flow intensity in a channel is barely

sufficient to entrain bed particles. The hydrodynamic forces of a fluid acting on the particles

are responsible for their motion (Kanellopoulos, 1 998). Since sediment is transported along a
riverbed in a complex manner, it can be challenging to define the flow conditions in which
sediment will move. For this reason studies focus on the flow conditions surrounding the

initial motion or incipient motion of sediment. As indicated in Figure 2, there are four primary
forces acting upon a sediment particle: Fl represents the lift or buoyancy force, Fw indicates

the force downward from the submerged weight, Fd is the drag force and Fr is the resistance

force.

11

Direction of flow
».

Figure 2. Overview of forces acting on a sediment particle.

In order for incipient motion to occur one of the following conditions must be satisfied:

Fd = Fr

2.1

Fl = Fw

2.2

A number of variables affect the lift, weight, drag and resistance forces acting on a sediment
grain. Shields, in 1936 was the first to develop a relationship between hydraulic variables and

sediment characteristics in order to satisfy the conditions for incipient motion. Through
dimensional analysis, Shields examined conditions in which bed particles are stable but on the

verge of being moved. Shields determined that the critical conditions in which sediment is
about to become entrained can be found by relating the critical Shields value (t£) and the shear
Reynolds number (Re,). The critical Shields value, also called the dimensionless shear stress,
is given by the following relation:
Tbc

(Ps-P^gD

2.3

12

where Tbc

= pU,c2 = critical bed shear stress for initiation of motion, U,c is the critical shear

velocity, (ps

- p) is the density of sediment and water respectively, g is gravity and D is the

diameter of the sediment particle.
The Reynolds number, R«, is a dimensionless quantity that represents the ratio of

inertial forces to viscous forces. The Reynolds number is typically used to characterize
whether a flow is laminar or turbulent. Laminar flows occur at low Reynolds numbers where
viscous forces are dominant, while turbulent flows are characterized by high Reynolds
numbers where inertial forces are dominant. The Reynolds number can also be adapted to
study the incipient motion of sediment through the particle Reynolds number. The particle
Reynolds number is given as follows:

m
where U is the flow velocity, D50 is the median particle diameter and v is the kinetic viscosity

of fluid. When studying incipient motion the shear Reynolds number R* is referred to as it

uses the shear velocity and is written as:

*

=

(^)

where U, is the shear velocity. The graphical relationship between the dimensionless shear
stress (Tc) and the shear Reynolds number (R,) is referred to as the Shields diagram which

represents values in which incipient motion of sediment occur.

The relationship is

dimensionless so that the driving forces of particle motion can be compared to the resisting

forces (particle size, particle density).

13

The incipient motion of sediment depends not only on hydraulic variables but also on
the characteristics of the bed material itself. Incipient motion of non-uniform sediment is more
complicated than uniform sediment since sediment in motion is influenced by the grain shape,
interactions between grains and orientation in the channel bed. The Froude number represents
the ratio of fluid inertial forces to fluid gravitational forces and is used to determine the
resistance of an object flowing through water. The greater the Froude number, the greater the

resistance exerted on water flow by the river bed material. The Froude number is given by,

where h is the flow depth. In order to use the Froude number to examine transport of bed
material, the dimensionless fonn of the Froude number is used. The densimetric Froude

number is the ratio of inertial forces to the submerged weight of the sand grain. The
densimetric Froude number is written as:

’J

gDso(Ps-p)
p

where Dso is the median grain size of the bed material.

While incipient motion of uniform sediment has been studied in the literature (Andrey

and Gareth, 2000; Beheshti and Ataie, 2008; Buffington and Montgomery, 1997;
Kanellopoulos, 1998; Vollmer and Kleinhans, 2007), studies that examine incipient motion of

non-uniform sediment are limited. Xu et al. (2008) established a formula for incipient velocity

of non-uniform bed material based upon the drag and uplift forces and verified the formula

with field data. They found that the incipient velocity required for coarse non-uniform
particles was less than the incipient velocity for uniform sediment. In addition, the authors

14

calculated that the incipient velocity for fine particles for non-uniform sediment was greater
than that for uniform sediment.

2.4 Incipient motion of sediment under ice conditions
River ice cover imposes an extra boundary on flow, altering the flow velocity and water

level in comparison to open channel flow (Shen and Wang, 1995). For ice cover conditions,
the portion of upper flow is mainly influenced by the ice cover resistance while the lower flow
is mainly influenced by the channel bed resistance (Sui et al., 2010). The maximum flow is
located between the channel bed and ice cover depending on the relative magnitudes of the ice
and bed resistance coefficients (Crance and Frothingham, 2008; Ettema et al., 2000; Lau and
Krishnappan, 1985; Smith and Ettema, 1997). Generally, the maximum flow velocity is closer
to the surface with the smallest resistance coefficient. In the case of narrow river channels or

near river banks, the maximum flow velocity will not occur at the surface but rather slightly
below the surface due to the resistance forces of the side banks (Wang et al., 2008).

Since ice cover imposes an added boundary on flow conditions, the incipient motion

of sediment under ice cover is different from that for open channel flow. Wang et al. (2008)
conducted a number of flume experiments examining the relationship between incipient
motion of bed material and ice cover conditions. Since near-bed velocity is higher under ice
covered conditions, a higher shear stress is exerted on the river bed (Wang et al., 2008). The
threshold velocity for the incipient motion of sediment under ice cover decreases as the ice
cover resistance increases. This is due to the increased kinetic energy exerted on the bed

material as the near-bed velocity increases. The flow velocity required for initial movement

of bed material under ice cover also increases with water depth (Wang et al., 2008). This
15

relationship is only valid if the resistance coefficients of the ice cover and channel bed remain

constant.
Generally, the larger the roughness coefficient for ice cover, the greater the near bed

flow velocity. In this situation, the larger the roughness coefficient for ice cover, the smaller
the densimetric Froude number for incipient motion of river bed material (Wang et al., 2008).

Since the near bed velocity is relatively high, larger sediment particles can be moved.
However, if the roughness coefficient of the river bed is high, the near bed velocity will
decrease; this in turn will increase the densimetric Froude number required for incipient motion

of bed material.
As depicted in Figure 3, winter ice cover and ice jams can significantly alter the river

bed. Based upon field observations of ice jam formation and frazil ice transport, Sui et al.
(2000) proposed that frazil ice formation and riverbed deformation reinforce each other. The

authors found:
(a) As an ice jam grows, the riverbed will be scoured. Once an ice jam begins to decrease

and diminish, the riverbed material will undergo deposition. During the entire life of an ice

jam, patterns of scour and deposition will be repeated.
(b) During ice cover, the decrease in river cross section causes river flow to take a path of

least energy consumption. As a result, the riverbed will become deformed.

16

Figure 3. Changes incurred at a cross section of Hequ reach of
the Yellow River during an ice jam period (from Sui et al., 2000).

2.5 Summary and conclusions
With the exception of three studies: Hains (2004), Olsson (2000) and Batuca and

Dargahi (1986), the majority of literature to date has focused on pier scour under open channel

conditions. Since ice cover imposes an added boundary on flow conditions, the flow pattern
and resulting incipient motion of sediment under ice cover is different from that for open
channel flow. Generally, the larger the roughness coefficient for ice cover, the greater the near
bed flow velocity and the greater the sediment transport around a bridge pier. This study

attempts to gain a better understanding of how ice cover is related to local scour around bridge

piers by conducting flume experiments as described in the following sections.

17

XOEXPERTiJIKlPAL SETUP: FLUME CONSTRUCTION

3.1 Background
There is no hydraulic laboratory at UNBC and securing use of a hydraulic flume at
another institution was not possible. As a result a flume was built at UNBC’s Quesnel River
Research Centre (QRRC).

Prior to commencing construction a survey was conducted of the flume literature to
investigate present flume configurations and set-ups in hydraulic laboratories around the

world. To the author's knowledge there are very few articles in the literature that review
theoretical and experimental considerations for flume design (Ettema and Muste, 2004; Nowell
and Jumars, 1987). The most extensive review by Nowell and J umars (1987), concludes that

no single flume design is applicable for all studies. Small flumes will have a smaller range in
flow depths and discharge while larger flumes will have greater work and construction costs
along with larger ranges in hydraulic parameters. Upon further investigation of the literature,

it was found that the majority of hydraulic flumes are located indoors, have an average length

of 20-30 m (or less) and an average width of 0.6 m. It is by these standards in which the flume
constructed for this research was classified as a large scale flume. The following is details of
the construction process.

3.2 Flume construction
The research centre has a number of outdoor fish rearing channels that are currently set

up as flow-through systems. During my first field season in 2011, a fish rearing channel was
converted into a hydraulic flume. The water source for the flume is a well situated uphill of
the flume allowing for a gravity operated system. The well water runs through an aeration

18

tower prior to flowing downhill into the flume channel. The aeration tower’s purpose is to

oxygenate the water for fish rearing. Since the flume water originates from groundwater the
water temperature in the flume ranged from 8.6-10.4 degrees Celsius. This is an important

aspect of the flume as it resides outdoors and was operated during the summer, fall and early

winter months.
The flume channel (original salmon spawning channel) is 80 m long, 2 m wide, 1 .3 m

deep and is located outside. Since the flume is outside exposed to the elements, a roof 40 m

in length was constructed to cover the channel downstream from the weir (Figure 4a). The

roof frame was constructed out of lumber and covered with greenhouse grade poly. Viewing
windows were built by removing two 4 m sections of concrete wall and replaced with 6 mm
thick Lexan Margard polycarbonate (Figure 4b). The viewing windows were framed with fir
and cedar so the frame would expand and contract with the variable water pressure exerted on

it (Figure 4c). The flume floor and viewing windows were sealed with industrial marine
silicone and allowed to cure for 2 weeks. Two enclosed viewing buildings (Figure 4d) were

also constructed so that the flume experiments could be viewed while being protected from
outside weather elements. The viewing buildings were partially sunken into the ground to
allow for viewing of the flume floor. The building foundations were constructed out of wood
and lined with gravel to allow for drainage. Heaters were placed in the viewing buildings to
maintain adequate temperatures for running pumps, computers and flow meters. Power was

supplied to the flume via extension cords from nearby power outlets. In order to obtain an
adequate range in flow velocity, three channel valves were diverted into the flume. A weir

was constructed in the middle of the channel at 40 m to create a head tank (Figure 4f). A debris

screen was installed prior to the weir in the head tank to control outside leaf debris from
19

entering the flow. Downstream from the weir a flow diffuser was installed in order to decrease

the flow turbulence prior to the first sandbox (Figure 4e). The channel bed downstream of the
weir was raised by 0.30 m, in order to create two sand boxes. This was done by constructing

a plywood framed floor that was sanded and sealed with two sand box openings. The flume
floor in front of each viewing window was a sandbox area measuring 5 m in length and 2 m in

width. An adjustable tailgate was also installed at the end of the flume channel so the water

depth could be altered. A schematic diagram of the flume with given dimensions is presented
in Figure 5 and Figure 6.

20

Figure 4. (a) Construction of flume roof, (b) Viewing window and frame, (c) Concrete wall removal and frame
for viewing window, (d) Enclosed viewing platform, (e) Flow diffuser, and (f) Flume head tank.

21

Figure 6. Side view of flume with exact dimensions.

22

3.3 Experimental runs
The main objective of this thesis is to investigate variables that affect pier scour depth,
namely: channel cover, sediment grain size (Dso), pier width, flow velocity and flow depth.
After a search of the flume literature, to my knowledge there is no criteria that researchers

follow that indicate minimum number of flume runs, experimental run time or

recommendations on selecting sediment grain size and flow velocity. Having never conducted
flume experiments prior to this thesis, extensive consideration was given to deciding the
experimental design criteria of the flume experiments.

In general, the procedures followed are that of the scientific method to manipulate
independent variables (channel cover, grain size, pier size) while observing the dependant

variable for change (scour depth). However many external factors influenced how the flume
experiments were set up, all of which are summarized in the following sections.

Flume calibration:
Determining the range in flow velocity and water depth for flume experiments was

dependant on finding a balance between the water supply system and the flume tailgate
configuration. Since the flume is operated by a gravity fed system the discharge is controlled
by opening and closing supply valves. The supply valves could only be opened in quarter

increments due to the original set up for spawning channels. It was determined that a
maximum of 1.25 valves could be opened to supply water to the head tank; any further supply

of water would cause the head tank to overflow. Given that a supply of 1.25 valves did not
provide sufficient approach velocity for all experiments, an additional valve was supplied to

the flume, downstream of the weir, as shown in Figure 7. With a 2.25 valve configuration the

23

flow velocity was sufficient to operate the flume with up to a 20 cm tailgate setting. After
completing water valve and tailgate testing, the upstream and downstream sandboxes were

measured for approach velocity and flow depth under various valve and tailgate settings. Table
1 indicates the minimum and maximum hydraulic conditions under the present flume
infrastructure. Considering the channel conditions in Table 1 along with using two pier sizes,
it was decided that experiments would be completed in the upper and lower sandboxes in order
to capture different flow depths and velocities. As a result of the flume tests it was concluded

that six experimental runs could be completed for each channel cover incorporating both pier
sizes. This resulted in 18 experimental runs per sediment type as outlined in Table 2.

Figure 7. View of flume showing second valve configuration with white pipe entering flume roof
just downstream of weir, white arrow indicates white pipe.

24

Table 1. Measured velocity and stage resulting from various valve and tailgate configurations. Configuration
one and four were chosen for experimental conditions. Note: testing was also undertaken using three tailgates
(0.30 m) and velocity was not sufficient for incipient motion to occur. In addition under the two tailgate
configuration the small pier was always placed in the upstream sandbox since under such channel conditions in
the downstream window no scour/incipient motion occurred.

Config¬
uration

#
valves

#
tailgates

1

1.25

2

1.25

Upstream sandbox
velocity
stage
(cm)

(cm/s)

1

10

2

21

23
12 velocity
too small for
incipient
motion
36 ’velocity

3

2.25

1

10

4

2.25

2

21

•

Downstream sandbox
stage
(cm)

velocity (cm/s)

13

20

26

1 1 ’velocity too small
for incipient motion

too large
13
leading to
live bed scour

25 ’small
pier only

26

33 ’velocity too large
leading to live bed
scour
23 ’large pier only

Table 2. Example of experimental flume schedule for one sediment type.

Run

Dso

#

(mm)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58

0.58
0.58
0.58

Cover
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice

Position in
flume

#
valves

tailgates

upstream
upstream

1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25

1
2
1
1
2
1
1
2
1
1
2
1
1
2
1
1
2
1

downstream
upstream
downstream
downstream
upstream
upstream

downstream
upstream
downstream
downstream
upstream
upstream
downstream
upstream
downstream
downstream

#

Pier
width
(cm)
11
11
11
22
22
22
11
11
11
22
22
22
11
11
11
22
22
22

25

3.4 Experimental simulation time
An important factor that was considered in the experimental design was the time

required for equilibrium scour depth to be reached. Upon review of pier scour studies in the
literature, it was found that the subject of equilibrium scour is either (a) not addressed or
spoken of in the paper at all, (b) defined by visual observation, (c) defined by a ratio of change

over time or (d) determined that equilibrium scour could not be reached under the experimental
conditions. One of the first studies to define equilibrium scour was Ettema's PhD thesis on
bridge pier scour (1980): the time to equilibrium scour was defined as the time at which no

more than 1 mm of incremental scour occurred within a timeframe of 4 hours. Most often

equilibrium scour is defined through visual observation and a ratio of change over time (an
overview of pier scour studies is provided in Table 3 along with equilibrium scour definitions).

Given that every flume is different in dimension, discharge and sediment type it makes sense
that equilibrium scour is a difficult topic to define. That being said since the influence of ice

cover on pier scour is the main topic of this thesis, it is important that equilibrium scour be
addressed.
During the testing/calibration phase of flume experiments (described under section 3.3
above) piers were placed in the sandbox and the scour hole was visually monitored at hourly

increments. The scour hole test runs were monitored at the highest velocity configurations
along with the highest depth configurations to ensure the maximum scour conditions would be
obtained. The flume was typically started at 8 am and monitored hourly for changes in scour

depth. After a period of 4 hours no change in scour depth was observed. After a period of 6

hours an armour layer had developed in the scour hole. The flume was run for 24 hours and
still no change in scour depth was observed. At such time the experimental test results were

26

discussed with Dr. Sui (UNBC), and it was decided that flume experiments should last at least

20 hours to ensure the equilibrium scour had been reached. After discussion with the research
centre staff it was decided that logistically the flume could run continuously throughout the

night allowing for one simulation per 24 hour period. Providing photographic evidence of

equilibrium scour tests proved difficult due to water ripples (Figure 8). Also, when looking
through the viewing window at the scour hole, photographs at any angle were unable to capture

inside the scour hole Figure 9.
Table 3. Summary of literature standards regarding the definition of equilibrium scour depth. Channel width is
indicated by B.

Study
Acharya
(2011)
BabaeyanKoopaei
and
Valentine
(1999)

Zhao et al.
(2010)

Flume Flume
Number
width length
of runs
(m)
(m)

0.6

2.5

4

Raikar and
Dey
0.6

12.2

22

7

12

Length

of runs
(hrs)

24

7-8

Parameter Time for equilibrium
of measure scour depth addressed
Spur dyke
scour

Yes, decided 24hrs
upon visual observation

Pier scour

No significant change
after 3-4 hours so
experiments stopped
after 7-8 hours

45

28

2.354.5

Pier scour

Equilibrium scour
reached when mean bed
level change rate at
front of cylinder was
less than 0.03B per
hour

12

40

18-36

Pier scour

Visual observation

Pier scour

Experiments stopped
when change in scour
depth did not exceed
0.05B during 24 hour
period (proposed by
Melville and Chiew,
1999).

(2005)

Sheppard
et al.
(2004)

6.1

38.4

14

41-616

27

Figure 8. Aug 26, 201 2, 2:32pm flume testing. Water ripples did not
allow for down facing photographs of the scour hole.

Figure 9. August 27, 2012, 3:35 pm, side angle photograph taken during flume testing,
photograph inside scour hole was not practical due to flume design.

28

3.5 Flume sediment
Since the flume is a large scale flume, each sandbox required a minimum of 3 m3 of

sediment to sufficiently fill the recessed area. During transport in and out of the flume
sediment was lost therefore it was standard to order 4 m3 for each sandbox equalling a total of
8 m3 of sand for each set of experimental runs. Given that 8 m3 of sand is the size of an entire

dump truck load (Figure 10), it was unrealistic and beyond the budget of this thesis to use

manufactured engineering sand. Additionally, the objective of this thesis was to investigate

scour depth under non-uniform sediment. Therefore landscape sand was used for flume
experiments. All known sediment quarries within a 500 km radius of the research station were
contacted. Sand samples were collected from the quarries, oven dried, sieved and median
particle size (Dso) was calculated. The mason, concrete and bedding mix sand types were
selected, with Dso = 0.47 mm, 0.58 mm, 0.50 mm, respectively (Table 4 and Figure 12). This

selection was based upon the fact that the mason, concrete and bedding sand were the three
most common sands mined from the surrounding quarries. In one instance, an alternative

variety of bedding sand was sampled; however with a Dso of 0. 1 8 mm it was too small for the

flume experiments as live bed scour would occur. Given that three sand types were selected
with 1 8 flume experiments per sand type, it was decided that a total of 54 flume runs were
required for sufficient comparison between channel covers and sediment types. Between
experiment sets, sand was manually moved in and out of both sandboxes (Figure 1 1).

29

f
t

[
[

t

Figure 10. Landscape sand sample that was used in
flume.

Figure 1 1 . Sand being shoveled from sandbox # 1 .

30

Table 4. Particle size distribution of sands used during flume experiments.

Dio

D16
Dio

D30
D40
Dm

Dm
D70
Dso
Dg4

Dw
D99

Mason sand (nun)
0.19
0.23
0.25
0.32
0.39
0.47
0.54
0.60
0.70
0.77
0.97
4.00

Concrete sand (mm)
0.21
0.23
0.31
0.39
0.48
0.58
0.70
1.00
1.58
1.90
2.60
4.00

Bedding sand mix (mm)
0.22
0.24
0.27
0.33
0.41
0.50
0.57
0.73
1.20
1.57
2.18
4.00

pas ing
%

sieve size (mm)

Figure 12. Particle size distribution of sediments used for flume experiments.

31

Variability of the three materials:

The Unified Soil Classification System along with sediment engineering textbooks

typically regard soil as uniform when the coefficient of uniformity, equation 3.1, is less than 4
(Assaad et al., 2004). Additionally, if studies are investigating sand for more industrial

application then it seems that a higher value of uniformity is also accepted. For example,
Olufayo et al. (2010) use a uniformity of less than 5 to classify sand as uniform in laboratory

experiments studying sand water storage systems for weir height classifications.

In the flume literature the geometric standard deviation, equation 3.2, is primarily used
when classifying sediments as uniform or non-uniform. As described by Hoffmans and

Verhejj (1997) sediment gradation is typically characterized using the geometric standard
deviation (ofl); for natural river sand the ag is approximately 1 .8 and for uniform sand the ag
is approximately 1.3. The standard deviation, coefficient of uniformity and geometric mean
were calculated for the three sands used in this thesis (Table 5). Table 6 is a summary of

standard deviation values found in the flume literature along with how the authors classified
the sediment in each flume experiment. Based upon the values and conclusions of Table 6,
the geometric standard deviation value of 1.83, 2.87, and 2.56 were used to classify the three
sediments used in this thesis as non-uniform.

Coefficient of uniformity (Allen Hazen, 1911)
Geometric standard deviation
Geometric mean Dg =

—

Cu = ^10

3.1

og =

3.2

7^84^16

3.3

32

Table 5. Summary of geometric mean (Dt), standard deviation (<r#) and uniformity
coefficient (CM) for the three sands used in flume experiments.

D,

Ctl

Mason sand
D$o = 0.47 mm
0.42
1.83
2.84

Concrete sand
Dso = 0.58 mm
0.66
2.87
3.33

Bedding sand mix,
Dso = 0.50 mm
0.61
2.56
2.59

Table 6. Summary of literature values pertaining to geometric standard deviation of sediments used in flume
experiments.

aa

Author

D$o (mm)

Muste et al. (2000)
Raikar and Dey (2005)

0.25
1.4
4.10-14.25 >1.4
3.1
2.15

Oliveto and Hager (2002)

Ataie-Ashtiani and AslaniKordkandi (2013)
Sheppard et al. (2004)
Alabi (2006)
Barbhuiya and Dey (2003)
Link et al. (2008)

Classification of sediment as
stated by authors
Uniform sand
Non-uniform gravel
Relatively non-uniform
Medium sediment non¬

1.2

1.80

0.71

1.2

Uniform sand

0.22-2.90
0.53
0.52
0.26

1 .2-1.5
1.23
1.21

Near uniform sediment
Uniform sand
Uniform sand
Uniform sand

1.37

uniformity

Measure of sediment shape:
There are three main ways to express sediment shape: ( 1 ) roundness- a measure of the
sharpness of the comers of the sediment grain, (2) sphericity - a measure of the degree of
similarity between the grain and a perfect sphere, and (3) form - the overall appearance of the

particle. A common measure of shape is the Corey Shape factor (SF) given by:

SF = ~
>Jab

3.4

33

Where ‘a ’ is the longest sediment axis, 'b’ is the median sediment axis and ‘c ’ is the shortest

sediment axis. A spherical particle will have a shape factor of 1.0 while natural sands have a
shape factor of 0.7 (Garcia, 2008). The shape factor is important as it relates to the fall velocity

and transport of a particle. The majority of sediment transport equations were developed by

studying uniform bed materials.

In the flume literature the incipient motion of sediment is often studied through the
dimensionless shear stress, shear Reynolds number and shear velocity. These parameters all
require a measure of the sediment diameter. Since a single grain size diameter cannot be

calculated for the three sediment examples used in this research (as they are non-uniform) it is
standard practice to use the Dso to represent the particle size D. The use of Dso in place of the
particle size D for non-uniform sediments was originally studied and proven by Yang (2003)
and can be found in the literature regarding flume experiments using non-uniform sand (Bong,

2012; Wang et al., 2008). It is proposed that the sediment Dso be used in all equations requiring

a measure of sediment diameter.

3.6 Flume scaling and pier selection
Prior to commencing construction, a survey of the flume literature was conducted to

investigate present flume configurations and set-ups in hydraulic laboratories around the

world. The lengths, widths, shape and pump capacities of flumes all vary as it depends on the
purpose for which the flume is being used. However, over the years there have been some

standard scaling guidelines published in the flume literature that were used in

designing/developing the present study. Table 7 is a summary table of the standard scaling
guidelines presented in the literature along with dimensions of the current study flume. The

34

flume constructed for this thesis falls within the current scaling guidelines taken from the
literature with the exception of flow depth to pier width. Further information which addresses

this scaling issue can be found in the critical analysis section of this thesis, Section 6.1.1.
Bridge piers were constructed out of PVC plumbing pipe and were spherical in shape.

Pier widths were determined by calculating the 5 and 10 percent of the flume width, resulting
in pier widths of 1 1 cm and 22 cm. The ratio of channels width to pier diameter should be
greater than 6.5 in order to avoid effects from the channel sidewall (Ataie-Ashtiani and Aslani-

Kordkandi, 2013). Piers were placed in the centre of the sandbox and attached to the flume

floor (Figure 1 3).

Figure 13. 22 cm PVC pier installed in sandbox.

35

Table 7. Summary of flume scaling guidelines and comparison to scaling ratios calculated from current study
flume dimensions. Present flume ratios are on the left side of table with literature guidelines on right side of
table.

Present
flume
dimensions

Width: 200cm Piet*l:llcm Pier#2: 22cm
D50: 0.47 mm, 0.50 mm , 0.58 mm
largest flow depth 26cm
smallest flow depth: 10cm
Scaling guidelines from literature
Ataie-Ashtiani and Aslani-Kordkandi. 2013: The ratio of
200cm: 1 1cm the channel width to pier w idth should be greater than 6.5
so to ensure that the flume wall has no effect on scouring.
Ratio: 18:1

ratio
Channel
width : pier
width

Alabi, 2006: For live bed scour the flume width should be
at minimum 10 times the pier width for scour depths not to
be reduced due to bed features being modified as they
propagate through constriction.

ratio
Flow depth
: pier width

200cm:22cm
Ratio: 9:1

1 0cm: 1 1cm
Ratio: 1:1.1

10cm:22cm
Ratio: 1:2.2

26cm:22cm
Ratio: 1:1.2

26cm: 1 1cm
Ratio: 1:2.4

1 1cm:
0.47
Ratio:

1:23

ratio
pier width :

D50

Shen and Schneider, 1969. the width of an experimental
flume should be at least 8 times the pier size for clear-water
scour conditions so that sidewall effects are minimized.

22cm:
0.47
Ratio:
1:47

1 1cm:
0.50
Ratio:
1:22

22cm:
0.50
Ratio:
1:44

1 1cm:
0.58
Ratio:
1:19

22cm:
0.58
Ratio:
1:38

Scaling guidelines from literature
Ataie-Ashtiani and Aslani-Kordkandi, 2013:
The ratio of the flow depth to pier width
should be greater than 4 so the scour depth is
independent of the flow depth.
Scaling guidelines from literature
Ataie-Ashtiani and Aslani-Kordkandi, 2013:
The ratio of the pier diameter to the D50
needs to be greater than 50 so the size of
sediment particles has no influence on scour
depth.
Melville and Coleman, 2000: defined
sediment coarseness as the ratio of pier
width (b) to the mean grain size (D50), b/Dso,
the local scour is affected by sediment size
as long as the ratio b/Dso<5O. If b/Dso>5O
the local scour is not influenced by the
sediment coarseness. For b/Dso < 8, the
individual grains are so large relative to the
pier that scour is mainly due to entrainment
at the flanks of the pier (Melville and
Coleman, 2000).

36

Ettema, 1980: found that grain size does not
affect the scour depth if the pier width to
grain size ratio exceeds the value of 50.

ratio

Flume
width :
water depth

200cm:10
cm
Ratio:
20:1

200cm:26
cm
Ratio:
7.7:1

Scaling guidelines from literature
Nowell and Jumars, 1987: a minimum value for
the width to depth ratio is 5 in order to reduce
boundary wall effects and effects of secondary
circulation. “It is wise to try and get a ratio of
1 0: 1 , but designing and operating flumes is full
of compromises between conflicting
dimensionless groups and pragmatic limitations
such as pump capacity” (p.100)

3. 7 Channel cover
Flume simulations were conducted under open channel, smooth ice and rough ice

conditions. Ice cover was created using standard 1 .2 m x 2.4 m (4 x 8 foot) Styrofoam panels
purchased at a local hardware store. Styrofoam density was 0.024 g/cm3. The Styrofoam was
placed inside the flume so that it floated on top of the water during simulation (Figure 14A).
Rough ice cover was simulated by attaching small Styrofoam cubes to the underside of
Styrofoam panels using toothpicks (Figure 14B). Styrofoam cubes were 2.5 cm x 2.5 cm x 2.5

cm in dimension and spaced 3 cm apart.

37

Figure 14. A: Styrofoam floating around the pier inside flume. B: Styrofoam cubes attached to create a
rough ice cover.

3.8 Measuring flow velocity
Approach velocity and flow temperahire were measured using a 2D flow meter by

Sontek Incorporated. San Diego, USA (Figure 15, Figure 16). The Sontek IQ is rectangular in
shape and is designed to mount on the channel bottom. The impact to flow is minimal as the

Sontek IQ is sleek in shape. For this research, a constant approach velocity is required to

ensure that the flume hydraulics were constant over a 24 hour period. A staff gauge was also
installed in each sand box to manually verify water depth. The scour hole flow field was
measured using a 10-Mhz acoustic Doppler velocimeter (ADV) by Sontek (Figure 17). The

sampling volume of the 10-Mhz ADV is 10 cm from the sensor head. The ADV operates on

the principal of a Doppler shift, measuring the phase change when the acoustic signal reflects
off particles in the flow. The scour hole flow field was measured at 1 cm increments for the
depth of the scour hole in front of the bridge pier for each experimental run (Figure 18 and 19).

For ice covered experiments, removable flaps were cut in the Styrofoam to allow placement
of the ADV for flow field measurements (Figure 18).

38

Alwn AmH Bom

Stew team

"

"

V^’x*1 team
RfWSWMW

Figure 15. Diagram of 2D flow meter Sontek IQ (Sontek, 2014).

Figure 16. Sontek IQ mounted to flume floor.

39

Remote Sampling Volume
Diameter 0 6cm, Height 0 9 cm
Figure 17. Sontek 10 MHz 3-dimensional ADV (Sontek, 2014).

Figure 18. Use of Sontek 10-Mhz ADV
to measure scour hole velocity under ice
cover.

Figure 19. Sensor head of Sontek ADV
measuring velocity inside scour hole.

40

3.9 Scour hole measurements
Upon completion of each experiment the flume was drained and the scour hole was

contoured using manual calipers. Contour measurements were taken at 0.5-1 cm increments
at preset numbered measuring points along each bridge pier (Figure 20). Three-dimensional
contour data were recorded and entered into Surfer plotting software (Golden Software

Incorporated, Golden, USA) and scour plots were generated.

Measuring points around small pier (10 cm)

Measuring points around large pier (22 cm)

Figure 20. Numbered measuring points drawn on the outside of the small bridge pier (10 cm width) and large
bridge pier (22 cm width) for scour hole contour points of reference.

3.10 Armour layer sampling
After each experimental run was completed, the flume was drained, the scour hole was

contoured and samples of the armour layer were taken. The following section describes the
methodology followed in sampling the armour layer. Armour layer thickness typically extends

from the bed surface to the bottom of the largest (Dmax) surface particle or the most commonly
occurring particle size (Ddom). Here Ddom is approximated to equal D90 (Bunte and Abt, 2001).

When sampling the armour layer, care should be taken to ensure the entire depth of the armour
layer is considered otherwise finer particles are not sampled leading to a distribution that is

41

incorrectly large in diameter. The sampling methodology followed for collecting armour
samples was taken from Bunte and Abt (2001). To my knowledge, the authors provide the
most extensive review to date of armour layer sampling protocols. The following is a brief
excerpt from their report (p.189):

Armour thickness is approximated by:
a) the c-axis of the

particle of the surface (Ettema, 1984):

b) the b-axis of the Dna particle size (Diplas and Fripp, 1992):
c) two times the b-axis of the Dw surface particle size (Simons and Sentiirk

1992, p.654);
d) the embedded depth of the reach-average Ddom particle size;
e) the embedded depth of the local Dma* particle size.

The five prediction criteria listed above result in different armour-layer depths

when applied to the same deposit.

This is demonstrated in Figure 21. Since measuring the embedded depth of the armour layer

was difficult due to the small scale of the sediments, the protocol followed under points (d)
and (e) above was not applied. As the sediment was non-uniform the a, b and c-axis
measurements were not used. Instead, two times the Dw of each sediment was used as a

baseline for determining armour depth. The D90 values were 2.18 mm, 2.60 mm and 0.97 mm.
The task of sampling the armour layer to the exact depth of two times the D90 values was

virtually impossible. Therefore, the armour samples were taken from the top 5 mm of each

scour hole (small tick lines on the pier represent 5 mm intervals), using the ruler drawn on the
pier for a visual guide for sampling depth (Figure 22).
42

1 cm

2 cm
3 cm

Figure 21. Differences in armour-layer thickness determined for the same deposit using various
prediction criteria.

Figure 22. Example of armor layer in scour hole after experimental run # 1,
September 22, 2012.

43

3.11 Flume operation procedures
Flume experiments were conducted for (1) open-flow conditions, (2) smooth ice conditions

and (3) rough ice cover. The flume was operated by the following steps:

• position bridge pier in flume, level bed surface, set tail gate, place Styrofoam for ice
covered runs, fill with water, start Sontek IQ. Note: for each simulation the flume was

filled with water slowly so to prevent initial scour from occurring. This process would
take approximately 20 minutes.

• maintain uniform water flow for duration of simulation time (20 hours).
• after 20 hours measure scour hole velocity profile in front of pier using Sontek ADV.
Note: during all velocity measurements no person was standing in the flume, but rather

a wooden platform was suspended above the water so measurements could be taken
while not impacting the water flow.

• stop water, drain flume, download Sontek IQ data, take photographs, manually contour
the scour hole with calipers and collect sediment samples from armour layer.

3.12 Flume data and error analysis
The data set generated from the 54 flume experiments is large and a number of software

programs were used for data analysis. The following is a brief summary of the software and
quality assurance procedures taken.

Sontek ADV: scour hole velocity and approach velocity were analyzed using the WinADV

software supplied by Sontek. Velocity data were filtered for correlation, signal to noise ratio

44

and data spikes. Average values for each sampling time were calculated. Velocity data were

filtered for correlations values above 70 (SonTek, 1997). Low correlation values indicate the
output data was dominated by noise and no coherent velocity signals was used in velocity

calculations (Khorsandi et al. 2012). The values of the correlation coefficient is a direct
indication of the random errors produced by Doppler noise for the velocity data (Khorsandi et
al. 2012). Once ADV data is filtered for correlation, signal to noise ratios and data spikes, it

is assumed that velocity measurements are accurate within 0.25 cm/s (SonTek, 2014).

Sontek IQ: approach velocity for each sandbox was calculated using the IQ software statistics

module. According to manufacturer specifications, (SonTek, 2014), the SonTek IQ velocity
is accurate to within +/- 0.5 cm/s and water depth measurement are accurate to within +/0.0003 meters. The measured IQ water level was compared against the recorded staff gauge

level for additional quality assurance. The measured IQ water temperature was compared
against the ADV measured water temperature and was accurate to within 0.2 degrees Celsius.

Scour contours: since scour contours were manually measured with calipers and a measuring
tape, profile measurements were accurate to within 0.01 cm. Contours were manually plotted

on graph paper for determination of x, y, z data points. Standard 1 cm x 1 cm graph paper was

used therefore transcription of plotted data points was subject to error within 0.01 cm. The
Surfer12 plotting software by Golden was used to plot scour contours and calculate scour hole

volume and area.

45

r

3.13 Summary and conclusions
This study required building a hydraulic flume.

Prior to commencing flume

construction, a survey was conducted of the flume literature to investigate present flume

configurations and set-ups in hydraulic laboratories around the world. It was discovered,

similar to conclusions by Nowell and Jumars (1987), that no single flume design was best

suited for this study. Design criteria such as pier width, simulation time, equilibrium scour
and number of simulations were decided based upon a survey of the literature; other design
criteria such as flow velocity and flow depth were governed by the constraints of the flume
infrastructure. Overall, flume construction and design allowed for a total of 54 experimental
runs that operated within the scaling guidelines presented in the flume literature. The next

sections describe results found during the investigation of the impact of ice cover on pier scour.

46

4.0 RESULTS AND DISCUSSION: VELOCITY AND TURBULENT FLOW FIELDS
UNDE^ OPEN AND ICE COVERED CHANNEL

Given that this research required building a flume, this chapter firstly presents the experimental

conditions of the flume. The flow fields, velocity profiles and Reynolds stresses are
investigated under open, smooth and rough channel cover.

4.1 Experimental conditions
The following is the experimental conditions under which the flume was operated. As
indicated in Table 8, each pier size ( 1 1 cm and 22 cm) were tested under two depth ranges,
namely one tailgate configuration (9.5-1 3 cm depth) and two tailgate configurations (21-26 cm

depth). Velocity varied slightly (up to 2 cm/s) for each replicate experiment as velocity was
controlled by opening and closing gravity fed supply valves. Scour depth was investigated
under three channel conditions; open, smooth ice and rough ice cover. Scour depth was also
investigated under three sediment sizes; Dso = 0.58 mm, 0.50 mm, 0.47 mm. A total of 16

flume experiments were completed for each sediment size, examining scour depth under

various channel conditions for each grain size.

47

Table 8. Summary of experimental conditions

Run H

Dso
(mm)

Cover

6
7

0.58

8

0.58

9

0.58
0.58
0.58
0.58
0.58

rough ice

0.58
0.58
0.58
0.58
0.58
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47

rough ice

2
3
4

5

10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36

0.58
0.58
0.58
0.58
0.58
0.58

Water level
(cm)

(cm/s)

open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice

1

Approach
velocity

rough ice
rough ice
rough ice
rough ice

open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice

23.2
23.7
24.0
23.2

9.5
21
13
9.7
24.4
13.3
9
21.6
13.1
9
25.6
13
9.8
20.5
13.5
10
25.5
13.4

downstream

23.5
27.1

9.8

19.4

23.5
22.8
22.6

21.7
21.1
23.3
20.0
22.6
23.3
22.6
21.8
25.0
26.0
18.3
23.5
25.4
18.2

23.3

23.3
20.1
23.3
24.6
22.9
25.1
28.1
21.6
25.1
24.0
21.6

Pier
width

Position in
flume

(cm)

upstream

11

upstream

upstream

11
11
22
22
22
11
11

downstream

11

upstream

downstream

22
22
22

upstream

11

downstream
upstream

downstream
downstream
upstream

downstream

upstream

11

downstream

11

upstream

upstream

22
22
22
11

21.4

upstream

11

13

downstream

9.8
24.3
13.4
9.7
21.5
9.8
21.4
13
12.9

upstream

downstream

11
22
22
22

upstream

11

downstream

downstream

upstream

11

downstream

11

upstream

downstream

22
22
22

9.6

upstream

11

25.8
13

upstream

11
11

12.8
22.0
13.0

downstream

downstream
upstream

downstream

downstream

22
22
22

48

37
38
39

3^2
43
44
45
46
47

0.50
0.50
0.50

open channel
open channel
open channel

23.2
24.1
20.5

11.2
23.1
13.4

upstream

11

upstream
downstream

11

open chanty

23J
22.7
20.2

10

upstream

21.1
13.2

downstream

22
22
22

rw- open channel
0.50

open channel

0.50
0.50
0.50

smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice

48

0.50
0.50
0.50

49
50
51

0.50
0.50
0.50

52
53
54

0.50
0.50
0.50

rough ice

rough ice
rough ice
rough ice
rough ice

22.0
23.3
22.6
22.0
22.5
21.2
22.9
26.0
22.2
22.9
26.3

223

downstream

9.9
24.6

upstream

13.1
9.8
21.6
13.1
9.8
25.9
13.2

downstream
upstream
downstream
downstream
upstream

upstream

11

11
11
11

upstream

22
22
22
11
11

downstream

11

13.0

upstream

22.0
13.0

downstream
downstream

22
22
22

4.2 Flume velocity profiles
The average velocity profiles corresponding to the various experimental conditions
(open water, smooth cover and rough cover conditions) are presented in Figure 23. The

velocity profiles were measured with the 10 MHz ADV under the two tailgate configuration
(21-26 cm flow depth). In Figure 23, the depth of flow on the ordinate axis has been non-

dimensionalized (vertical distance from bed/approach flow depth) to allow for comparisons
between experiments with slightly different flow depths. The approach flow depth is measured

as the vertical distance from the channel bed to the water surface and represented by the letter
H (for the remainder of this thesis, H is used to denote the total flow depth or a fractional

distance from the stream bed to the ice cover). As the ADV measuring volume is located 1 0

49

cm from the probe head, the velocity profile for each channel condition does not extend entirely
to the water surface (Figure 23).

The velocity profiles under open channel conditions are logarithmic with flow velocity
highest closer to the water surface and lowest near the channel bed (Figure 23). A logarithmic
velocity profile is standard for open channel flow. The drag forces exerted on water near the

river bed generally account for the decrease in flow velocity for open channel flow. Of note,
however, since the ADV operates on the principal of a Doppler shift, it may be inferred that

the near bed velocity most likely represents a mixture of both sediment and water velocity.

Even though clear water scour was achieved, it is impossible to achieve water velocity only
measurements within 1 cm of the bed. This was also noted by Muste et al. (2000).

River ice cover imposes an extra boundary on flow, altering the flow velocity and water
level in comparison to open channel flow (Shen and Wang, 1995). For this study, the
maximum velocity for both smooth and rough covered conditions is located approximately

half way between the channel bed and ice cover (Figure 23). Under all Dsos the rough ice
cover produced a larger maximum velocity in comparison to smooth ice cover (Table 9). The
maximum velocity under rough ice cover is 1 7-20 percent larger than that under smooth ice

conditions (Table 9). Zabilansky et al. (2006) also conducted flume experiments comparing
smooth and rough ice cover and found that the maximum velocity for rough ice cover was 20
percent greater than that for smooth ice cover.

The velocity profile under ice conditions depends on the relative roughness of the ice.
Under ice cover conditions, the portion of upper flow is mainly influenced by the ice cover

resistance while the lower flow is mainly influenced by the channel bed resistance (Sui et al.,
2010). The maximum flow is located between the channel bed and ice cover depending on the

50

relative magnitudes of the ice and bed resistance coefficients. Generally, the maximum flow
velocity is closer to the surface with the smallest resistance coefficient. As found by Robert

and Tran (2012), Sui et al. (2010) and Wang et al. (2008) it is expected that as the ice resistance
increases (from smooth ice to rough ice cover), the maximum flow velocity will move closer
to the channel bed. This was also noted by Muste et al. (2000), who measured maximum

velocity under smooth cover at approximately 0.8H, while maximum velocity under rough

cover was approximately 0.6H. Crance and Frothingham (2008) also found that under rough
ice cover the maximum velocity was 2/1 0th’ lower than under smooth ice cover.

For this study, the roughness coefficient for rough and smooth ice cover is 0.02 1 and
0.013 respectively, which are greater than the roughness coefficient for the respective channel

bed Dsos (Table 10). The location of the maximum velocity for all Dsos and ice covers varies

from 0.36-0.43H (Table 9). Contrary to previous findings, the maximum velocity for smooth

ice cover under Dso=O.58 mm is 0.41 H which is lower than the maximum velocity for rough
ice cover at 0.43H (Table 9). However, under Dso=O.5O mm the average maximum velocity

under rough ice cover is 0.36H and for smooth ice cover is 0.41 H (Table 9). Turbulence and
coarse ADV volume measurements may account for the smooth ice cover maximum velocity
to occur closer to the channel bed than the rough ice cover maximum velocity (for Dso=O.58

mm only).

It is important to note that in a natural river system, the velocity profiles under ice cover
would be under constant flux. This is supported by evidence from Crance and Frothingham
(2008). The authors found that river flow erodes the ice underside and generally maintains a

similar texture/roughness across a stream section. Areas of low discharge had rough ice while
areas of high discharge had smoother ice cover. The authors also found that while a cross

51

section would have a consistent roughness, the texture could change from rough to smooth to
back to rough again on a weekly basis. This is an important consideration when modeling or
estimating velocity under ice cover.

Table 9. Maximum velocity and z/H values represented in Figure 23.

Maximum velocity
(cm/s)

Open channel
Smooth ice
cover
Rough ice
cover

Dso = 0.47 mm
Dso= 0.50 mm
Dso = 0.58 mm
D50 = 0.47 mm
Dso = 0.50 mm
D50 = 0.58 mm

27.1
ADV error
23.3
23.3
28.0
28.6
29.2

Stage of maximum velocity
(z/H)
0.47
ADV error
0.41

0.41
0.41
0.36
0.43

Table 10. Roughness coefficients for channel material and ice cover. Associated equations used in calculating
the roughness coefficient can be found in Chapter 5, Section 5.12.

Channel cover/bed material
Channel D50 = 0.58 mm
Channel D50 = 0.50 mm
Channel D50 = 0.47 mm
Rough ice cover
Smooth ice cover

Roughness coefficient

0.0113
0.0110
0.0109
0.021
0.013

52

water surface

1

open channel

a 050=0.58 mm

0.8 .

D50=0.50mm
050=0.47 mm

•

0.6
0.4

0.2
0

5

0

10

15

20

25

30

35

25

30

35

25

30

35

velocity (cm/s)

smooth ice cover

water surface
a D50=0.58 mm

D50=0.50 mm

0.8

•D50=0.47 mm

0.6

0.4
0.2

0
0

15

20

velocity (cm/s)
water surface

rough ice cover

1
a D50=0.58 mm

0.8

D50=0.50 mm
D50=0.47 mm

•

0.6

X

N0.4
0.2
0

0

5

10

15

20

velocity (cm/s)
Figure 23. Velocity profiles for Dso = 0.58 mm, 0.50 mm, 0.47 mm under open channel, smooth cover and
rough cover channel conditions. Velocity profiles were measured with the 10 MHz ADV and presented under
the two tailgate configuration (21-26 cm flow depth). ADV measurements are accurate to +/- 0.25 cm/s.

53

4.3 Flume turbulent intensity
Since an ice cover imposes an added boundary on flow conditions the flow velocity

distribution and turbulent intensity is different under an ice covered channel than that of open
conditions. As discussed in the previous section, the near bed flow velocity under ice covered
conditions is higher than that of open channel conditions. As the near bed velocity increases,
the kinetic energy exerted on the bed also increases which can have implications for sediment
transport hence why it is being discussed here. The turbulent kinetic energy indicates the

energy that is extracted from turbulent eddies and is expressed as (Clifford and French, 1 993):

TKE = 0.5(RMSx2 + RMSy2 + RMSZ2)

4.1

The turbulent kinetic energy can be decomposed by examining the turbulent intensities for the

streamwise turbulent intensity (RMSX), the lateral turbulent intensity (RMSy) and vertical
turbulent intensity (RMSZ) components. The turbulence intensity values were determined from
the standard deviation of instantaneous velocity fluctuations (RMS or root mean square). The
RMS values were calculated based upon the velocity profiles measured in the approach flow

section of the flume.
Figure 24 and Figure 25 show the turbulent intensities (RMS) for the streamwise (x)

and vertical (z) flow components. The lateral turbulent component (y) was not presented as

fluctuations for all measurements did not exhibit any meaningful patterns; this is similar to
other studies reporting turbulent intensities (Faruque, 2009; Muste et al., 2000; Robert and
Tran, 2012). The turbulent intensities were normalized by the shear velocity in order to make
a dimensionless comparison.
54

As indicated in Figure 24, the streamwise turbulent intensity is a maximum just above

the channel bed for rough ice cover and open channel conditions. Muste et al. (2000) suggests
that this is due to the increased turbulence of sediment movement near the bed. The turbulent

intensity reduces towards the surface for both open and rough cover conditions.

Of note,

Robert and Tran (2012) reported the streamwise and vertical turbulent intensities for rough

cover as maximizing just above the channel bed but also at the water surface. In this study
however, measurements were taken to a point 10 cm below the water surface due to the
measuring volume of the ADV. In Robert and Tran (2012) they used a micro-ADV which had
measuring capabilities of 5 cm below the sensors.

DS0=0.58 mm

D5o=O 47 mm

0.7

0.7

Oopen
A rough

0.6

p

<A

0.5

a
X

o

X

O A

0.4

OA

p

O

0.3
OA

0.2

A

°A

0.2
OA

O

0.1

0.1
O

A rough

OA

OA

0.5

z/H

oopen

0.6

A

£

A

A

0

0

0

0.2

RMSX/U.

0.4

0

RMSX/U.

0.2

0.4

RMSX/U.

Figure 24. Spatially-averaged profiles of turbulent intensity for the streamwise (x) velocity component,
normalized by the shear velocity (U.), for open water and rough ice cover conditions.

55

D50=0.47 mm
0.7

Oopen

a rough

0.6

Oa
CM

0.5
X

O*

0.4
0

0.3

o

0.2
0

A
A

0.1

o

A

0
0

RMS/U.

RMS/U.

0.1

0.2

RMSZ/U.

Figure 25. Spatially-averaged profiles of turbulent intensity for the vertical (z) velocity component, normalized
by the shear velocity (U,), for open water and rough ice cover.

The streamwise turbulent intensities for the rough flow were greater than those for open
channel conditions (Figure 24).

Faruque (2009) also found that streamwise turbulent

intensities for rough ice cover were greater than those for open channel conditions. The
vertical turbulent intensities are approximately one half than the streamwise turbulent
intensities (Figure 24 and Figure 25). This indicates the flow is anisotrophic (directionally
dependant as opposed to isotrophic- which would indicate identical measures of flow in all
directions). This indicates that the turbulence is mainly associated with fluctuations in the

streamwise velocity (Muste et al., 2000). There is no significant observed effect between open
channel and smooth cover on both the streamwise and vertical turbulent intensity for the entire

flow profile (Figure 26). For this reason, data points for each Dso were plotted on one graph
rather than separated out.

56

The turbulent intensities are of interest when considering sediment transport as the
strength of the turbulence will affect sediment suspension (Muste et al., 2000). A greater

turbulence and therefore more kinetic energy under rough ice cover infer that there is a greater

capacity for sediment transport to occur.
Flow reversal occurs in the wake region downstream of the pier which can also

contribute to the decrease in sediment ridge size. Also, the turbulent kinetic energy behind the
pier is isotropic and stronger than the turbulence upstream of the cylinder (Graf and Istiarto,
2003).

Figure 26. Spatially-averaged profiles of turbulent intensity for the vertical (z) and streamwise (x) velocity
component, normalized by the shear velocity (U,), for open water and smooth ice cover conditions.

57

&

4.4 Reynolds stresses

The Reynolds stress is defined as

-w = - jEJUa, - ux) (a, - u,)

42

where ux and uz are the streamwise and vertical turbulent fluctuations, Ux and U2 are the time
averaged streamwise and vertical velocities over n measurements and ux and uz are the
instantaneous streamwise and vertical velocity components. As the Reynolds number is a

measure of the balance in inertial to viscous forces, the Reynolds shear stress corresponds to a
balance of mean linear momentum (Speziale, 1990) and is important in examining turbulent

flows and sediment transport.

In general, there were only slight differences in the Reynolds stress values for open,
smooth and rough covered flows (Figure 27 and Figure 28). Figure 27 indicates that in the
upper portion of flow the Reynolds shear stress is smaller under rough cover in comparison to

open channel conditions; however, of note the ADV sampling volume was 1 0 cm from the
probe. Values of Reynolds shear stress obtain a maximum value closer to the channel bed in
comparison to open channel conditions (Figure 27). Values of Reynolds shear stress gradually

become smaller towards the surface for rough ice cover. The values of Reynolds shear stress
do not show any significant differences in pattern between open channel and smooth cover
conditions (Figure 28). For all sediment sizes, Reynolds stresses are greater in the near bed
region under rough ice cover conditions.
Faruque (2009) found that surface roughness increases the contribution of turbulent
events which produces a greater change in Reynolds shear stress distributions which can

58

potentially influence sediment transport. The shear stress distribution changes the most under
rough cover for D5o=O.47 mm (Figure 27).

With such small scale change in Reynolds shear stress and a sampling volume 10 cm
away from the ADV probe, there is no clear trend as to whether smooth and rough ice cover

significantly altered the Reynolds shear stress distribution over the flow profile.

59

r
t

DS0=0.58 mm
0.03
Oopen

-

A

.0.02

A rough
A
A

*>

°A

°

A

A O

A
O

o

O
A
A

A

0.01

0.00
0

1

1

4

1

1

1

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Z/H

DS0=0.50 mm

0.03

„

O open
A rough
A

0.02

o

j>
3

0.01

-

o

O

A

A

O

O

A

A

0.5

0.6

0 00

0

0.1

0.2

0.3

0.4

0.7

Z/H

D50=0.47 mm

0.03

O open
A rough

O
n

0.02

ZD
O

J>

3

O

O

O

0.01
O

O

O

0.00

0

0.1

0.2

0.4

0.3

0.5

0.6

0.7

Z/H
Figure 27. Reynolds shear stress, normalized by the shear velocity, for open and rough cover channel
conditions.

60

Dso=O.S8 mm

0.03

Oopen

x smooth
« 0.02

o
v
X

j>

x

°

°

o
x

3 0.01

°
x

X
X

0.00
0

0.3

0.2

0.1

0.4

0.6

0.5

z/H

0.7

Dso=O.5O mm
0.03

Oopen

x smooth

X
cm

0.02

o

O

X

O

X

3 0.01

o

0.00
0

1

1

1

0.1

0.2

0.3

x

Ox

»

Z/H

1

0.4

0.6

0.5

Dso=O.47 mm

0.03

o

O open

x smooth

~ 0.02
o

j>
3 o.oi

0.7

-

O

o*

X

X

O

o

o

°

X

X
X

0.00
0

0.1

0.2

0.3

0.4

0.5

0.6

0.

z/H
Figure 28. Reynolds shear stress, normalized by the shear velocity, for open and smooth cover channel
conditions.

61

4.5 Summary and conclusions
The purpose of this chapter was to examine the experimental hydraulic conditions

under open, smooth and ice covered conditions. In doing so the velocity profiles, turbulent
intensities and Reynolds shear stresses were examined. All data for this chapter were derived

from ADV measurements for two tailgate runs (2 1 -26 cm depths) at Dso = 0.47 mm, 0.50 mm
and 0.58 mm under open, smooth and rough channel conditions. The main limiting factor in

some measurements was the fact that the ADV flow volume was 10 cm from the probe end,
which did not allow for flow measurements immediately adjacent to the ice covers. While
previous studies by Crance and Frothingham (2008) and Sui et al. (2010) have measured flow

profiles under ice cover, this study builds upon work by Muste et al. (2000) and Robert and

Tran (2012) by decomposing flow profiles and examining turbulent intensities which have
important implications for sediment transport. As one of the main objectives of this thesis is
to examine pier scour under ice cover, the causal mechanisms for sediment transport are

important. The main findings from this chapter are summarized as follows:

(1) The average maximum velocities under rough ice cover were 0.36-0.43H and for smooth

ice cover were 0.41H. The range in rough cover velocity compared to smooth cover is most
likely due to the greater turbulent intensity measured under rough ice cover conditions.
Smooth and rough ice cover causes the maximum velocity to migrate towards the channel bed.

A higher velocity close to the channel bed indicates a greater capacity for sediment transport.
There was not a consistent difference between the height of the maximum velocities between
smooth and rough ice cover.

62

(2) The streamwise turbulent intensities for the rough flow were greater than those for open

channel conditions and the vertical turbulent intensities were approximately one half than the
streamwise turbulent intensities.

The strength of the turbulence will affect sediment

suspension; a greater turbulence and therefore a greater kinetic energy under rough ice cover
infer that there is a greater capacity for sediment transport under rough ice cover.

(3) There were only slight differences in the Reynolds stress values for open, smooth and rough

covered flows. For all sediment sizes, Reynolds stresses were slightly greater in the near bed
region under rough ice cover conditions. With such small scale change in Reynolds shear stress

there is no clear trend as to whether smooth and rough ice cover significantly altered the
Reynolds shear stress distribution over the flow profile.

63

5.0 RESULTS AND DISCUSSION;

S<;OUR CHARACTERISTICS UNDER ICE COVER

The purpose of this chapter is to examine various hydraulic parameters as they relate
to pier scour. Experimental flume results are presented relating pier scour depth to pier size,

median size of bed material and channel cover. The scour area and scour hole morphology are
also examined along with velocity profiles within the scour hole. Since non-uniform sediment

was used, the armour layer within the scour hole is also investigated. Dimensional analysis is
presented for hydraulic parameters as they relate to scour depth and the validity of multiple
regression analysis is investigated. Also, since the flume was subject to scaling, information

is provided as to how scaling effects were addressed in applicable sections.

5. 1 Scour depth versus sediment grain size
Open channel:

Scour depth is compared under various grain sizes under the same flow and depth for
open channel conditions. Results indicate that the largest scour depth occurs under smallest

D50 while the smallest scour depth occurs under the largest D50 for all cover conditions (Figure
29). On average scour depths for open channel conditions are 50 percent larger for D5o=O.47

mm and 13.64 percent larger for Dso=0.50 mm compared to the largest sediment grain size

D5o=O.58 mm. Of exception is the results measured for Dso=O.58 mm for runs 1 and 6 (Figure
29); the measured scour depth is larger than the scour depth for D5o=O.5O mm for the same

runs. These two results do not align with the overall pattern of smaller scour depths with
increasing D50 size. It is difficult to determine why in this case there was a deeper scour hole

64

for D5o=O.58 mm than for Dso=0.50 mm. The average armour layer size for these runs is very
similar, varying by only 0.1 mm and velocity, varying by only 0.9 cm/s (Appendix A and C).

A possible explanation for a smaller scour depth under Dso=0.50 mm may be due to the extent

of the embedded depth of the armour layer grains for that specific run. Since non-uniform
sediment was used, during each experimental set up, the bed was leveled around the pier; in
doing so larger grains may have been raked around the pier which may have become embedded

in the scour hole. Also, perhaps fluctuations in near bed turbulent intensity were slightly
greater for D5o=O.58 mm leading to greater sediment transport around the bridge pier.

Figure 29. Maximum scour depth under open channel conditions for Dso=O.47 mm, Dso=O.5O
mm, Dso=O.58 mm. Numbers 1-6 indicate experimental run number. For associated data values
refer to Appendix B. Scour depth measurements subject to +/-0.02 cm error.

Smooth Cover:

Scour depth is compared under various grain sizes for the same flow and depth under

smooth channel conditions. For all experimental runs, the smallest grain size resulted in the
largest scour depth under smooth ice cover (Figure 30). The average scour hole depth for

Dso=O.47 mm is 6.33 cm, D5o=O.5O mm is 3.66 cm and Dso=O.58 mm is 3.16 cm. On average,
65

maximum scour depth for D5o=0.47 mm is 56 percent greater than maximum scour depth for

Dso=O.58 mm. The maximum scour depth for D5q=0.50 mm is 22.77 percent greater than that

Figure 30. Maximum scour depth under smooth ice channel conditions for D5o=O.47 mm, Dso=O 50
mm, Dso=O.58 mm. Numbers 1-6 indicate experimental run number. For associated data values refer
to Appendix B. Scour depth measurements subject to +/-0.02 cm error.

Rough Cover:

Scour depth is compared under various grain sizes under the same flow and depth for
rough ice cover channel conditions. Just as with open and smooth channel conditions, results

indicate that largest scour depth occurs under smallest D50 while the smallest scour depth
occurs under the largest D50 for rough cover conditions (Figure 31). A reduction in grain size

from 0.58 mm to 0.47 mm produces a 58 percent increase in maximum scour depth (under the
same flow conditions). On average scour depths for rough ice cover channel conditions are 33
percent larger for Dso=O.5O mm compared to the largest sediment grain size Dso=O58 mm.

66

Figure 31. Maximum scour depth under rough cover channel conditions for D5o=O.47 mm,
D$o=050 mm, Dso=0.58 mm. Numbers 1-6 indicate experimental run number. For associated data
values refer to Appendix B. Scour depth measurements subject to +/-0.02 cm error.

Summary of scour depth versus grain size:
When comparing scour depth to grain size the scaling of the experimental flume must

be considered. Melville and Coleman (2000) stated that if the ratio of pier width to flow depth
(b/h) is between 0.7-5.0 then local pier scour is dependent on both flow depth and pier width

while if b/h is less than 0.7 than local pier scour depth is dependent only on pier width. For
this thesis, the ratios of pier width to flow depth range from 0.42 to 2.2. In order to work
around these scaling issues, maximum scour depth was compared between covers and

sediment size while keeping pier width and flow depth constant for each corresponding run.

In keeping all hydraulic parameters constant, the smallest sediment size D5o=O.47 mm yielded
the largest pier scour depth under all channel covers. The largest scour depth occurs under
smallest D50 while the smallest scour depth occurs under the largest D50. It is interesting to
note that even though the bed sediment sizes only differ by up to 11 mm the maximum scour

67

depths vary by up to 58 percent. For example, under rough cover the maximum scour depth
is 38 percent larger for Dso=O 47 mm in comparison to Dso=O.5O mm, even though the
difference in sediment size is only 3 mm.

It should also be noted that the choice of sediment size in laboratory models, and in
this thesis, distorts the value of pier width to sediment size (b/Dso) in comparison to the
prototype (real world bridge). In reality, bridge piers are much wider than their model

counterparts, yet bed material typically remains within the same diameter ranges as used in

flumes. As examined by Lee and Sturm (2009) this scaling issue leads to larger scour depths

in the laboratory than in the field. Further discussion on scaling issues regarding the flume
experiments undertaken in this thesis can be found in Chapter 6, Section 6.1.1.

5.2 Scour depth versus pier size
When analyzing local scour around bridge piers the scaling ratio between flow depth

and pier width must be considered. As outlined by Melville and Coleman (2000), b/h <0.7

indicates that scour depth is directly proportional to pier width. Given 0.7<b/h<5, scour depth
is proportional to both pier width and flow depth. Therefore when comparing scour depth and
to pier size, a subset of the experimental runs were selected: namely those runs where b/h >

0.7. For each comparison flow depth and sediment size were the same.

For the majority of experiments under the same hydraulic conditions a larger maximum
scour hole was produced for the 22 cm pier (Figure 32, Figure 33, and Figure 34). For run 6
under open channel conditions the maximum scour depth for the 1 1 and 22 cm piers was the

same. This also occurred for runs 2, 4 and 5 under smooth ice cover (Figure 33). On average,
the maximum scour depth for the 22 cm pier was 25 percent larger than the 1 1 cm pier under

68

open channel conditions. The maximum scour depth for the 22 cm pier was 1 2 and 29 percent
larger than the 1 1 cm pier for smooth and rough channel conditions respectively.

Since scouring is caused in part by the horseshoe vortex and the dimensions of the
vortex system are a function of pier diameter, it is concluded that local pier scour is related to

pier size (Breusers et al., 1977). Shen et al. (1969) also found that the vortex is proportional
to the pier Reynolds number, which is a function of pier diameter. With a larger horseshoe
vortex, the sediment transport capacity is greater hence the larger maximum scour depth under

all conditions for the larger 22 cm pier.

Figure 32. Maximum scour depth under open channel conditions for 1 1cm and 22 cm width pier. Numbers 1-6
indicate experimental run number. Scour depth measurements subject to +/-0.02 cm error.

69

Smooth Cover

Figure 33. Maximum scour depth under smooth channel conditions for 1 1 cm and 22 cm width pier. Numbers 1 6 indicate experimental run number. Scour depth measurements subject to +/-0.02 cm error.

Rough Cover

Figure 34. Maximum scour depth under rough channel conditions for 11cm and 22 cm width pier. Numbers 1-6
indicate experimental run number. Scour depth measurements subject to +/-0.02 cm error.

70

I

5.3 Scour depth versus channel cover
As previously discussed, under the scaling conditions of the flume, scour depth
automatically depends on both flow depth and pier width. Experimental runs with a pier width
to flow depth (b/h) ratio of less than 0.7 were not considered for comparison in this section as

scour depth at a 0.7 ratio is only dependant on pier width and not flow depth (Melville and
Coleman, 2000). In addition, for some experiments the flow velocity was slightly greater than

2 cm/s between experimental runs which will increase variability in sediment transport. This
was a result of the limitations in the gravity fed water system and differences in the flow
velocity between the downstream and upstream sandboxes.

Under all sediment sizes rough ice cover scour depth is greater than open channel

conditions for nine out of 15 experimental groupings (60 percent). Rough cover scour depth

is equal to open channel scour depth for 2/15 experiments, and rough cover scour depth is less
than open channel scour depth for 4/15 experiments (Figure 35, Figure 36 and Figure 37). On

average, for all sediment sizes, rough ice cover scour depth is 37 percent greater than open
channel scour depth (Table 11, Table 12, Table 13). Of note, the smaller the D50, the greater
the difference between ice cover scour depth and open channel scour depth. Under D50S of

0.47, 0.50 and 0.58 mm the scour depth under rough ice cover was 46, 33 and 29 percent
greater than open channel scour depth respectively (Table 11, Table 12, Table 13). Smooth

ice cover produced a greater scour depth than open channel conditions for eight out of 1 5
experiment groupings (53 percent). On average, for all sediment sizes, smooth ice cover scour
depth is 20 percent greater than open channel scour depth (Table 11, Table 12, Table 13).
Smooth ice cover scour depth is equal to and less than open channel scour depth for 3/1 5 and
4/15 experiments respectively. The findings presented here, indicating that ice cover does not

71

always produce a greater pier scour depth, are different than the two previous studies in the

literature.

Batuca and Dargahi (1986) and Olsson (2000) found that scour depth under smooth
and rough cover is greater than scour depth under open channel conditions for all experimental

runs. Batuca and Dargahi ( 1 986) compared local scour around a pier under open water and
floating ice cover conditions. A total of 34 experiments were performed under ice conditions

using plywood and aluminum for channel cover. Sediment with a median diameter of

Dso=O.41 mm and og=1.88 (non-uniform sand) was used. Through visual observation they
found that the general size of the scour hole was greater under ice covered conditions. Olsson
(2000) compared pier scour depths under open, smooth and rough ice cover and found that the

rough ice cover resulted in a greater scour depth than the smooth cover. Uniform sand with

Dso=O.42 mm was used along with Styrofoam sheets to simulate ice cover. Experiments were
run for 4 hours. Olsson (2000) found that ice cover may increase the local scour depth by 25-

35 percent compared to open channel conditions.
The differences found in this study compared to Olsson’s (2000) findings are most
likely due to the sediment types used. This study used non-uniform sand and Olsson (2000)

used uniform sediment. The mechanisms involved in non-uniform sediment transport are
complex as grain shape, grain distribution and interactions between grains all influence
incipient motion of sediment inside the scour hole (Xu et al., 2008). Wu et al. (2000) examined

non-uniform sediment transport in alluvial rivers and developed a correction factor for the
hiding and exposure factors associated with non-uniform sediment. The hiding and exposure

factors of non-uniform sediment are stochastically related to the size and gradation of bed
materials. Given that between each experiment the sediment in the flume was raked and

72

leveled, the arrangement of particles would be different for each experimental run. The down

flow and associated vortices at the bridge pier would therefore transport sediment depending

on the hiding and exposure factor of individual sediment grains (since after leveling the flume
sediment would be arranged in a different way for each experiment). The micro environment
around each bridge pier would be different in terms of turbulence since the fluid interactions
with grains are different. This is reinforced by evidence of the large variability in the average
armour layer size within each scour hole (Table 1 1, Table 12, Table 13). Also, the flume was

2 m wide with 11 .3 m of flow between the head tank and first sandbox. Even though a flow

diffuser was installed, each experimental run would have slightly different approach velocities,
turbulence, and water temperature. (Water temperature fluctuated from 8.6-10.4 degrees

Celsius between each run since the flume was outside. This would change parameters such as

the water viscosity). Given that the arrangement of sediment, flow velocity, turbulence and
water temperature were different for each experimental run the sediment transport potential

around the bridge pier would also different for each experiment. The variation in scour depth
with channel cover for this study may be a result of the variation in sediment and associated
transport mechanisms as discussed above.

Of note however,. Batuca and Dargahi (1986), found that ice cover produced a greater
general size of scour hole compared to open channel conditions all the while using non-

uniform sand. The authors did not physically measure the maximum scour depth but rather
visually inspected the scour hole in drawing their conclusion. The sediment used in Batuca
and Dargahi (1986) closely aligns with the D5o=O.47 mm sand used in this study. For the

Dso=O.47 mm sediment, this study found that rough ice cover produced a greater scour depth
for three out of five experimental groupings (Figure 37). The scour volume and area for each

73

scour hole was also measured and correlated with maximum scour depth (as discussed in the
next section, Figure 41). As indicated in Figure 41, using the scour volume and area (or

visually estimating the general size of the scour hole) is not always a good indication of
maximum scour depth. In order to compare results between this study and Batuca and Dargahi
(1986) actual measurements of maximums scour depth are required. Also, using plywood and

aluminum for channel cover would create a different pressure at the water surface compared
to using Styrofoam. It is assumed that plywood and aluminum would be much more rigid

surfaces than Styrofoam and have less buoyancy.
As previously mentioned, this study resulted in rough ice cover producing greater pier

scour depths for 9/15 experimental groupings. These results partially align with findings from

Batuca and Dargahi (1986) and Olsson (2000). Rough ice cover produces greater down flow
then open channel conditions (as discussed in detail in section 5.8, Figure 56 and Figure 57),
and therefore under rough ice cover there is a greater capacity for sediment transport around

the bridge piers. While 40 percent of the data for this study did not align with previous
research, it is important to consider the characteristics of non-uniform sediment and that even

in an experimental flume environment hydraulic parameters are not constant. Heterogeneous

environments are more representative of natural rivers: highly variable with non-uniform
sediment. While it has been previously found that ice cover produces greater down flow

leading to greater pier scour, this study indicates that while ice cover does produce greater

down flow, sediment uniformity is also important in considering pier scour.

74

Figure 35. Maximum scour depth under open, smooth and rough channel condition for Ds<r=0.58 mm. Numbers
1 -6 indicate experimental nm number. Scour depth measurements subject to +/-0.02 cm error.

Table 11. Scour depth values associated with Figure 35.

Experimental run
# Dso = 0.58 mm

open

Scour
depth
(cm)

Dso
armour
layer

smooth
Scour
D50
armour
depth
(cm)
layer

(mm)

1
2
3
4
5

5
2
5.5
1
3
average

0.60
1.05
0.62
0.65
0.78
0.74

4.5
1.5
5.5
2
1.5
average

(mm)
1.25
0.85
1.81

0.83
0.94
1.14

rough
Scour
D50
depth
armour
(cm)
layer
(mm)

5
1
6
2
1.5
average

1.25
1.42
1.60
0.85

1.57
1.34

75

Figure 36. Maximum scour depth under open, smooth and rough channel condition for D5o=0.50 mm. Numbers
1-6 indicate experimental run number. Scour depth measurements subject to +/-0.02 cm error.

Table 12. Scour depth values associated with Figure 36.

Experimental run
# Dso = 0.50 mm

open

Scour
depth
(cm)

1
2
3
4
5

4.5
2
6
3
2
average

D50
armour
layer
(mm)
0.60

smooth
Scour
D50
depth
armour
(cm)
layer

rough
Scour
D50
armour
depth
(cm)
layer

(mm)

(mm)

0.80
1.43
0.60
0.68

6
2.25
6
2
3

0.82

average

1.40
0.59
0.68
0.79
0.65
0.82

6
3
7
3
5
average

0.75
0.73
0.60
0.59
0.73
0.68

76

Figure 37. Maximum scour depth under open, smooth and rough channel condition for Dso=0.47mm. Numbers
1-6 indicate experimental run number. Scour depth measurements subject to +/-0.02 cm error.

Table 13. Scour depth values associated with Figure 37.

Experimental run
# D50 = 0.47 mm

open
Scour
depth
(cm)

2
3

8
4
11

4

4

5

5

1

average

D50
armour
layer
(mm)
0.48
0.50
0.59
0.43
0.45
0.49

smooth
Scour
D50
armour
depth
(cm)
layer

rough

Scour
depth
(cm)

(mm)

8.5
5
12
4.5
5
average

D50
armour
layer
(mm)

0.40
0.49
0.55
0.30

6
7
9
8

0.53
0.45

9

0.55
0.45
0.57
0.49
0.48

average 0.51

77

5.4 Scour area and volume
Much of the current literature focuses on the value and location of maximum scour
depth around piers, however for purposes of bridge construction it is also important to consider

the scour volume and scour area. Riprap and flow dispersal devices are common pier scour
countermeasures (Beg and Beg, 2013) and it is essential to understand potential pier scour area

and volume in countermeasure design.

Scour hole contours were manually measured and input into Surfer 1 2 plotting software.
The Surfer12 software was then used to calculate scour hole volume and area for each
experimental run. The scour area (A) versus scour volume (V) was plotted in Figure 38, Figure

39 and Figure 40. Based upon these figures the following three relationships were developed:

For open channel:

Vncmpier = 3.93A - 1787.5

5.1

^22 cm pier = 3.81A — 2731

5.2

vu cm pier = 3.95A- 1367.4

5.3

v22 cm pier = 4.88A- 5054.4

5.4

Vncmpier = 2.63A - 733.36

5.5

^22 cm pier = 3.84A - 2835.6

5.6

For smooth cover:

For rough cover:

78

It should be noted that the above equations are only practical for piers under the flume’s
scaling ratios. In all scenarios, there is a linear relationship between scour volume and area.
Only two previous studies have looked at scour volume and area around bridge structures

(Khwairakpam et al., 2012; Wu et al., 2014). Wu et al. (2014) found that there was a linear
relationship between scour depth and volume around bridge abutments under ice cover.
Khwairakpam et al. (2012) found an increasing relationship between flow depth, densimetric

Froude number and scour hole geometry. The main conclusion drawn from this study is that
the linear relationship between scour area and volume remains valid even under ice cover

conditions. Also, scour area and volume increase with pier size.
Maximum scour depth was also plotted against scour volume and area (Figure 41). As
indicated by the R values, there is not a strong correlation between maximum scour depth,
scour area and scour volume. This study found that while the maximum scour depth typically

occurs at the front of the pier, the remainder of the scour hole can be shallow and contain areas
of sediment deposition. This is discussed more in the next section outlining scour patterns. It
is important to understand scour depth is not a good indicator of scour area or volume when
considering riprap for a scour countermeasure.

Also of note, using scour area as a measure of scour depth brings into question findings
by Batuca and Dargahi (1986): the authors found that ice cover produced a greater general size

of scour hole compared to open channel conditions all the while using non-uniform sand. As

found in this study, a larger scour area may not always correlate to a larger maximum scour
depth.

79

Figure 38. Variation of scour volume and area around bridge pier under open channel condition.

Figure 39. Variation of scour volume and area around bridge pier under smooth ice cover condition.

80

rough

area (cm2)
Figure 40. Variation of scour volume and area around bridge pier under rough channel condition.

(cm2)

area

scour

maximum scour depth (cm)
maximum scour depth (cm)
Figure 41. Maximum scour depth as related to the scour area and scour volume.

81

5.5 Scour patterns under open, smooth and rough ice cover
Scour hole contours were manually measured and input into Surfer 1 2 plotting software.
Surfer12 was used to generate 2-D and 3-D scour patterns in order to gain a better
understanding of sediment transport patterns around bridge piers under ice cover. The 2-

dimension contour plots, Figure 42, Figure 44 and Figure 46 show the scour depth
characteristics for 1 1 and 22 cm piers. Negative or dark shaded contour values indicate erosion
while lighter or positive values indicate deposition. Figure 43, Figure 45 and Figure 47
illustrate 3-dimension scour patterns produced by both pier diameters (b = 1 1 cm and 22 cm)
under open, smooth and rough cover channel conditions.
Across all three Dsos, scour patterns were similar between pier sizes and channel cover;

therefore a subset of plots were chosen to display typical scour and deposition patterns. All
plots, Figure 42 to Figure 47, present scour patterns for Dso=0.47 mm, b = 1 1 and 22 cm, h =

0.21-0.26 m (two tailgate configuration) and U = 0.2 1 -0.28 m/s. Photographs and scour plots
were produced for each experimental run; however for the sake of avoiding repetition, scour
plots presented here are a representative subset of the entire 54 run data set.
The following is a summary of scour characteristics based upon Figure 42 to Figure
47. Regardless of cover conditions, the location of maximum scour depth is always at the pier

face. This phenomenon indicates that maximum scour depth is not overly influenced by
contraction scour (otherwise maximum scour would occur along the pier sides). The scour
depth decreases towards the downstream side of the pier and behind the pier is typically an
area of deposition. Under all cover conditions, at lower velocities, (U< 0.22 m/s) scour is
relatively limited downstream of the pier. Under higher velocities the scour hole extends
around the entire pier circumference. Under all channel covers deposition downstream was

82

1

greater for the 22 cm pier and a distinct deposition ridge was formed downstream of the pier.

The deposition ridge was most pronounced under open channel conditions and least defined
under rough cover conditions. The decreased size of the sediment ridge for ice covered flows
is most likely caused by greater turbulent intensity under ice cover, which would lead to

increased sediment entrainment downstream of the pier.

83

flow direction

open, 11 cm pier, h = 0.21 m, U = 27 m/s

open, 22 cm pier, h = 0.24 m, U = 0.23 m/s

c
Figure 42. Scour pattern for 11 cm and 22 cm pier in units of centimeters, D50 = 0.47 mm, under open channel
conditions.

84

F

open, 11 cm pier, h = 0.21 m, U = 27 m/s

open, 22 cm pier, h = 0.24 m, U = 0.23 m/s

Figure 43. Scour patterns for 11 cm and 22 cm pier, Dso = 0.47 mm, under open channel condition.

85

flow direction

smooth, 11 cm pier, h = 0.22m, U = 0.23 m/s

smooth, 22 cm pier, h = 0.26 m, U = 0.25 m/s

cm
Figure 44. Scour pattern for 11cm and 22 cm pier, Dso = 0.47 mm, under smooth channel cover.

86

smooth, 11 cm pier, h = 0.22m, U = 0.23 m/s

smooth, 22 cm pier, h = 0.26 m, U = 0.25 m/s

Figure 45. Scour pattern for 11 cm and 22 cm pier, D50 = 0.47 mm, under smooth ice cover channel condition.

87

rough, 11 cm pier, h = 0.21 m, U = 0.28 m/s

-15

-10

-5

0

5

10

15

20

25

30

35

40

45

50

55

rough, 22 cm pier, h = 0.23 m, U = 0.21 m/s

cm
Figure 46. Scour pattern for 11 cm and 22 cm pier, Dso = 0.47 mm, under rough ice cover channel condition.

88

rough, 11 cm pier, h = 0.21 m, U = 0.28 m/s

rough, 22 cm pier, h = 0.23 m, U = 0.21 m/s

Figure 47. Scour pattern for 11 cm and 22 cm pier, D50

— 0.47 mm, under rough ice cover condition.
89

!

rough, 11cm pier, h = 0.21 m, U = 0.28 m/s

rough, 22 cm pier, h = 0.23 m, U = 0.21 m/s

Figure 47. Scour pattern for 1 1 cm and 22 cm pier, Dso = 0.47 mm, under rough ice cover condition.

89

5.6 Scour profiles
The scour holes from each experiment were manually contoured in centimeter

increments. The scour profiles, Figure 48, Figure 49 and Figure 50 illustrate the side view of
the scour hole for representative experimental runs. The purpose of this section is to illustrate

scour and deposition patterns under open, smooth and rough cover conditions. Scour profiles
were measured for each experimental run; however for the sake of avoiding repetition, scour

profiles presented here are a representative subset of the entire 54 run data set.
The following is a summary of scour and depositional morphology around the 1 1 and

22 cm piers as shown in Figure 48, Figure 49 and Figure 50. For all the cross sections the
maximum scour depth is located at the upstream face of the pier at an angle of 90 degrees to
approach flow. This is due to the downflow and resulting horseshoe vortex within the scour

hole. Dey and Raikar (2007) found that within a scour hole there is a higher magnitude of
turbulence that increases as the scour hole grows larger. The horseshoe vortices along with
turbulent intensities located within the scour hole at the pier face are responsible for the
location of the maximum scour depth. Additionally, as previously discussed, the smaller the
grain size the larger the maximum scour depth as indicated by the differences in the profiles
between the three Dsos.

There is also a decrease in scour depth located on the downstream side of the pier. As
flow passes the sides of the pier, the flow separates and wake vortices form. The wake vortices

are transferred downstream by the approach flow and are responsible for transport of sediment
that is already entrained by the downflow and horseshoe vortex (Melville and Coleman, 2000).

The strength of the wake vortices are typically less than the horseshoe vortices and therefore
cannot transport the same sediment load as the horseshoe vortex. Since the wake vortices are

90

weaker behind the pier, sediment deposition may occur downstream of the pier. This is most
evident in Figure 50 for the 22 cm pier. The amount of scour or deposition that occurs in front
and behind the bridge pier is a balance between the strength of the horseshoe vortices and wake
vortices. A larger pier size, flow depth, approach velocity and rougher channel cover all lead
to increased scour depth. As scour depth increases, there is more sediment for the wake

vortices to transport away from the pier base.
Scour mitigation at bridge sites often involves the placement of riprap around the pier

base as a scour countermeasure. Riprap can act as a resistant armour layer to the hydraulic
shear stress and provide protection to the smaller credible sediments underneath (Deng and
Cai, 2010). Lauchlan (1999) found that placing riprap at a specified depth below the bed

surface improved the performance of the riprap layer. For this study, the average scour hole
slope ranged from 22-25.5 percent (Table 14), with a standard deviation of up to 13 percent.
As bed-form undermining (where the riprap layer is undermined) is a dominant failure

mechanism (Lauchlan, 1999) it is important to understand the ranges in possible scour hole
slope for successful installation of riprap. Also highlighted in this section was that the

maximum scour depth occurs at the upstream pier face under all channel covers which is also
important for successful installation of riprap as a scour countermeasure.

91

-60-50-40-30-20-10 0 10 20 30 40 50
cm
cm
Figure 48. Scour profile for 1 1 (left) and 22 (right) cm pier under open, smooth and rough channel conditions
for Dso=O.47 mm. Scour depth measurements subject to +/-0.02 cm error.

‘60 '50 ’40 ‘30 '20 10
10 20 30 40 50
-60 -50 -40 -30 -20 -10 0 10 20 30 40 50
cm
cm
Figure 49. Scour profile for 1 1 (left) and 22 cm (right) pier under open, smooth and rough channel conditions
for Djo=O.5O mm. Scour depth measurements subject to +/-0.02 cm error.

°

92

cm
cm
Figure 50. Scour profile for 1 1 (left) and 22 (right) cm pier under open, smooth and rough ice conditions for
Dso= 0.58 mm. Scour depth measurements subject to +/-0.02 cm error.
Table 14. Slope of scour hole at upstream pier face for each run.

slope (%), 11 cm pier
slope (%), 22 cm pier
open
20.4
24.0
16.4
11.2
25.81
17.7
mm
smooth
9.1
20.4
26.5
26.1
6.25
12.5
Dso=O.58
rough
30.0
21.1
7.1
31.3
18.2
open
22.3
15.4
9.4
16.1
34.5
30.0
29.4
8.8
22.2
26.3
12.5
14.7
Dso=O.5O mm smooth
rough
18.2
28.6
12.1
23.5
32.3
18.2
open
38.1
20.2
34.2
20.0
22.4
33.3
54.8
9.7
32.0
36.4
Dso=O.47 mm smooth
41.9
14.9
rough
20.0
38.9
33.3
28.8
42.4
30.8
*average and standard deviation calculations based upon open, smooth and rough
cover inclusive of pier size and sediment grain size.
open
open
22.85
8.31
standard
average
smooth
smooth
22.46
13.08
oevianon
rough
rough
25.57
9.37

93

5. 7 Armour layer analysis

In the flume literature, very few studies use natural or non-uniform sediment due to its
complexity. However, since river beds are composed of a mixture of different sediment sizes

it is more representative of natural systems to use non-uniform sediment in laboratory flume
experiments. In an experimental flume environment, an armour layer is formed when nonuniform sediment is exposed to a constant discharge and no sediment is input to the system.

Initially, bed degradation occurs, the bed decreases in slope and the sediment transport rate is
at its maximum, gradually decreasing as time passes (Aberle and Nikora, 2006; Mao et al.,

2011). Finer bed materials are transported faster than coarse materials under the same flow

conditions; the coarse material typically rolls and jumps along the river bed until such point

that hydraulic conditions cause it to become stationary. This process is repeated and the river

bed sediments effectively become coarser acting as a protective layer for underlying finer
material (Proffitt, 1980). The coarser sediments form a static armour layer when the flow

entrains only the finer elements of non-uniform bed material (Church et al., 1998). The
armouring process depends on the flow velocity and the grain size distribution of the bed

mixture (Church et al., 1998).

The armour layer that developed within the scour hole was measured for each
experimental run and the average armour layer D50 is provided in Appendix C. Figure 51,
Figure 52 and Figure 53 present a subset of the experimental data and are examples of the

armour layer grain size distribution for three individual experiments for each of the three D50S.
The armour layer generated in D5o=O.58 mm is coarser than the armour layer for Dso=O 47 mm
and 0.50 mm (Figure 54). As the D50 decreases, more fine sediments make up the bed material
and the sediment size in the armour layer decreases. Smaller particles in the armour layer will

94

provide less scour protection around the pier and result in a larger local scour depth. Figure

55 presents armour layer size and scour depth under the 1 1 and 22 cm piers. As indicated in
Figure 55 armour layer size does not discriminate against pier size. Both the 1 1 cm and 22 cm

piers have similar ranges in armour layer size. This is an important note as while pier size

does affect scour depth (Figure 32, Figure 33, and Figure 34), the size of the armour layer that
forms around a pier is dependent on the median size of bed material rather than the change in

flow vortices with pier size. When measuring median size of the armour layer it should be
noted that the non-uniform nature of the three experimental sediments presents large variation

in geometric standard deviation (Appendix C). The armour layer D50 presented in this section
is subject to both systematic and random error. Systematic uncertainty arises in the methods

of armour layer sampling, namely the top 5 mm of the armour layer was sampled based upon
visual observation. Slight differences would arise from each sample depth and sampling area
within the pier scour hole. The upstream, side and downstream areas of the scour hole were
sampled, however, slight differences in each sample taken would produce systematic error.
Random error arises due to the variation in the transport of non-uniform sediment. Depending

on the individual grain embedded depth, turbulence and shear stress, each experimental flume
run would result in different fluid forces for sediment transport and subsequent armour layer
development. The random error is represented by the geometric standard deviation values for

each armour layer sample as indicated in Appendix C.
Under all channel covers, the scour depth decreases as the armour layer size increases.
Similar results were also found by Raudkivi and Ettema (1985) and Dey and Raikar (2007).

Both studies found that bed armouring acts to reduce pier scour development compared to the
expected scour development if no bed armouring layer was present. Chiew and Melville

95

(1989) also investigated scour in non-uniform sediments. The authors found that local scour

in non-uniform sediments was less than that of uniform sediments due to the formation of an
armour layer. Raudkivi and Ettema (1983) also found that as the standard deviation of
sediment increases, the rate of scouring and final equilibrium depth decreases. They found

that the decrease in equilibrium scour depth was due to the formation of an armour layer, which

was a significant process for standard deviation values greater than 1 .4.
Scour mitigation at bridge sites often involves the placement of riprap around the pier

base as a scour countermeasure. Riprap can act as a resistant armour layer to the hydraulic
shear stress and provide protection to the smaller erodible sediments underneath (Deng and

Cai, 2010). Riprap is the most common and cost effective scour countermeasure used today.
The main drawback in using riprap as a scour countermeasure is that once in place, it is difficult
to monitor the riprap placement and it may move allowing finer material to erode. Studies by

Lagasse et al. (2007) and Lauchland and Melville (2001 ) both recommend rip-rap be placed at
a depth below the average bed level in order to act as an effective countermeasure.
The findings in this study support previous research indicating that as armour layer size

decreases pier scour depth increases. The presence of an ice cover did not however influence

armour layer size for Dso=O.5O and 0.47 mm (Table 15). The average armour layer size for

Dso=O.47 mm remained constant while average armour layer size for D5o=O.5O mm varied by
0.12 mm. Of note however, under Dso=O.58 mm the average armour layer size did increase
with channel cover roughness by up to 0.67 mm (Table 15). Since all sediments used were

non-uniform, it is unclear as to why the average armour layer size varied under Dso=O.58 mm
but not under the other two sediments used. Since the critical bed shear velocity would be

96

slightly different for each run due to the initial sediment arrangement, the process of armour
layer development would be different for each experimental run.

Figure 51. Example of armour layer and related distribution curve for Dw = 0.58 mm.

D50 = 0.50 mm

Figure 52. Example of armour layer and related distribution curve for D50 = 0.50 mm.

97

Dm = 0.47 mm

Figure 53. Example of armour layer and related distribution curve for Djo = 0.47 mm.

channel bed Djolmm)

Figure 54. Variation of scour hole armour layer size with median bed grain size. The average geometric
standard deviation for D$o = 0.47, 0.50, 0.58 mm is 2.57, 3.20 and 2.96 mm respectively.

98

300.00

22 cm pier
X 11 cm pier

250.00
„

200.00

x>

o’ 150.00

"xn
E

100.00

50.00
0.00

0

armour/Djo bed
Figure 55. Variation of maximum scour depth and average scour hole armour layer size under 1 1 and 22 cm
pier width.

Table 15. Average size of scour hole armour layer for each bed material and channel cover.

median sediment
size channel bed
(mm)

Dso = 0.58
Dso = 0.50
D50 = 0.47

channel cover
open
rough
smooth
median sediment size in scour hole armour layer (mm)
0.75
1.42
1.09
0.87
0.76
0.88
0.50
0.50
0.50

5.8 Scour hole velocity profiles
One of the key features of this thesis is that the scour hole velocity profiles were
measured under ice cover. Previous work by Unger and Hager (2007), Graf and Istiarto (2002)
and Kumar and Kothyari (2012) have investigated scour hole velocity profiles for open channel

flow, but to my knowledge this is the first study to investigate scour hole velocity flow fields
under ice cover.

99

Scour hole velocity profiles were measured for depths varying from 22-24 cm under
the two tailgate flume configuration for both the 11 and 22 cm piers. Velocity measurements

were taken at the end of the experimental run at which time the scour hole was fully developed.

Scour hole velocity profiles for 10 cm depths (1 tailgate) were unavailable due to limitations

of the ADV. Of note, while it is expected that scour hole velocity will vary slightly under
various Dsos due to the variation in the size of the armour layer, the purpose of this section is
to highlight the overall patterns and distribution of the velocity profile between open, smooth

and rough channel covers. Therefore, the focus of this section is examining the differences in

scour hole velocity profiles between open, smooth and rough channel conditions. As the ADV
measures velocity in three dimensions, the velocity profiles presented in this section comprise
the streamwise (Ux), lateral (Uy) and vertical (Uz) velocity components. The following is a
summary of the profile patterns for the individual velocity components under each channel

cover.

Streamwise velocity, Ux:
Figure 56 and Figure 57 represent scour hole velocity profiles for individual runs under

the 22 and 11 cm piers respectively for various channel covers. For the streamwise (x) velocity
component the scour hole profiles take on a logarithmic pattern. For both the 1 1 and 22 cm

piers, the streamwise velocity for rough cover is generally higher than the scour hole velocity
for smooth and open channel conditions. For the 22 cm pier, the streamwise velocity for rough

cover is between 15-47 and percent higher than the scour hole velocity for open channel
conditions (Table 16). This is most clearly represented for the 22 cm pier configuration under

Dso=O.58 mm and 0.47 (Figure 56). For both pier sizes and channel conditions the streamwise
100

velocity is smallest in the scour hole and increases towards the water surface. Within the scour
hole, streamwise velocity decreases with scour depth, eventually reaching negative values.

This is indicative of flow reversal that occurs due to the horseshoe vortex that is located at the
pier face in the scour hole. Since the horseshoe vortex and downflow increase with pier size
(Shen et al., 1969) the patterns in the scour hole flow field are most obvious under the 22 cm

pier configuration (Figure 56).
Table 16. Average velocity measured for each profile presented in Figure 56 and Figure 57. In addition, the
percent the velocity values are greater or less than open channel velocity are presented. Data are missing due to
ADV error for Dw=0.47 mm under smooth cover conditions.

channel cover and velocity component (cm/s)
open smooth rough open smooth rough open smooth rough
Uy
Uy
Ux
Ux
Ux
Ux
Ux
Ux
UY
channel bed
1 1 cm pier
8.5
mm
0.58
17.8
14.8
4.7
-2.6
-5.9 -1.7
-4.9
-4.9
D50 =
% +/- open Ux
+43
+52
-36
-65
-56
-65
0.50
10.6
11.7
mm
11.4
4.7
-7.6
-3.4
-1.2
-6.7
-6.2
D50 =
% +/- open Ux
+9.5
+7.1
+7.3
+35
+58
-12
13.2 -1.1
-4.6 -5.8
-4.3
D50 = 0.47 mm 9.4
% +/- open Ux
+29
-76
+26
22 cm pier
7.9
9.8
-1.3
-3.6
0.2
1.4
-3.1
-0.7
D50 = 0.58 mm 6.6
% +/- open Ux
+16
+ 13
+ 14
+33
+51
+80
8.9
9.1
-3.2
-0.4
-2.7
3.4
-1.9
-1.7
D50 = 0.50 mm 4.9
% +/- open Ux
+47
+45
+89
+47
-85
+61
7.4
9.2
-3.5
-2.9
-0.7 -4.3
D50 = 0.47 mm 6.3
-4.2
-4.2
%+/- openUx
+15
+ 18
+32
+2.4
+80
+2.4
median

Lateral velocity, Uy:

For the lateral (y) velocity component, there does not appear to be any pattern in flow

for both the 11 cm and 22 cm pier. The lateral velocity profile appears turbulent and changes
between positive and negative values along the entire profile.

101

Vertical velocity, U2:

For the vertical velocity component (z) the scour hole profiles exhibit patterns in the
shape of the letter C. Figure 56 represents the C-shaped profile to a greater extent than the

profiles for the 11 cm pier (Figure 57). The downward velocity is generally greatest under
rough cover conditions and smallest under open channel conditions (Table 16). In examining

the C-shaped profile, the smallest velocity typically occurs at the transition zone between the

channel bed and scour hole. From the channel bed to transition zone the velocity decreases,
while from the transition zone to the water surface the velocity increases.

Discussion:
For open channel flow, the logarithmic and C-shaped scour hole velocity distributions

for streamwise and vertical velocity are similar to those found by Kumar and Kothyari (2012).
Under open channel conditions the horseshoe vortex and downflow increase with pier size
(Shen at al., 1969) therefore patterns in the scour hole flow field are most obvious under the

22 m pier configuration. Under the experimental conditions for this study, it was found that
the streamwise and vertical (Ux, Uz) velocity profiles for smooth and rough ice cover have a

similar distribution and pattern to open channel flow, however, ice cover leads to higher
velocity values. The streamwise and downward velocities at the pier face are greater under ice

cover than open channel flow. For all sediment Dsos under the 22 cm pier, the average rough

cover velocity is 37 percent greater and the average smooth cover velocity is 1 5 percent greater
than the open channel velocity (calculated from Table 16). For all sediment Dsos under the 11
cm pier, the average rough cover velocity is 29 percent greater and the average smooth cover
velocity is 26 percent greater than the open channel velocity. As ice cover shifts the location

102

of maximum velocity towards the channel bed (Sui et al., 2010), while also increasing the
maximum velocity, the overall bulk velocity value is generally higher under ice cover than
open channel flows.
A greater flow down the face of the pier to the riverbed can lead to greater scour around

the pier base. Under open channel conditions, approaching flow velocity meets the pier face
and flows downward to the channel bed or upward to form a bow wave. Under ice cover

conditions, the extent of the upward flow and bow wave is diminished due to the ice cover

boundary. This causes increased downflow at the pier face resulting in a greater capacity for

sediment transport. These conditions described above of increased downflow under ice cover

are consistently present under the 22 cm pier for the vertical (Uz) velocity component. For all
sediment D50S under the 22 cm pier, the average rough cover vertical velocity is 43 percent
greater and the average smooth cover vertical velocity is 26 percent greater than the open

channel velocity (calculated from Table 16). The larger scour hole velocity under ice cover
supports the findings of greater pier scour depth under ice cover given a consistent armour

layer size in comparison to open channel conditions.

103

flow direction

Figure 56. Scour hole velocity profiles for the streamwise (Ux), lateral (Uy) and vertical (Uz) velocity components under open, smooth and rough ice cover for
22 cm pier. All profiles were measured under the two tailgate flume position with water depths from 22-24 cm. ADV measurements are accurate to +/- 0.25
cm/s.

104

flow direction

Figure 57. Scour hole velocity profiles for the streamwise (Ux), lateral (Uy) and vertical (Uz) velocity components for 1 1 cm pier. All profiles were measured
under the two tailgate flume position with water depths from 22-24 cm. Smooth cover velocity profiles for D5<>=0.47 mm are missing due to ADV file error.
ADV measurements are accurate to +/- 0.25 cm/s.

105

K

5. 9 Dimensional Analysis
Dimensional analysis is the relationship between various physical quantities by

identifying their dimensions. The dimension of any physical quantity is simply the actual
physical dimensions that create it. In order for engineers to understand pier scour the
parameters and associated dimensions responsible for local scour must be defined. The

following are parameters that are associated with local pier scour as suggested by Ozalp (2013)
and Melville and Coleman (2000). The parameters are classified in terms of time (T), length
(L) and mass (M).
Table 17. Parameters associated with local pier scour.

Parameters describing flow
gravitational acceleration
g
approach flow depth
H
approach flow velocity
U
shear velocity
density of water
pw
V
kinematic viscosity of water
Parameters describing the flume
channel slope
s
channel width
B
n,
Manning's coefficient for ice cover roughness
Parameters describing channel bed material
sediment density
Ps
median sediment size
D50
standard deviation of sediment size
°g
cohesion
C
Manning’s coefficient for channel bed
Ob
median sediment size of the armour layer
DsOa
Parameters describing bridge pier
pier diameter
b
pier shape factor
Ki
pier surface roughness
Kr
angle of attack
k2

u.

Time
t

| flow duration

Units

Dimensions

m/s2

LT2

m

L

m/s
m/s

kg/m3
m2/s

LT1
LT1
ML3
L2T-1

-

-

m

L

-

-

kg/m3
mm
-

ML3

kg/m s2

L

-

ML'T2

-

-

mm

L

m

L

-

min

-

T

106

The above parameters associated with local pier scour can be presented by the following
relationship, where dm„ represents the maximum scour depth:

dmax =

U, U, pw. v, s, B, nt,

Dso. ag. C, nb, Dsoa. b, Ki, Kr, K^t) =0

5.7

Under dimensional analysis the following parameters from equation 5.7 are considered:
^max _ f ( D

'

d

*

££so Dso
v

P$p Djoo Ps~Pw Ut n^
H ’ U ' B ' Dso ' pw ' B ’

'

nb’

C

B‘ pU*

b

b

B' H '

„

11

„

r'

„

2

\
/

In considering the experimental conditions of this study equation 5.8 can be simplified by the
following conditions:
(a) the flume has a constant slope and channel width so the 5 and B terms are ignored

(b) cylindrical piers are used which are smooth, and the angle of attack for all experiments

is zero, so the Ki, Kr and K2 terms are ignored
(c) the bed materials remain constant for each experimental set, so the og term is ignored

and the

—

term is considered instead.

(d) the densities of the flume water and sediment are constant for each experimental subset

—

so the Ps^Pw is ignored.
1

•

f

Pw

(e) the bed material is non-uniform and considered non-cohesive (C = 0) so the

ignored.

—

j is

(f) after 24 hrs of experimental run time, an armour layer is formed and equilibrium scour

depth is reached; therefore scour time,
(g) for each experimental subset, the

—

, is ignored.

is defined in relation to

as it depends on approach

flow depth.

107

Given the above conditions, equation 5.8 is simplified and the following parameters are related
to scour depth under the experimental conditions of this study:

^max
d

Where

_ f ( U UPso £so £so £soa b b
' \JgH' v ' H ' B ’ Dso ’nb'B' H

is the Froude number,

n(

V

5.9

is the particle Reynolds number,

— —

relationship between median sediment grain size, flow depth and pier width,
the impact of sediment composition on armour layer particle size,
b

roughness and the channel bed roughness and - and
B

—
b

H

and

HQ

Dsoa

is the

represents

represents the ice cover

is the pier diameter in relation to

channel width and water depth. Since three types of sand are used for flume experiments, the
densimetric Froude number can also be used in addition to the Froude number, written as:

Fo = ,

5.10

Also of interest in this research is the specific particle Reynolds number, which is indicated by
the shear velocity U., so the shear Reynolds number is substituted for the Reynolds number,

written as:

Re. =

V

5.11

In the following sections, the parameters in equations 5.9 will be compared to maximum scour
depth through correlation and regression analysis.

108

5.10 Correlation of scour depth versus the Froude number
The ability of an object to move through water will depend on its size (object length
and area) as well as the relative velocity and depth of water. The Froude number, U/jgH,
represents the ratio of fluid inertial forces to fluid gravitational forces. The greater the Froude

number, the greater the resistance exerted by water flow on the river bed material. It is standard

engineering practice to relate pier scour to flow depth, flow velocity, pier size and the Froude
number (Molinas, 2003). One of the first studies to document the positive relationship between
Froude number and scour depth was Aziz ( 1 983). The author found pier scour was dependant

on both flow depth and velocity and pier scour increased with increasing Froude number. Wu
et al. (2014) also documented that the Froude number increased with abutment scour depth

under ice cover and open channel flow. Of note however, it is important to consider that the

Froude number used in laboratory experiments may be larger than that for similar field
conditions as pointed out by Melville and Coleman (2000). Since the flume sediment relative
to pier size is larger than the prototype, the flow velocity required for incipient motion may be

larger than the velocity derived from Froude scaling of prototype flow (Melville and Coleman,
2000). As such, the Froude number used in flume experiments may be larger than that for

field conditions, leading to overestimation of pier scour. While scaling issues related to the
Froude number are discussed here, further scaling issues are addressed in Chapter 6. As the
Froude number incorporates flow depth and velocity, and scour depth is dependent on flow

depth and flow velocity, the relationship between scour depth and the Froude number is
examined in this section.

109

Since the Froude number is a dimensionless variable, it was plotted against the

dimensionless maximum scour depth (dmax/H) in Figure 58. For all experiments, there is a
positive relationship between the Froude number and maximum scour depth (Figure 58). To
further investigate variation in Froude number with scour depth, rough ice cover and open

channel conditions are compared in Figure 59; however, only open channel and rough cover
were compared for all three Dsos since including smooth cover created a visually busy graph.

The relationship between Froude number and maximum scour depth under smooth ice cover
for all three Dsos is presented in Figure 60.
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
0.

Figure 58. Variation of maximum scour depth with Froude number for all experimental runs.

110

0.10

0.15

0.20

0.25

Fr
0.70
0.60
0.50

x 0.40

J 0.30
0.20
0.10
0 00
0.

Figure 59. Variation of maximum scour depth with the Froude number under open and rough cover for all
three sediment sizes.

Ill

1.40

smooth cover

Figure 60. Variation of maximum scour depth with Froude number for smooth channel cover under various
sediment sizes.

Generally, under the same Froude number, maximum scour depth is greater under
rough ice cover conditions (Figure 59). This is clear for D5o=O.58 and 0.50 mm; however,

under D5o=O-47 mm maximum scour depth is not greater under ice cover conditions for
Froude>0.22 (Figure 59). There is also a large range in measured scour depth exhibited for

Dso=O.5O mm, rough cover at Fr~0.20 and for D5o=O.47 mm, rough cover, around Fr~0.19.
Under smooth ice cover, maximum scour depth also increases with Froude number. For

smooth ice cover conditions, under the same Froude number, the largest scour depth occurs
under Dso=O.47 mm, while the smallest scour depth occurs under Dso=O.58 mm (Figure 60).
The large range in measured scour depth exhibited for Dso=O.5O mm, rough cover at

Fr~0.20 and for Dso=O.47 mm, rough cover, around Fr~0.19 is most likely due to variation in

the initial non-uniform sediment arrangement and subsequent armor layer development. In

112

general, under uniform sediment, an ice cover will increase resistance and turbulence and

therefore potential for sediment transport - which supports the trend of larger scour depths
under rough ice cover then open channel under the same Froude number. However, for the

same Froude number, non-uniform sediment transport processes will vary according to initial
sediment arrangement and interactions between sediment grains.

In order to incorporate sediment size in relation to the Froude number, the densimetric
Froude number, Fo, can be used as the mass density of sediment is considered (Equation 5.10).
Hodi (2009) conducted a series of flume experiments using uniform sediment, Dso=O.85 mm,
investigating pier scour under Fo = 2.1 and 2.5 under 10, 20 and 30 mm piers. The author

found that when the pier size is held constant and Fo increases from 2. 1 to 2.5 the area of scour
increases significantly. For this study, in order to investigate if the Fo has a relationship with

scour depth, the dimensionless scour depth is plotted against the densimetric Froude number
in Figure 61. As indicated by Figure 61, relating the scour depth to the densimetric Froude
number provides no guidance when applied to non-uniform sediment. The scatter is large and

no significant relationship is present. When the same velocity is applied to both a uniform and
non-uniform sediment of the same D50, each sediment will have different critical velocities
due to the arrangement and interactions of individual bed grains.
Molinas and Wu (1998) point out that a single sediment size such as the D50 does not

reflect the range in non-uniform sediment sizes nor the impact various size fractions have on
transport calculations. The authors evaluated commonly used sediment transport formulas for

bed-material load of non-uniform sediment and found that the transport formulas gave
considerable scatter for bed-material load relationships. As a result, Molinas and Wu (1998)
developed a size gradation compensation factor, which is a function of the geometric standard

113

deviation and flow conditions. They included this in the original transport formulas and found

that agreement between the measured and computed bed-material concentrations improved.
As indicated by Molinas and Wu (1998), the D50 value in the densimetric Froude

number does not adequately represent the sediment composition and therefore the Fo does not
present a strong relationship with pier scour depth for non-uniform sediment. This study does

however indicate that there is a relationship between scour depth and the Froude number. For
D$o= 0.58 and 0.50 mm, under the same Froude number, rough ice cover produces a greater

scour depth then open channel conditions. Even though flume scaling under Froude similarity
may lead to overestimation of pier scour, it is still important in bridge design to note that ice

cover may produce a larger scour depth in comparison to open channel conditions under the
same Froude value.

Fo
Figure 61. Variation of dimensionless maximum scour depth with densimetric Froude number for all
experiments.

114

-

5.11 Correlation of scour depth versus shear stress incipient motion
A sediment grain will move when the shear stress acting on it is greater than the
resistance of the particle to movement. The magnitude of shear stress required to move a

particle is known as the critical shear stress (Tw). The resistance of the particle to movement
and entrainment will vary depending on the particle size, its orientation, particle size relative

to surrounding particles and embedment. The relationship between the dimensionless shear
stress and the Reynolds number is known as the Shields Diagram which is often used for

predicting incipient motion of sediment.

The critical Shields value, also called the

dimensionless shear stress, T„ used to calculate the initiation of motion of sediment is given
by the following equation:

T* =

pU.c2
A

512

where p is the fluid density, bp is the difference between the fluid density and sediment
density, U.c is the critical shear velocity, g is gravity, and D50 is the median sediment size of

bed material. The logarithmic velocity distribution assumption is one of the generally accepted

methods for calculating the shear velocity based on the Prandtl and Einstein correction factor
(Einstein, 1950).

*c“

5.751og10

I
fI12.27;^
Ao J

5.13

where Rh is the channel hydraulic radius, u is the average cross sectional velocity, D50 is the
median bed grain size and % is the Einstein multiplication factor, / =1. For ice covered flow,

115

the ice cover can be included under the channel hydraulic radius. In practice, the shear

Reynolds number is usually used to study sediment incipient motion, which is given by:

Ree = —v—

5.14

in which, U,c is the critical shear velocity, D is the grain size diameter and v is the kinetic
viscosity of the fluid. Since the sediment used here is the non-uniform sediment, the grain size

diameter will be replaced by D50, then the above equation can be written as following,

^«c^50
r>*
RI
e =
v

5.15

The calculated dimensionless shear stress versus Reynolds number is presented in

Figure 62. Under all sediment sizes, as the dimensionless shear stress increases so does the

shear Reynolds number. For the same shear Reynolds number, the finest sediment D5o=O.47
has the highest dimensionless shear stress. The greater dimensionless shear stress, the greater
the capacity for sediment transport. With a higher proportion of finer particles in non-uniform
sediment, high dimensionless shear stress values will lead to greater sediment transport. For

the same dimensionless shear stress, the coarser the bed material, the larger the shear Reynolds
number. A larger Reynolds number for coarser sediment particles indicates a larger shear
velocity is required for incipient motion of sediment.
Figure 63 presents the maximum scour depth as it relates to dimensionless shear stress

for both the 11 and 22 cm pier. As the maximum scour depth increases the dimensionless
shear stress increases correspondingly. Figure 63 also indicates that, for dimensionless shear
stress>0.030, the 22 cm pier will produce a greater maximum scour depth. This supports

findings by Breusers et al. (1977) that the dimension of the horseshoe vortex is a function of

116

pier diameter. With a larger horseshoe vortex, the sediment transport capacity is greater hence
the larger maximum scour depth for the larger 22 cm pier.

Figure 64 and Figure 65 present the maximum scour depth in relation to dimensionless
shear stress for open channel and rough ice cover conditions.

For both open channel and ice

covered conditions the maximum scour depth and dimensionless shear stress increase
correspondingly. This positive trend is expected as increasing dimensionless shear stress
allows for greater sediment transport and larger scour hole development. For both the 1 1 and

22 cm piers, under the same dimensionless shear stress open channel conditions produce a
larger scour hole. In other words, under rough ice cover conditions less dimensionless shear
stress is required for sediment transport and scour hole development.

Figure 62. The variation of shear Reynolds number with dimensionless shear stress

117

300

•11 cm pier
o 22 cm pier

0.010

0.015

0.020

0.025

0.030

0 035

0.040

0.045

0.050

0.055

dimensionless shear stress T.
Figure 63. Variation of maximum scour depth with dimensionless shear stress around 1 1 and 22 cm pier.

dimensionless shear stress T.
Figure 64. Variation of maximum scour depth with dimensionless shear stress under rough ice cover and open
channel conditions for 1 1 cm pier.

118

22 cm pier

300
250
200

150
100

50
0
0.010

0.015

0.020

0025

0.030

0035

0.040

0.045

0.050

0.055

dimensionless shear stress T.
Figure 65. Variation of maximum scour depth with dimensionless shear stress under rough ice cover and open
channel conditions for 22 cm pier.

5.12 Multiple Regression Analysis
A basic linear regression model assumes that the contributions of various independent

variables to the prediction of the dependant variable are additive. Given that two independent

variables (Xi and X2) may contribute additively to the dependant variable (Y), the prediction
equation would be as follows,
F(t) = b0 +

M2(0

5.16

where if all things are equal and Xi increases by one unit, then Y is expected to increase by bi
units. When there is an absolute change in Y that is proportional to the absolute change in X/,
then the coefficient bi represents the constant of proportionality. For this study the dependant
variable is the maximum scour depth and the independent variables can be written as follows,

b b
u yP5P £so °so Psoa
f (
—
’
’
'
H
B ’ Dso nb 'b ’H
v
H

dmax

nj

5.17

119

For this study, the maximum scour depth is affected multiplicatively by the independent
variables, meaning that the percentage change in Y should be proportional to the percentage

change in Xi and Xj and so on (unlike linear regression where the change is additive). Since

linear regression is not appropriate for this situation, a multiplicative regression model should

be applied,
F(0= b0(Xi(t)b')(X2(t)b’)

5.18

where Y is proportional to the product of Xi and X: which are each raised to some power based
upon the data set. For analysis purposes and equation development, a multiplicative model

can be converted into an equivalent linear model by taking the natural logarithm as follows,

LN (r(t)] = LNlb^X^) (X2(t)^)]

5.19

where the coefficients bi and bi can be positive or negative and will provide the direction and
magnitude of the effect of the associated variable, Xi or X? on Y.

In order to gain a better understanding of the impact of various dimensionless hydraulic
parameters on maximum scour depth, a multiplicative regression model was applied and the

natural logarithm was used in order to present linear graphical analysis. Using multiplicative
regression provides a quantitative description into the dependence of maximum scour depth

on the Froude number, sediment grain size, pier size, ice cover and armour layer size. Various
parameter combinations from equation 5.17 were explored through regression analysis since

for example sediment Dso is presented by both D50/H and D50/B.

Not all parameter

120

combinations produced strong regression models and therefore are not included in the sections
following but rather in Appendix D. Reasons that parameter combinations did not produce
strong regression models are as follows:

(a) the regression model produced regression coefficients that did not support findings

from previous sections in this study. For example, the regression coefficient for
sediment D50 would be positive, when in fact findings under section 5, indicate that

sediment D50 actually decreases with increasing scour depth.
(b) the regression model produced supported findings of this study however a stronger

correlation was found using various other parameters. For example, it was found that
the parameters D50/H and D50/B both typically produced valid regression models,
however, for the most part, D50/B would produce a stronger correlation.

The following sections present the regression models for maximum scour depth as it relates to

open channel conditions, ice covered conditions and armour layer development.

Scour depth under open channel conditions:
Using principals of dimensional analysis described in section 5.9, the maximum scour

depth around the 11 and 22 cm piers under open channel flow can be described by the
following variables:

5.20

121

where U I JgH is the Froude number and D3Q/B is the change in sediment size with pier width.
By using regression analysis the following equations were derived from open channel
experiments (Figure 66 and Figure 67).

For 11 cm pier:
h

= 5E -11 fe)3
XJgH/
X bJ

°019

5 21

f^2)’4'93 - 0.017

5.22

“

For 22 cm pier:
°3

= 4E"16

For both the 1 1 and 22 cm piers, the Dx/B regression coefficients indicate that as sediment

size decreases the maximum scour depth increases. This trend supports findings in section 5. 1
(scour depth versus sediment size) of this thesis. Finer channel particles require less shear

stress for incipient motion and can produce a larger maximum scour depth. Both Froude

regression coefficients for equations 5.2 1 and 5.22 indicate that as the Froude number increases

the maximum scour depth also increases. This is also confirmed in section 5.9 (correlation of

flow depth with Froude number). The greater the Froude number, the greater the resistance
exerted by water flow on the river bed material; therefore, there is a positive relationship
between Froude number and scour depth (Aziz, 1983). Similarly, Wu et al. (2014) also
documented that the Froude number increased with abutment scour depth under open channel
flow. The R2 value for the 22 cm pier, Figure 67, is also greater than the R2 value for the 1 1
cm pier, Figure 66, indicating a stronger correlation.
Using dimensional analysis, the pier size can also be incorporated into the regression
model given by the following equation,

122

where b/B is the change in pier size with channel width. Through regression analysis, the
following equation was derived for open channel conditions, representing the change in
maximum scour depth with flow Froude number, sediment D50 and pier size (Figure 68).

= 2E”3

ter (^r er -

For equation 5.24, the Froude regression coefficient of 3.77 indicates that maximum scour
depth increases with Froude number, the D50 coefficient of -4. 1 5 indicates that as sediment D50
decreases scour depth increases, and the pier size coefficient of 0.03 indicates that as pier size

becomes larger, scour depth also increases. Here, when comparing the two positive regression
coefficients, the Froude coefficient (3.77) is larger than the pier size coefficient (0.026); this

indicates that the value of the Froude number has a greater influence on scour depth than the
pier size. In addition, the sediment D50 has the largest regression coefficient (4.15) indicating
that the sediment size has the largest influence on scour depth.

123

11 cm pier, open channel
1.00

Figure 66. Variation of scour depth for 1 1 cm pier under open channel conditions.

22 cm pier, open channel

Figure 67. Variation of scour depth for 22 cm pier under open channel conditions.

124

1.20
o

Figure 68. Variation of scour depth in relation to Froude number, median bed sediment size and
pier width under open channel conditions.

Scour depth under ice conditions:
Using dimensional analysis, the maximum scour depth around the 11 and 22 cm piers

under ice cover can be described by the following variables:
dmax _
H

a

( U

\ B J

f ni A

\nbJ

C

5.25

where ni/nb is the ice cover roughness compared to the channel bed roughness. In order to
calculate the roughness coefficients for the channel bed and ice cover, equations from the

literature were examined. Li (2012) analyzed large sets of field data and found that Manning’s
coefficient for ice cover for ice covered rivers averages from 0.013 to 0.040 during the winter.
The following equation was used to calculate the Manning’s coefficient for rough ice cover
(Li, 2012):

125

n(

= 0.039 ks1/6

5.26

where ks is the average roughness height of the ice underside in meters. Using equation 5.26
the Manning's coefficient for rough ice cover was calculated as 0.021. This value also

coincides with findings by Carey (1966) and Hains and Zabilansky (2004). The height of
roughness of the ice underside was based upon the 2.5 cm Styrofoam cubes that were attached
to the original 1 .2 m x 2.4 m (4 x 8 foot) Styrofoam sheet. For smooth ice cover, the Manning's

coefficient was based upon the Manning's value for smooth concrete, and the value of 0.013

was used (Mays, 1999). To calculate channel bed roughness for non-uniform sediment the
following equation was used (Hager, 1999):
nb = 0.039

DS01/6

5.27

Based upon equation 5.27, the channel bed roughness for the various D50S in this study were

calculated as follows: D5o=O.58 mm nb = 0.01 13, Dso=O.5O mm nb = 0.01 10, Dso=O.47 mm
nb = 0.0109. By using regression analysis the following equations were derived for ice

covered experiments (Figure 69 and Figure 70).
For 11 cm pier ice cover conditions:

5.28
For 22 cm pier ice cover conditions:

£max _ 2£-14

5.29

126

Just as for open channel conditions, the D50/H regression coefficients for ice cover indicate
that as sediment size decreases the maximum scour depth increases. For both equations 5.28
and 5.29, the Froude regression coefficients indicate that as the Froude number increases the

maximum scour depth also increases. The nJ nb regression coefficients however do not
indicate similar trends for both the 1 1 and 22 cm piers. The ice cover roughness coefficient

for the 22 cm pier indicates that as the ice cover roughness increases the scour depth will also
increase (Figure 70). Opposite is indicated by the roughness coefficient for the 11 cm pier
(Figure 69). This discrepancy reflects the findings outlined in section 5.3 ‘scour depth versus

channel cover’ where rough ice cover resulted in greater scour depth then open channel
conditions for 60 percent of experiments and smooth ice cover produced greater scour depth

then open channel conditions for 53 percent of experiments. Various other parameters were
tested for this regression model as indicated in Appendix D, however for the njnb parameter

an appropriate model could not be found.
Using dimensional analysis, pier size was also incorporated into the regression model
given by the following,

By using regression analysis the following equation was derived for ice covered experiments

taking into consideration pier size (Figure 71).

127

In equation 5.31, the regression coefficient for the Froude number and pier size are both
positive, indicating that both parameters have a positive relationship with maximum scour
depth. The Froude coefficient, 2.87, is larger than the pier size coefficient, 0.03, indicating

that the Froude value has a greater impact on scour depth than pier width under ice cover

conditions. Additionally, under the regression model in equation 5.3 1 , the coefficient for the
ice roughness parameter is positive, indicating that as ice cover roughness increases so does

scour depth. As over 50 percent of experimental runs measured a greater scour depth under
ice cover, theoretically, the coefficient should be positive; however, as previously mentioned,

smooth and rough ice cover only produced a greater scour depth than open channel flow for

57 and 60 percent of experiments, therefore the coefficient value is small. The ice cover
condition does not always produce a greater scour depth due to variations in non-uniform

sediment transport and subsequent armour layer development, as discussed in section 5.3,
‘scour depth versus channel cover’. To further explore the effects of the armour layer on scour

depth, the next section includes the armour layer in regression analysis.

128

11 cm pier, smooth + rough ice cover

Figure 69. Variation of scour depth for 1 1 cm pier under ice covered conditions.

Figure 70. Variation of scour depth for 22 cm pier under ice covered conditions.

129

smooth + rough ice cover

Figure 71 . Variation of scour depth in relation to Froude number, median bed sediment size, ice cover
roughness and pier width.

Effect of armour layer on maximum scour depth:
Using principals of dimensional analysis and taking into consideration the armour
layer, the maximum scour depth under open channel conditions and ice cover can be described

by the following variables:

For open channel conditions:
^max
H

_

( U

X (^\b f-V (Dsoarmour\d

\fgHj \ B J \b) \Dsochannel)

5.32

For ice cover conditions:
dmax
H

_ A ( U V (^f\b

fDsoarmour\d (b\e

\fgHj \ H J \nb) \Dsochannel) \h)

5.33

130

where Djo armour /D30channel is the armour layer size relative to the channel D50. By using
regression analysis the following equations were derived from open channel and ice covered

experiments (Figure 72, Figure 73).

For open channel conditions:
79

H

_

(£so)-408 \B/
P)° \D$ochannel/
= 3E-13 m3
\7pH
)

01

\ B /

q qq9$

5 34

For ice cover conditions:
54

H

= 20.58

(^^V0A7 pf + 0.0 2 54
m3 f^)’033(^)°-2S
J
\ H

\nt,/

\D$ochannel/

\H/

5.35

In this case, the interest is in comparing the regression coefficients for armour layer D50 as they
relate to maximum scour depth. For both open channel and ice covered conditions the armour
layer regression coefficients are negative indicating that as armour layer D50 size decreases,

scour depth increases. This correlation supports results of this study outlined in section 5.7
‘armour layer analysis’. Bed armouring acts to reduce pier scour development compared to

the expected scour development if no bed armouring layer was present. For non-uniform
sediments, local scour is typically less than that of uniform sediments due to the formation of

an armour layer (Chiew and Melville, 1989). The negative regression coefficient for the

armour layer parameter also aligns with the negative regression coefficients for sediment D50
for open channel and ice cover conditions (D50/B and D50/H for equations 5.34 and 5.35
respectively). As previously discussed, this study found that scour depth increases as sediment

D50 decreases; the D50 coefficients in equations 5.34 and 5.35 also support this trend.
Considering all the regression coefficients in equations 5.34 and 5.34, the Froude number,

sediment D50 and armour layer D50 appear to be the most important parameters affecting

131

maximum scour depth, while ice cover roughness and pier size typically have smaller
regression coefficients indicating a smaller influence on maximum scour depth.

Also of note, the correlation in Figure 73 is weaker than the correlation in Figure 72.
As the number of independent variables increase in regression analysis the number of scaling

issues also increases. Using dimensional analysis and multiple regression, a number of
parameter combinations were examined investigating the effects of the armour layer on

maximum scour depth. The parameter combinations that produced strong regression models
are presented in Figure 72 and Figure 73, while remaining regression models can be found in
Appendix D.

Figure 72. Variation of maximum scour depth in relation to Froude number, median sediment size, pier size
and armour layer under open channel conditions.

132

smooth + rough cover

1.4

d^/H = 20.58 X 0.0254

1.2

=

R2 0.63

X

1

o

0.6
0.4

0.2

o

o

o

0

0.01

0.02

o o
O

o

v°

0
0

X =

—Y
\JgH
(—

/

'W'

0.03

0.05

0.04

/b\°
f^so^rmour \-047 \H
f— Y ” \D^0Channel)
)

°5

'nb'

Figure 73. Variation of maximum scour depth in relation to Froude number, median sediment size, ice cover
roughness, armour layer and pier size.

5.73 Summary and conclusions

The purpose of this chapter was to examine various hydraulic parameters as they relate
to pier scour. In doing so, 54 flume experiments were completed investigating pier scour under

open, smooth and rough ice cover conditions. In order to avoid repetition, a summary of major
findings is not provided here, but rather in the conclusion section of this thesis (Chapter 7).

In order to address whether the findings of this study are useful and relevant to bridge
design a critical analysis of bridge research and management is undertaken in the next section

of this thesis. In order to assess the validity of the physical experiments conducted, scaling
issues associated with physical models must be explored. This is an important consideration

for this study as exact similitude between a model and prototype has not yet been achieved in
any known study. The results presented in Chapter 4 and Chapter 5 are all subject to scaling
effects and experimental results typically cannot be translated directly to real world

133

applications. However, parameter relationships presented in the multiple correlation analysis
do present possible causal relationships to scour hole formation. In conducting a critical
assessment of bridge research and design, the goal is to investigate and highlight current

approaches to bridge research and highlight areas of current bridge manuals where this
research addresses knowledge gaps.

134

6.0 CRITICAL ASSESSMENT OF BRIDGE RESEARCH AND DESIGN
6.0 Introduction

To critically assess something means to examine and judge something in order to assess
its worth. In academia critical analysis involves interpretation and evaluation of a body of
work which often takes the form of an essay. Academics are often asked to explain how their
research is applicable in the real world, as the dissemination of information from academia to
industry and government can range from open and effective to difficult and confusing (Rynes
et al., 2001). Armstrong (2004) argued that only 3 percent of published papers in climate and

population forecasting contained useful knowledge that could contribute to better decision
making. The author proposed that universities could ask faculty to write short reports on: what
they discovered, how they made the discovery, and how the findings will lead to better decision
making.
As researchers have investigated pier scour for the past 50 years and published hundreds

of journal articles, one must question what further contributions this thesis can offer the field
of scour research or is it merely an ‘exercise in academia’. Secondly, can information
presented in this thesis be used or incorporated into bridge design or form the groundwork for

further studies relating river ice and pier scour.
The purpose of this chapter is to examine the various research components that involve

bridge scour research, namely flume work, scour equation development and computational

fluid dynamics, and assess where components of this thesis work relate and the overall
usefulness in current day practices. Bridge standards in North America along with scour
prevention practices are then reviewed in order to gain insight as to how academic bridge

research is used and applicable to real world bridge design problems around pier scour.

135

6.1 The use of flume experiments and their relation to field conditions
Both physical hydraulic modeling in a laboratory and numerical computer modeling are
the standard techniques used to evaluate bridge design and scour. Physical models fulfill an
important role in hydraulic analysis by verifying solutions that cannot be obtained by
numerical methods. Physical hydraulic models for bridge scour research consist of hydraulic
flumes. The model component is a small scale replica of a bridge pier or abutment while the
prototype is the fully functional real world bridge pier or abutment.

6.1.1 Similitude theory and scaling
This section addresses scaling principles that were considered under Chapter 3.

Similitude is the relationship between prototype and model with geometrically similar

boundaries (Zevenbergen et al., 2012). In the realm of fluid mechanics there are three types
of similitude: geometric, kinematic and dynamic.
(a) geometric similarity involves x, y, z length scales and imposes that the corresponding

lengths between the model and prototype have the same ratios.
(b) kinematic similarity requires that the length and time scales be similar between the model

and the prototype. Under kinematic similarity streamline patterns for water flow around a pier
model and pier prototype would be similar. In order for flow-field similitude the geometric
scale relating to the point of the stagnation head along the face of the pier must be scaled
directly with the pier size (Ettema et al., 1998). The stagnation head refers to the point in

which the water hits the pier face and the fluid velocity becomes zero. When this occurs the

kinetic energy is converted into pressure energy.

136

(c) dynamic similarity occurs between the model and prototype when the forces at

corresponding points are similar. Fluid forces that are considered in dynamic similarity are
inertia, viscosity, gravity, surface tension, pressure and the elastic compression force. This

similarity is the hardest to achieve in fluid mechanics since all non-dimensional numbers

between model and prototype relevant to the flow must be preserved. For perfect dynamic
similarity the following equations must be identical for model and prototype (Heller, 2011).

The ratio of the inertial force to the gravity force is represented by the Froude number as:

Froude number F = (-4= I

= (-==)
Xy/Wp

6.1

where M is the model and P is the prototype. The ratio of the inertial force to the viscous force

for the model and prototype is represented by the Reynolds number written as:

Reynolds number R =

6.2

The ratio of the inertial force to the surface tension is known as the Weber number:

Weber number W =

X a

X a

JP

6.3

where L is the characteristic length and o is the surface tension. The ratio of inertial force to
elastic force is represented by the Cauchy number:

137

Cauchy number C

) (—
)
= (—
E 'M =
E
'

\

/p

6.4

where E is Young's modulus, also known as the tensile modulus representing the elastic force.

The ratio of the pressure force to inertial force is represented by the Euler number:

Euler number E = (-£-)

\pu2/M

(-£-)
= Vpu1/
p

6.5

where P is the pressure.

If a hydraulic model were to have perfect geometric, kinematic and dynamic similitude
than the model would be the same size as the prototype. Perfect scaling is however not
possible. Take for example studying the flood stages of the Mississippi River which is 2,092

km long, 0.4 km wide and 6 m deep. Construction of a downscaled physical model of the
Mississippi with a length of 1 00 m would cause the model depth to decrease to a miniscule
scale of 0.0286 cm. In this case the effects of surface tension (cohesion of water molecules)

would be exaggerated in the model in comparison to the prototype (Grabel, 2001).

In order to minimize scale effects, the most common practice in flume studies is to
follow a set of limiting criteria (rules of thumb). Limiting criteria for scour studies involve
maximum and minimum ratios for pier width, flume width and sediment grain size as outlined

in Table 7, section 3.6, ‘Flume scaling and pier selection’. The limiting criterion that was not
met for all experiments in this thesis was the ratio pier width to flow depth (b/H). The ratios

were all less than 4, indicating that the resulting experimental scour depths cannot be
considered independent of flow depth. Melville and Coleman (2000) state that if the ratio of
pier width to flow depth (b/H) is between 0.7-5.0 then local pier scour is dependent on both

138

flow depth and pier width, while if b/h is less than 0.7 than local pier scour depth is dependent
only on pier width. For this thesis, the ratios of pier width to flow depth range from 0.42 to

2.2 as indicated in Table 7, Chapter 3. While the purpose of this thesis was not to investigate
scour depth in relation to flow depth, this scaling is present in the design of the flume
experiments. In order to investigate scour depth under various channel covers all the while
having flow depth as a scaling issue, the experiment parameters (depth, velocity and pier size)
remained constant with only the channel cover changing (as outlined in Table 8, Chapter 4).

With all flow and bridge parameters remaining constant, the impact of ice cover could be
investigated.

While flume scaling guidelines (rules of thumb) were followed in this thesis (with the

exception of flow depth to pier width ratio), the importance of kinematic and dynamic
similarity must be discussed. As indicated by (Heller, 201 1) the most common and important
similarity criterion applied to open channel hydraulics is Froude similarity. Under Froude

similarity, the additional force ratios outlined in equations 6.2-6 5 cannot be identical between
the model and prototype. The additional flow forces such as surface tension, elasticity and
pressure are therefore not considered and represent scale effects, however these scale effects

are small and often considered negligible (Heller, 2001; Zevenbergen et al., 2012).
While Froude similarity was considered in this thesis, in doing so the similitude of
particle mobility is affected. Similitude of particle mobility (U/Uc) and similitude of the
Froude number (U /J gD with b/H) cannot occur simultaneously (Ettema et al., 1998). The
natural scale to which river bed particles can be scaled does not allow for both criteria to be

satisfied. The majority of flume experiments use particles that are comparable in size to natural

139

river bed sediments (D50 from 0.25-4.00 mm); therefore the ratio of pier diameter (b) to particle
diameter (D50) is always larger in the prototype than the physical model (Table 18). Due to
the similar size of river bed particles for model and prototype, larger approach velocity values

are required to satisfy incipient motion (U/Uc) similitude than Froude number (U/JgD')
similitude. The result of using larger approach velocities to satisfy incipient motion similitude
leads to larger values of scour depth than those naturally occurring at bridge piers in actual
rivers (Ettema et al., 1998). In this thesis the flume was designed with the goal of obtaining

subcritical flows (Froude number < 1 ) and incipient motion. As previously indicated, flume
testing was undertaken using three tailgates (30 cm water depth) and under this depth, velocity

could not be adjusted for incipient motion to occur. In addition under the two tailgate
configuration the small pier was always placed in the upstream sandbox since under such

channel conditions in the downstream window no scour/incipient motion would occur. Due
to the design of the gravity flow water system, the flume velocity was adjusted to obtain
particle mobility similitude all the while retaining a Froude number < 1 .

In the design of this experimental study, geometric similitude was considered by
following the flume scaling guidelines and dynamic similitude was considered by the Froude
number and occurrence of incipient motion. As indicated in Mount (1995) the Reynolds
number is used to characterize whether flow is turbulent or laminar while the Froude number
characterizes whether the flow is subcritical versus supercritical. Using the Reynolds and
Froude numbers virtually all flow falls into one of four categories:

140

1. subcritical-laminar (Fr < 1, Re < 500), rare in rivers, more common in groundwater or very
slow moving deep water.

2. supercritical-laminar (Fr > 1, Re < 1), rare in rivers, occurs in sheetwash and in fast moving
water that forms a thin moving layer over smooth rocks.

3. subcritical-turbulent (Fr < 1, Re > 2,000), most common flow condition in rivers where
inertial forces exceed viscous forces.
4. supercritical-turbulent (Fr > 1, Re > 2000), found in high gradient, shallow rivers with high

velocities, spatially limited.

For the purpose of discussing similarity, a reach of the Nechako River, including the
John Hart Highway Bridge around Prince George was used as a real world comparison for how
representative the experimental flume used in this research is of real river channels. Of note
however, the flume was not built or scaled to the Nechako River or Hart Highway Bridge, this

example is merely used for the purpose of highlighting a real world bridge and channel
configuration. The reach of the Nechako around Prince George is subject to annual ice events
and the channel is composed of fine sediment. As indicated in Table 18, the Froude and

Reynolds numbers for the flume fall within the most common flow conditions found in rivers
and streams. The flow in the flume was turbulent (Re > 2,000) and subcritical (Fr < 1).

Subcritical flow can be described as slow, deep flow where ripples can travel upstream.
Reynolds turbulent flow is always irregular with eddies and vortices that can occur on many

scales. Froude values for the flume and Nechako were similar, while Reynolds numbers for
the Nechako were considerably larger due greater inertial components of velocity and
hydraulic radius. Also, as previously mentioned, the ratio of pier width to sediment D50 is

141

much larger for the Nechako and all natural rivers in comparison to the flume setup. This is
due to the fact that for this study, flume experiments use sediments that are comparable in size
to natural low gradient rivers. (Here, only natural low gradient rivers are used for comparison

as high gradient mountain streams are not considered for bridge experiments).

Summary: While there is a long standing practice of using physical hydraulic models to

investigate sediment transport around bridge structures, there is currently no known study that
successfully models such phenomena without scale effects (Heller, 2011). The design and

construction of the flume and experiments undertaken in this thesis were subject to scale
effects; the scale effects of flow depth to pier width were addressed while the scale effects of

sediment size could not be avoided. Many scour equations are developed from physical model
tests in flumes; therefore, scour equation development has evolved and is addressed in the next

section.
Table 18. Similarity parameters. Water Survey Canada data for March 4, 2014 was used to calculate Froude
and Reynolds numbers, WSC Station 08JC002, Isle Pierre.

John Hart Highway
Bridge, Nechako River,
Prince George
1.22 m
240 m
1.243 m

Parameter

Flume environment

Pier width

0.1 1 m pier
2m
0.10 m, 0.13 m,
0.21 m
0.47, 0.50, 0.58
0.17-0.26
15,400-31,400

0.22 m pier
2m
0.10 m, 0.13 m,
0.21 m
0.47, 0.50, 0.58
0.15-0.26
15,700-34,000

assume 0.50
0.26-0.29
926,721

189, 220, 234

379, 440, 468

2,103

Channel width
Channel depth

Dso (mm)
Froude number
Reynolds
number
Pier width:Dso

142

6.1.2 Scour equation development and field data

The process of developing and validating equations for pier scour depends entirely on
how representative the equations are of the real world system. Experimental flumes are subject
to scale effects and can over simplify river channel morphology and hydrology. As a result,

pier scour equations can underestimate or overestimate scour hole development. Over the past

30 years numerous studies have developed pier scour equations for open channel flow.

In

Table 19 are the most commonly cited equations for computing pier scour under open channel
conditions.

143

Table 19. Pier scour equations with associated author and study.

Author
Equation
0.65
Colorado State
University (CSU)
ds 2.0Hk,k2k3
equation (FHWA,
1993):
where ds is scour depth, H is flow depth upstream of the pier, ki is correction factor for pier
nose shape, kj is correction factor for angle of attack flow, kj is correction factor for bed
condition, b is the pier width and Fr is the Froude number.
Melville and
Sutherland (1988):
where ki is the flow intensity factor, kj is the sediment size factor, ky is the flow depth
factor, ka is the pier alignment factor, ks is the pier shape factor and b is the pier width.
Shen (1969):
/1lb\03
ds = 0.00022 (-

(£) C43

=

V /

where U is the approaching flow velocity, b is the pier width and v = 1 .0 x 1 0‘6 m2/s
Froehlich (1989):
/h \0 62 /.

ds = O.32b0Fr°-2

(y) (■"O-Lf082
b) Wen/

where 0 is the shape coefficient based on the shape of the pier nose, Fr is the Froude
number, be is the width of the bridge pier projected normal to the approach flow, b is the
pier width, H is the flow depth and Dso is the median sediment size.
Breusers et al. (1977):
ds = bfkxk2[2 tanH(H/h)]
f=

2U

U

Qfor—^Q.S

f^-lfor

0.5<

—
U

≤l

f = 1 for U/Uc > 1
width,
kl and k2 are the correction factors for the pier nose shape and
where b is pier
angle of attack respectively, H is the flow depth just upstream of the pier, V is the
approach flow velocity, Vc is the critical velocity.
if (Fr- Frc) >2
Jain and Fischer
u
(1979):

ds=2.0b(.Fr-Frcr-25(.-rs
if(F,-F„)<0
LT

ds = 1.8St(Frc)»-25(-)°-5
where Fr is the Froude number, Frc is the critical Froude number calculated by Uc/(gH)° 5.
If 0 < (Fi-Frc) < 0.2, the largest commuted scour depth is used.

144

As field data measurements of pier scour are very limited, large uncertainty exists around how

accurate laboratory equations are in predicting pier scour. Johnson (1995) examined the above

listed bridge pier scour equations and tested them on four sets of field data ( Dongguan g et al.,
1993; Froelich, 1988; Jain and Modi, 1986; Zhuravlyov, 1978). The field data sets included

pier width, flow depth, flow velocity, median particle size, pier-shape coefficient and observed

scour depth. Johnson ( 1995) found that the equation developed by Shen ( 1 969) performed best
at low Froude numbers. The scour depth equation developed by Breuser et al. (1977)

performed best for ranges of U/Uc greater than 0.5. For low H/b ranges Breuser et al. (1977)
and Shen et al. (1969) performed best, while for H/b > 1.5 the Colorado State University
equation performed best. Similarly, Mohamed et al. (2005) also used pier scour field data

from Pakistan, Canada and India to test the accuracy of the above equations. Mohamed (2005)
found that the Melville and Sutherland and the Jain and Fisher equations generally over
predicted scour depth while the Colorado State University equation gave reasonable
predictions of scour depth.

The most extensive study to evaluate scour equations based upon field data is that by

Landers and Mueller (1996). A total of 139 field pier scour measurements were obtained at
high flow in clear water conditions by the United States Geological Survey (USGS) and

various state departments. After comparing computed (CSU, Froehlich, Shen scour equations)
and observed scour depths the author’s concluded that none of the selected equations estimated

scour depth for the selected field conditions. Many of the scour equations tested over-predicted
scour depths but would also under predict very large scour depths.
One of the main issues in estimating local scour depth around bridge piers is trying to
identify the relationship between pier width and scour depth. Some studies have found a linear

145

relationship between scour depth and pier width (Landers and Mueller, 1996) while other

studies report that there is not a linear relationship between pier width and scour depth unless
complete geometric similitude is obtained between pier size, flow and bed particles (Ettema et
al., 1998). These discrepancies in combination with flume scaling issues mentioned above are

the cause of much scepticism about hydraulic flume experiments and scour equation
development. As a result bridge engineers and scientists have to use a variety of measures to

arrive at satisfactory pier scour equations for bridge design.

For guidance on evaluating pier scour, a commonly referenced document is the
Evaluating Scour at Bridges, Hydraulic Engineering Circular No. 18 (HEC-18) Manual

developed in the United States. This manual presents the current state of knowledge regarding
bridge design and construction. The HEC-18 manual also provides guidelines for: designing
new and replacement bridges to resist scour, evaluating existing bridges for vulnerability to

scour and inspecting bridges for scour. The HEC- 1 8 manual is currently in its fifth edition. In
conjunction with the HEC-18 manual, the US Department of Highways also developed HEC

software packages that accompany the HEC-1 8 manual. The CSU pier scour equation is used

in the HEC software programs for calculating pier scour. Hence, the most widely used and
accepted pier scour equation is the CSU equation. As outlined in the Federal Highway

Administration (FHWA) HEC-18 manual, the following options are available to bridge
engineers:
(a) bridge engineers can use a chosen pier scour equation to calculate an upper estimate of

scour depth.
(b) engineers can continue to use the leading pier-scour equation based upon HEC-18,
currently the Colorado State University pier scour equation.

146

(c) engineers can use a pier scour equation that was empirically developed based upon

field measurements.

While it was not an objective of this thesis to develop a pier scour equation for ice

covered conditions, it is of interest whether or not the currently used CSU scour equation can
accurately calculate the scour depths measured in this thesis. Open channel scour depths were

calculated using the CSU equation and compared to flume scour depths as indicated in Table
20. The maximum scour depth as calculated by the CSU equation is consistently larger than
the maximum scour depths measured from this study. This further reinforces that the

experimental flume used in this study was subject to scale effects that must be taken into

consideration when interpreting experimental results. This is similar to findings by Mohamed
et al. (2005) where the authors found the CSU equation over predicted scour depth when

compared to laboratory flume data. Of note however, Mohamed et al. (2005) also compared

CSU calculated scour depth to measured scour depths from field data; the authors found that
the CSU equation gave a reasonable estimate of local field scour depth. Johnson (1995) also

found that the CSU equation performed well for very low Froude numbers (Fr < 0.1 ); however,
for Froude numbers 0.25 < Fr < 0.5 the CSU equation did not perform well when applied to
field data. Mueller (1996) also compared field data to CSU calculated scour depth and found

that the CSU frequently over predicted the observed scour depth.

147

Table 20. CSU calculated and measured maximum scour depths for open channel flow. Measured scour depths
are from open channel flow conditions for this study.

Measured
CSU calculated
% CSU greater then measured

1 1 cm pier
2 - 8 cm
12- 14.5 cm
43 - 87 %

22 cm pier
2 - 1 1 cm
19-22 cm
50- 90 %

Summary: Just as found in other studies (Johnson, 1995; Mohamed et al., 2005; Mueller,
1996), the CSU equation also over predicts scour depth when compared to scour depth as

measured in this thesis. Studies that have compared scour equations to field and laboratory

data typically find that most scour equations will overestimate scour depth however, under
various hydraulic conditions equations from Table 19 have also underestimated scour depth
(Johnson and Ayyub, 1996; Ataie-Ashtiani and Beheshti, 2006; Lu et al., 2008). Any equation

that underestimates scour depth is typically not considered in bridge design as it is better to
overestimate scour depth then to underestimate it in the interest of public safety. The CSU
equation is recommended by the FHWA as it most consistently represents scour depth or

overestimates scour depth and is used as a conservative measure (Deng and Cai, 2010).

6.2 The use of CFD as it relates to bridge construction
Bridge engineers include a safety margin in bridge design in order to compensate for

variations in building materials, unknown vehicular loadings and flood events. Prior to the use

of modem bridge codes and computer software programs, precise calculations regarding scour
and loading conditions were not always possible. Since many bridges in North America were

built in the post World War II era many bridges were not constructed under modem day bridge
codes. Hydraulic issues remain the leading factor in bridge failure (Zevenbergen et al., 2012).

148

Prior to computers, equations governing fluid flow were solved by hand but now the

process is completed by computer software. The advantage of computer models is that they

do not suffer from the scale effects that physical models do (Yang, 2005). The ultimate goal

of numerical modeling is to replace the need for costly physical models; at present the scientific
community is still a long way from adequately capturing flow routines that physical models
display. The process of modeling fluid flow is referred to as computational fluid dynamics
(CFD). In order to simulate the path a fluid takes numerical equations that describe the flow

routing are selected and solved.

There are four general steps that are followed when operating CFD programs:
(1) the geometry of the problem must be defined which is completed using pre-processing

software. The boundary and initial conditions must be set.
(2) a mesh is generated. In doing so the geometry domain is divided into finite elements.

Mesh generation can be unstructured or structured. Unstructured meshes involve filling
geometry with control volumes in an irregular fashion whereas structured meshing

involves creating regular control volumes throughout the entire domain.
(3) the solver is specified and resulting algebraic equations are solved. Appropriate solvers is

still an active area of research as the topic of convergence is explored (Davidson, 2002).
(4) the solution is examined to obtain the desired information. This step can often involve

post-processing software that can load and display complex illustrations of vector flow,

field data or contour lines. The data set can be quite large with up to a million points or
more, so post-processing can be time consuming especially when comparing to field or

experimental data sets.

149

There are a number of CFD software packages available on the market today; the most

common CFD models are FLOW-3D and FLUENT. Both software programs are highly
sophisticated and widely used in industry and academic publications for solving fluid flow
around bridge piers.

6.2.1 Critical assessment of FLUENT for use in this thesis

FLUENT was originally selected for use in this thesis to simulate flow and scour hole
depth around bridge piers under ice cover. The objective was to recreate flume boundary

conditions and validate the software against experimental flume results; however, after review
and software trial it was decided that using FLUENT was not a viable endeavor. The following
sections outline a brief description of the FLUENT software and describe why FLUENT was
not used in this thesis. The purpose of this section is to shed light on the capabilities of CFD

as they relate to hydraulic problems.

FLUENT is owned and distributed by ANSYS Incorporated - a large engineering
simulation software company based in the United States. Most academic research groups
purchase a short term use licence renewable on an annual basis. Since FLUENT simulates

complex 2D and 3D flows the software cannot be run on a standard desktop computer. For

this reason FLUENT is operated out of high performance computing (HPC) centres. A one
year license for FLUENT was purchased and installed at UNBC’s HPC laboratory.

Pre-processing with FLUENT was initiated by using FLUENT'S meshing mode.
Meshing requires building geometry by the user for the flume environment, the pier and

associated ice cover through manual input. Depending on the complexity of the mesh this
process takes between 3-6 months. While the FLUENT code was written to satisfy the
150

requirements of setting up a mesh for open channel flow, there is no provision written in the

program code to create an ice covered channel. Boundary conditions can be set for the channel
bed and channel sides as is normal for open channel CFD modelling. In order to create an ice

cover for a channel the only option is to create a boundary for the channel surface similar to
those created for the channel bottom and sides. The problem this inherently presents is that
ice acts much differently than conditions set for the side or bottom of a channel. Specifically,
ice cover floats, creating dynamic forces from buoyancy different from that of the channel bed
and sides. The options for boundary conditions in FLUENT are creating slip or no-slip
boundaries with a specified roughness. A no-slip boundary condition is most appropriate for
defining an ice cover as flow velocity is expected to approach zero as it contacts the ice cover.
Boundary roughness is specified by roughness height Ks and a roughness constant Cs. The
roughness height in FLUENT is typically specified as the D50 of the bed material; however, in

the instance of defining an ice cover the roughness height would be the height of the Styrofoam
cubes used to create the flume rough ice cover and Ks would equal zero for the smooth ice

cover. The roughness constant, Cs, is typically set to 0.5 when modeling uniform sand-grain
roughness; currently there is no guideline for choosing Cs for other types of roughness (Fluent

Inc. 2006). It is suggested that for other roughness types the value of Cs should increase,
however to select an appropriate Cs value to represent rough ice cover, flow field results would
have to be compared with experimental results and calibration procedures would have to be

undertaken.
While it is unknown if an appropriate roughness constant, Cs, can be found for ice
cover, the main limiting feature in using FLUENT to model ice covered channels is that
boundaries can only be specified to move in the x or y directions. A wall or boundary can be

151

defined as moving or stationary. Moving walls are defined by velocities specified in either the
x ory direction but problems where the wall has a motion normal to itself cannot be simulated
using FLUENT (Fluent Inc. 2006). The ice cover in the flume environment was stationary in

the x and y direction due to the experimental design. However the ice cover used in this study
moved in the z direction continuously throughout the entire experimental period as the

Styrofoam floated in the flume channel. With an average of 1 million points in a generated
mesh, the micro-scale flow environment around ice cover and bridge piers is sensitive to even

minor changes in boundary roughness and movement. Being able to accurately define ice
cover is the main key to successful flow field simulation using FLUENT. Flow simulation

needs to accurately represent flume flow fields in order for scour hole depth and sediment
transport to be modeled. Without the provisions to accurately define the flume ice cover in

conjunction with the allotted time of 18-24 months required to set up, calibrate and run

FLUENT (the time required to learn the FLUENT software is significant, typical CFD projects
are undertaken during the course of Master's or PhD degrees: Acharya, 2011; Ali and Karim,
2002; Escauriaza, 2008; Inkratas, 2007; Ou, 2007; Yang, 2005; Zhang, 2005) it was assessed

that use of FLUENT in this thesis was not appropriate.

It should be noted however that one study has used FLUENT to simulate flow in an ice
covered channel; however, no winter field data was used for verification of results so the
accuracy of the entire study cannot be verified. Inkratas (2007), completed his master’s thesis
on modeling 3-dimensional flows in a scour hole of the Mackenzie Delta, Canada, using

FLUENT. There are a number of natural scour holes in the Mackenzie Delta and proposed
gas pipeline crossings have questioned the stability of the scour holes. Two publications were
produced from the Inkratas (2007) Thesis: that by Gharabaghi et al. (2007) (supervisor) and

152

Inkratas et al. (2009). In the first study, looking only at open channel conditions, Gharabaghi

et al. (2007) investigated a 30 m deep scour hole in the Mackenzie Delta. FLUENT was set

up and calibrated against open channel flow data gathered by Environment Canada using an
acoustic Doppler current profiler (ADCP). The scour hole flow field was simulated by

FLUENT with a correlation coefficient of 0.68 between the field and simulated velocity values.
In the second study, Inkratas et al. (2009) used the open channel model parameter set-up from
Gharabaghi et al. (2007) with the addition of a no-slip boundary condition set to represent ice

cover. FLUENT simulated a velocity flow field in and around the scour hole; however, there
were no field data to calibrate and validate against (since collecting field data during ice events

on the Mackenzie Delta is too dangerous). The accuracy and validity of FLUENT in modeling
flow fields under ice cover is not known from the latter study since there are no winter field
data to validate against. In addition, the flow fields under ice cover were simulated based upon
open channel correlation accuracy of 0.68; in doing so, it is suspected that compounding errors

would be present in numerical results. After review and software trial it was decided that using

FLUENT was not a viable endeavor for use in this thesis.

6.2.2 CFD programs used for hydraulic design of bridges
In North America, the most widely used program for hydraulic bridge design is the
Hydrologic Engineering Center River Analysis System, HEC-RAS, software package by the

FHWA. HEC-RAS is capable of modeling steady and unsteady flow, sediment transport and
water quality. The software performs one-dimension hydraulic calculations for natural and

constructed channels. HEC-RAS can output surface profile and cross section information such
as channel depth, channel width and total discharge. The model is also capable of producing

153

F

rating curve information and estimates of various return periods. Sediment transport potential

is calculated based on grain size fraction.
The HEC-RAS model also allows for input of ice data at bridges when calculating
water surface profiles. Model users can input ice information with either the presence of a

bridge stipulated or left absent from the model set up. Ice cover at bridges can be input as a
constant thickness or a dynamic thickness. In the case of constant thickness, the model uses
the ice thickness immediately upstream of the bridge; in the case of dynamic thickness the

model calculations will be performed at the bridge cross section. Input of ice data, discharge,

surface profiles and cross section information can be used in HEC’s pier scour software
package HEC-18.

The HEC-18 software for pier scour can calculate scour depth in 1 -dimension. The

software is set up for the river in which multiple cross sections are entered, a water surface
profile is generated and used for calibration. A design event is selected (100 year flood) and

the Colorado State University (CSU) equation is used to compute pier scour under both livebed and clear water conditions. Presently there is no provision in the CSU equation for
calculating pier scour under ice cover. Often modellers do not incorporate the presence of ice
at bridges since historically observed jams did not contact the low bridge steel (Brunner, 2010).

The HEC manual states that “little is known about the ways in which a river ice jam interacts
with the various components of a bridge” (Ameson et al., 2012, p.l 18).

6.2.3 Success and limitations of CFD

It is accepted that a well-developed numerical model can assist design engineers in
identifying crucial cases in which tests should be conducted (Yang, 2005). As scaling effects

154

are not present in numerical modeling the governing equations which best simulate the
experimental environment will be of use to the engineering community in design of bridge
structures.

However, the inability of current CFD software to accurately solve turbulent and
unsteady flow is a significant limitation. Turbulence can occur at a fine scale so a spatial grid

small enough to capture even the smallest length scale of turbulence is required.
Computational resources and issues of time efficiency arise when addressing turbulent flow

with CFD. In order to solve turbulence CFD programs employ turbulence models that use the
Reynolds-average equations. The Reynolds-average equations represent fluid motion over a

larger time scale than that of actual instantaneous turbulent time scales. In using time-averaged
scales to represent instantaneous scales for turbulence, modeling inaccuracies can occur and

users of CFD software must be aware of these limitations. Since there is no standard
turbulence model that is accepted for use in CFD the selection of turbulence model depends

on the level of accuracy required, computational resources and the actual physics of flow
(FLUENT, 2006).

As an alternative to CFD, large eddy simulation can be used to address turbulence

modeling by solving large eddy motion by using the Navier-Stokes equations for fluid. Large
eddy simulation solves large scale eddies which are assumed to be dependent on the flow
geometry. In solving for the large eddies, eddies at the smallest scale are implicitly included

in solving for the largest eddies.
The second main limitation in CFD is the inability to accurately model multi-phase

flows. A multi-phase flow consists of phases which do not chemically relate to each other.
The phases have separate volume fractions and velocity fields. The flume experiment set up

155

for this thesis is an example of multi-phase flow: water flow, sediment transport and air flow.
There are a number of models set up to handle multiphase flows, the most common being the
Euler-Euler approach and the Euler-Lagrange approach. Under the Euler-Lagrange approach

it is assumed that the first phase is fluid and is treated as a continuum by solving the timeaveraged Navier-Stokes equations. The second phase is a dispersed phase solved by tracking

a number of particles through the calculated flow field. The second phase always occupies a
lower volume fraction than the first phase and particle interactions are neglected (Zhang,
2009). In the Euler-Euler approach the phases are treated as interconnected continua. Under

this approach interactions between sediment particles are considered. Each phase is divided
into fractional volumes, continuous through space and time with their total sum equal to one.
The FLUENT user manual also states that the Euler-Lagrange model is appropriate for particle
laden flows (FLUENT, 2006). The multiphase models do not however incorporate particle
shape, deformation rate or particle coalescence and therefore do not accurately represent nonuniform sediment. The inability of FLUENT to accurately simulate non-uniform sediment
transport was also a key factor considered in deciding FLUENT was not a viable endeavor to

pursue in this thesis.

Summary: As Davidson (2002) points out, the use of CFD remains in the hands of specialists.

Significant expertise in transport phenomena, turbulence and multiphase flows is required in

order to properly select a modeling approach and define boundary conditions. Additionally,

one must have a high understanding of the relationship between mesh quality, convergence
and solution accuracy which all rely on CFD code and ultimately overall efficiency of the
modeling process (Davidson, 2002). CFD also requires significant time investment for

156

meshing and post-processing meaning that solutions cannot be calculated in the short term.
(For example, the time required to learn the FLUENT software is significant, the user manual

is 2,426 pages in length). Typical CFD projects are undertaken during the course of Master’s
or PhD degrees (Acharya, 201 1; Ali and Karim, 2002; Escauriaza, 2008; Inkratas, 2007; Ou,
2007; Yang, 2005; Zhang, 2005) where multiple years can be spent on simulating one specific

problem.

6.3 Bridges in Canada

6.3.1 Governing bodies and financial management
In Canada bridges are managed by three main administrative levels: federal, provincial
and municipal governments. However, some bridges are also managed by private forestry and
mining companies.

Federal bridge management in Canada is undertaken by Transport Canada and the
Federal Bridge Corporation Limited (FBCL). Transport Canada and the FBCL provide
management of intraprovincial, interprovincial and international bridges and associated
structures.

Specifically, the FBCL manages the Jacques Cartier, Champlain, Seaway

International, Sault Ste. Marie and Thousand Islands Bridges. All of these bridges are located

in Ontario and provide important trade routes and links between Ontario, Michigan and New
York. The Confederation Bride, connecting Prince Edward Island and New Brunswick, is

managed by Strait Crossing Development Incorporated, a private sector developer that is

responsible for operating and managing the bridge until 2032; at such time operations will be

transferred to the Government of Canada.

157

Provincial bridge management is undertaken by the ministry or department of
transportation within each province. Presently, provinces and territories (except for Ontario)

do not have a provincial regulation for bridge inspection. (Provinces typically have written

policies but no enforceable legislation). Provinces generally download the responsibility of
bridge management to local municipalities.

Municipalities are the largest and most important bridge owners in all provinces and

territories of Canada. Each province and territory has a different bridge management system

that municipalities use. For instance, Alberta uses BEADS, the Bridge Expert Analysis and
Decision Support system and for the past 20 years British Columbia has been using BMIS,
The Bridge Management Information System. British Columbia, Alberta, Ontario, Quebec

and Prince Edward Island all have computerized bridge management systems while the
remaining province and territories manually enter bridge data to spreadsheets to create an

inventory database. Bridge management in Canada is a challenge since not all provinces and
territories have central databases of bridges. Already established databases are often lacking
records, reports and inspections for bridges (Khanzada 2012).

Building Canada Fund

The Building Canada Fund, introduced in 2014, provides a total of $53 billion towards

infrastructure investment over the next 10 years (2014-2024). This is the largest provision of
infrastructure funding in Canadian history. This funding is divided amongst the provinces and
territories for infrastructure improvements. The Building Canada Fund consists of a number

of funding categories:

158

(a) Federal Gas Tax Fund: the federal Gas Tax Fund was introduced in Canada’s Economic

Action Plan for 2013. The Gas Tax Fund is legislated and set to operate over the next 10
years and provide a projected $2 billion per year to municipalities for infrastructure needs.
Funding is dispersed twice per year to provinces and territories who then allocate funding
to municipalities to support local infrastructure priorities. Municipalities are allowed to
borrow against future allocations in order to invest in long term infrastructure projects.

(b) P3 Fund: the pubic-private partnerships (P3s) are long term infrastructure projects where a

private sector company assumes responsibility for construction of infrastructure projects in
Canada. P3 projects consist of rapid transit system improvements, airport improvements to

infrastructure projects. A total of $1.5 billion has been allocated for P3 projects.

(c) National Infrastructure Component: a total of 4 billion dollars is allocated for projects of

national significance. Projects considered are those that provide the greatest economic
impact such as: port infrastructure, major roads, public transit and airports.
(d) National and Regional Projects: a total of $10 billion is being given to provinces and

territories over the next 10 years for infrastructure projects at the regional level. Eligible
projects can be anything from wastewater treatment plants to public transit to highways and
bridges. Funding is allocated to province and territories based upon per-capita calculations.

As mentioned, bridge responsibility typically falls onto municipalities. As the Canadian

government recently rolled out a large infrastructure funding package, the Building Canada

159

Fund, it is up to local municipalities to plan and lobby for funding in order to adequately

maintain bridges or replace bridges as deemed necessary.

6.3.2 Construction standards
The Transportation Association of Canada (TAC) is an organization independent from
government that offers idea exchange, technical guidelines and best practices for transportation

needs in Canada.

The association considers itself a neutral forum that does not set

transportation standards but acts as a source for technical documents and national guidelines.

In Canada, the legal mandate for construction standards of highways and bridges is the
responsibility of the provincial and territorial governments. Prior to 2000 there was three
design codes used across Canada for bridge construction: the Ontario Ministry of

Transportation’s Bridge Design Code OHBDC-91-01, the Canadian Highway Bridge Design

Code used by all province and territories except Manitoba and Ontario, and the bridge code
published by the American Association of State Highway and Transportation, which was used
by the province of Manitoba.

In 2000, the Canadian Highway Bridge Design Code

amalgamated the previous Canadian code with the Ontario code to produce one bridge design
code across Canada. Since then it has been updated and the latest code is the CAN/CSA-S6
Bridge Design Code. The Canadian Bridge Code is written and produced by numerous expert

committees from across the country; it is distributed through the TAC. In Section 1.9 of the
Canadian Bridge Code it specifies that hydraulic design of bridges must comply with the
regulatory authority (local province regulations) or in their absence the TAC’s Guide to Bridge
Hydraulics (TAC, 2004).

160

There are a number of manuals issued by the TAC that specify various construction
standards for steel, concrete bridge applications, acceptable loads, surface design and hydraulic
design. As is relevant to this thesis, only the standards pertaining to hydraulic design of

Canadian bridges will be discussed here.
The following guidelines are specified in the CAN/CSA-S6 code book and the TAC

Guide to Bridge Hydraulics (TAC-GBH), and must be considered in the design and
construction of a bridge.
(a) the opening should be large enough to avoid backwater effects and not cause realignment

of natural flows (Chapter 4).
(b) hydraulic performance of existing structures near the build site must be reviewed (Chapter
3).

(c) the bridge should be able to withstand design discharge and severe ice conditions if

applicable without compromising the structural integrity of the bridge (Chapter 4).
(d) climate change and variability should be considered in evaluating bridge design floods.

Regression analysis of two or more close-by stations is recommended (Chapter 3).
(e) computation of water surface profiles and hydraulic properties can be modeled with

common software such as HEC-RAS (Chapter 4).
(f) scour estimates based upon models such as HEC-RAS should not be relied upon without

independent checks since scour is a complex phenomenon and is not adequately

represented in these models (Chapter 4).

In the TAC-GBH, Chapter 4, Section 4.5, ‘Local Scour due to Piers and Abutments’, the
manual offers two suggested methods for measuring local pier scour. Specifically, Melville’s

161

(1997) “curves and formulas... [where] pier diameter is usually the primary parameter

affecting scour depth” (p. 81) and the US FHWA procedure based upon the CSU equation
(Richardson and Davis, 1995). Under Melville’s (1997) method, which is a slight modification

of the Melville and Sutherland (1988) equation listed in Table 19, the following procedure is
suggested (p. 84):

• Step 1: Determine effective pier width or diameter b.

• Step 2: Determine depth of approach flow under design or other desired flow
conditions, allowing for general scour if appropriate. For a pier on a floodplain use

local depth of floodplain flow.

• Step 3: Using the envelop lines of Figure 74, calculate the scour depth ds.

• Step 4: Multiple the primary scour depth, ds, by a set of K. factors to obtain the design
local scour depth (dse) as presented in the following equation.

dse = kdkskgds

6.6

Where dse is the local design scour depth, kd is the sediment size factor, ks is the pier
shape factor and kg is the pier alignment factor.

162

Figure 74. Basic local scour relationship for aligned or circular piers where d, is scour depth, b is pier
width and y is approach flow depth. Data represented in the above figure were measured from uniform
sediments. Data are also independent from sediment size effects as b/Dso>5O, (taken from Melville,
1997; TAC, 2004).

The second method for calculating scour depth as outlined in the TAC-GBH is based upon
the CSU equation (Richardson and Davis, 1 995) written as:

6.7

where k1k2k3are correction factors for pier shape, angle of attack and bed condition
respectively. The Guide notes that for the CSU equation, scour depth increases with Froude

number and does not represent grain size; therefore, with uniform bed materials scour depth

may be greater for steep rivers with larger bed materials. The TAC-GBH recommends the use
of Melville’s method rather than the CSU equation as the CSU equation “contains no explicit

factor representing grain size exclusive of armouring considerations” (p.85).
163

Specific consideration is given to effects of ice and the Guide states:
(a) Investigation of ice conditions and ice history must be completed by examining: dates of

break-up and freeze-up, published and unpublished reports of ice problems and a search for

local physical evidence of high ice conditions.
(b) Scour can occur during an ice jam break-up event due to increased flow velocities.
Additionally, ice jam runs can transport and deposit piles of rocks and boulders. This is an

important consideration since riprap is often placed around bridge piers as a scour
countermeasure. The Guide states that riprap should have a diameter that is at least equal to

the ice thickness. In the instance of ice or debris accumulations, the TAC-GBH provides the

following equation, slightly modified from Melville (1997):

,

b' =

——

[0.5ti»a+ (y-O.SDb]

6.8

where be is the equivalent width or diameter of pier with drift accumulation, ba is the width

or diameter of drift accumulation, b is the pier diameter, t is the thickness of drift accumulation
and His the approach flow depth. Here the equivalent width or diameter of pier shall be used
in place of pier width in the Melville (1997) or CSU (Richardson and Davis, 1995) equations.

Lastly, since there are three series of manuals that govern bridge hydraulic standards
(CAN/CSA-S6, TAC guide to bridge hydraulics and individual provincial manuals) it is up to

the regulatory authority and bridge owner to reference appropriate manuals for hydraulic
design floods. The TAC manual dedicates an entire chapter regarding methodology for

164

calculating return periods. As each province is different in geography and hydrology, different
return periods may be used for bridge design floods (Table 2 1 ).
Table 21. Bridge design flood frequencies as stipulated by the various Canadian Bridge Design manuals.
Provincial guidelines were taking from the TAC manual. Guide to Bridge Hydraulics, with reference to Watt et
al. (1989).

Manual

Hydraulic design
flood frequency,
years

Scour design
flood
frequency,
years
refer to TAC
manual

CAN/CSA-S6

50 unless otherwise
specified by
regulatory authority

TAC Guide to bridge

refer to appropriate 1.7 times the
provincial manuals 100- year peak
Type of structure
bridges
low volume road bridges
♦low volume roads = 500 vehicles
per day or less
main highways
secondary highways
local roads
major bridges
others
main crossings
spans > 6 m
freeways

hydraulics
Province
British Columbia

Alberta

Saskatchewan

Manitoba
Ontario

Quebec

Newfoundland

arterials
local roads
spans < 6 m
bridges & highways
main roads
secondary roads
TransCanada highway
other roads

Check flood
frequency, years
at least twice the
normal design flood
unless otherwise
specified by the
regulatory authority.

500
Return period, years
200
100

100
50
25
100
50 or less
100-50
100
100
50
25
50/25/10
100
50
25
100
50

165

6.3.3 Bridge maintenance

In Canada there are four manuals used for bridge inspections: the British Columbia
Bridge Inspection Manual, Alberta’s Bridge Inspection and Maintenance Manual, Quebec’s

Manual d'inspection des structures and Ontario's Structure Inspection Manual. Currently, the
Yukon and Northwest Territories use Alberta's manual while Saskatchewan, Manitoba, New
Brunswick, Prince Edward Island and Nova Scotia are using the Ontario manual (Khanzada,

2012). Each manual stipulates inspection and maintenance standards. Inspections vary

between provinces, with most provinces conducting bridge inspections every 2 years and
British Columbia inspecting every year.

Underwater inspections are completed by all

provinces except Alberta, Saskatchewan and the Northwest Territories. This is a cause for

concern since it is estimated that 60 percent of bridge failures result from scour (Khanzada,
2012).

All provincial governments have in-house bridge managers which partake in routine
inspections and report writing. All inspections and reports are done to the standard of the

associated provincial inspection manual used.

British Columbia has privatized the

maintenance of bridges and highway systems; maintenance contractors are hired within each
management district to observe and conduct routine maintenance. The Yukon and Northwest

Territories hire engineering firms to conduct bridge inspections. Most bridge inspections are
based upon visual observations with limited physical testing such as tapping of concrete.
Bridges in some provinces such as Ontario are graded on a scale of excellent, good, fair and
poor, while other provinces such as Alberta rank bridges on a scale of 1 to 9. In all provinces

bridge inspections must be completed under the supervision of an engineer. Bridge inspector

166

training is provided in-house by each province except for New Brunswick where training is
sought through the United States Federal Highways Administration.

6.4 Bridges in the United States
6.4.1 Governing bodies
Bridges in the United States are regulated by the Department of Transportation (DOT),

Federal Highway Administration (FHWA). The FHWA provides stewardship, research and
development for the countries' highways and bridges along with research programs to support
design and construction practices. Within the FHWA there are three bridge and structures
research and development teams that conduct and manage research in the following areas:
(a) Design and construction: reviews use of structural materials (steel, wood, concrete),

shallow and deep foundations, river hydraulics, aerodynamics, bridge security and extreme
events (floods, seismic).

(b) Infrastructure management: investigates tools and practices for bridge inspections and

monitoring, improving bridge durability and preservation.
(c) Long-term bridge performance: this team looks at developing data and knowledge to

understand bridge performance under a variety of conditions.
The FHWA main research centre is the Turner-Fairbank Highway Research Center

based in Virginia. The centre has 20 laboratories and support facilities. The center reviews
and tests solutions to transportation topics such as: concrete coatings and corrosion, stream
stability and bridges, pavement testing, steel and timber structures, and bridge durability and

167

loads. The facility has over 100 scientists and engineers trained in transportation related

disciplines.

To assist with technology transfer along with national and international

collaboration the center also employs a team of support professionals focused on information
dissemination.

In addition to the Turner-Fairbanks Highway Research Center, the FHWA also directs a
number of transportation research initiatives outlined as follows.

University Transportation Centers: the main research program under the FHWA is the
allocation of University Transportation Centers (UTC). This program began in 1 987 under the
Surface Transportation and Uniform Relocation Assistance Act and established funding that
is allocated on a competitive basis to Universities located in each of the 10 standard federal
regions. To qualify, a UTC must be located within the United States, comprise a single or
consortium of universities and must obtain matching funding from non-federal sources. The
most recent allocation of funding came from the Moving Ahead for Progress in the 21st Century

Act, July 2012, which authorized $72.5 million per year from federal funds to be allocated for

up to 35 UTCs in the United States. During 2013 all of the grants were competitively selected
and research is currently being conducted. The UTCs provide the main transportation
knowledge base outside of the Department of Transportation.

Small Business Innovation Research Program: the FHWA also invests in technological

innovation regarding bridge infrastructure through the Small Business Innovation Research

Program (SBIR). This program awards contracts to small businesses to research and develop

168

technological solutions to various transportation challenges. Each year the SBIR solicits

proposals on a number of transportation topics that are needed by the FHWA.

State Planning: each state must also set aside 2 percent of funding they receive from core
federal-aid programs to fulfill State planning and research activities. Transportation research
activities must include practical application of technology and investigating new areas of
knowledge.

Exploratory Advanced Research Program: the Exploratory Advanced Research Program
(EAR) is based upon legislation that outlines the need to conduct transportation research on a

long time scale and in high risk areas. Projects typically funded through the EAR highlight

new technology and involve projects that are not likely completed within 1 0 years. The FHWA
appoints expert panels consisting of State, Federal, academic and international transportation
experts to scope topics and ensure technical quality of sponsored research. An example of

technology research produced by the EAR is the use of self-powered wireless sensors for realt

time monitoring of potentially dangerous cracks in steel bridges. This project, competed in

I
1
J

2013, is expected to increase cement crack detection awareness. Prior to 2012, approximately

$14 million per year was allocated for exploratory advanced research; under the current
Moving Ahead for Progress in the 21st Century Act 2013, funding for the program continues.

National Cooperative Highway Research Program: the National Cooperative Highway
Research Program (NCHRP) was created in 1962 to disseminate information and conduct

independent transportation research across the country. Each State receives planning and
research funds, and is expected to contribute 5.5 percent of their funds to the NCHRP to ensure

I

its operation. Each year the American Association of State Highway Transportation Officials
(AASHTO) solicits proposals on transportation issues from private and public organizations
(universities, non-profit institutions, consulting firms). A technical review panel is formed and

contracts are awarded. Research findings are published in the NCHRP report series. Annual

funding under this program is approximately $37 million per year.

6.4.2 Construction standards
The AASHTO is a non-profit organization that represents highway and transportation
departments in all 50 states. The AASHTO acts as a liaison between the states and the federal
government. The role of the AASHTO is to provide and set technical standards for all design

and construction of bridges and highways. The AASHTO has a number of committees that

consists of state transportation representatives; the committees meet annually to address
planning, design and construction of transportation infrastructure. Specifically, the Committee

on Bridges and Structures provides key standards manuals on topics such as bridge element
inspection, seismic design and bridge construction specifications.

There are a number of manuals issued by the AASHTO that specify various
construction standards for bridges, concrete bridge applications, acceptable loads, surface

design and hydraulic design. As is relevant to this thesis, only the standards for hydraulic
design will be discussed here.
The following standards were taken from the Hydraulic Design of Safe Bridges manual

(Zevenbergen et al., 2012). Bridge engineers have a variety of choices when deciding the site

and location of a new bridge. The following hydraulic components must be considered:

170

(a) the level of service the bridge will incur. If the bridge is in a remote area with low traffic

then the bridge can be designed with a lower hydraulic capacity; this enables construction costs
to be low. Alternatively, if the bridge is expected to have a high volume of traffic and

significant economic costs would incur if the bridge was out of service then the bridge is
designed with a higher hydraulic capacity.

(b) the bridge must provide enough hydraulic capacity to avoid excessive backwater so to

prevent adverse floodplain impacts. The design, geometry and occurrence of piers must not

lead to an increase in backwater over existing conditions. In the case of bridge replacement

over an existing floodplain crossing the backwater must not exceed 1 foot (0.3048 m).

(c) the bridge must provide enough hydraulic capacity to prevent excessive velocity and shear

stress within the bridge waterway.
!

(d) the bridge must have a freeboard for a 50 year flood for low-traffic routes, and a freeboard

for a 100-year flood for high traffic routes. Under the Federal Emergency Management
f

[

Agency, the floodplain regulations state that any bridge element must not cause in increase in

the 100-year flood water surface elevation compared to existing conditions.

i

r

Specific consideration is given to effects of ice in the Hydraulic Design of Safe Bridge Manual
(Zevenbergen et al., 2012). The guidelines indicate that:

(a) the bridge design team should conduct site-specific analysis to see if ice events are relevant.

171

(b) if ice forces are deemed relevant, the hydraulic engineer involved must provide the bridge

designer with information regarding ice forces. In doing so the hydraulic engineer can

model simulations of ice jam situations using HEC-RAS. The model allows user input of
ice cover thickness, ice cover Manning’s n value and maximum mean velocity under ice
cover.

The latest edition of the HEC-18 manual outlines the recommended minimum scour design

flood frequencies:
Table 22. Bridge design flood frequencies as stipulated in FHWA HEC- 18 manual (Zevenbergcn et al., 2012).

Hydraulic design flood
frequency, Qd
Qio

Scour design flood
frequency, Qs
Q25

Scour design check flood
frequency, Qc
Qso

Q25

Q50
Qioo
Q200

Qioo

Qso
Qioo

Q200
Qsoo

The flood frequencies for scour are larger than the hydraulic design flood frequencies
since there is a 39.5 probability of exceedance that during a 50 year design life a bridge will
experience a flood that is greater than the Qioo level (Zevenbergen et al., 2012). The same
rationale goes for the check flood levels being larger than the scour design flood frequency.

In bridge design it is generally accepted that rare floods do occur and bridges may incur
occasional damage; however, major damage or collapse can cause significant economic loss,
social impacts and safety hazards. For these reasons, there is a higher hydraulic standard for
design of bridge foundations in comparison to design of bridge hydraulic capacity. Lastly, the

172

HEC-18 manual states that all bridge foundations must have a minimum factor of safety of 1.0
under the scour design check flood conditions (Table 22).

6.43 Bridge maintenance
As bridges in the United States continue to age and deteriorate, it is increasingly

important that bridges are inspected and maintained. The DOT oversees the National Bridge
Inspection Standards (NBIS) which form minimum nationwide requirements for bridges. The

NBIS originated in 1971 after the Silver Bridge collapsed in the Ohio River in 1967 during

rush-hour killing 46 people. The NBIS is a set of Federal regulations that outlines inspection
procedures, inspection frequency, qualified personnel, inspection reports and maintenance of
bridge inventory.

Since 2000, approximately 59,000 bridges have been replaced and

rehabilitated under this federal regulation. A national Bridge Inspection Training Program
exists and bridge inspectors enter their data into the National Bridge Inventory (NBI). The
training program consists of many courses up to 2 weeks long regarding inspection safety,

inspection standards, stream stability, scour, underwater bridge inspection and bridge

maintenance.

It is the responsibility of each bridge inspector to: (a) provide a thorough bridge
inspection of conditions and defects and to (b) report deficiencies to supervisors and engineers
that may impact public safety or integrity of the structure and (c) provide recommendations to

close a bridge if necessary. Bridge inspections also serve to investigate minor problems that

can be upgraded before they lead to major repairs. It is also recognized that bridge inspectors
play a key role in maintaining and adding data to the National Bridge Inventory. Issues such

as abutment tilt may only be captured after looking at multiple years of historic data. Once a

173

bridge inspection is completed it becomes a legal document. Bridge inspectors are required to

be registered engineers and have completed the FHWA bridge inspection training program.

6.5 In situ scour measurement technology and countermeasures
As scour is one of the major causes of bridge failures (Lin et al., 2010), it is important

that time and money is allocated to technology for measuring and monitoring pier scour. In
addition to providing public safety, scour monitoring technology can also provide hydraulic
data for engineers in which scour equations can be improved upon. Real world scour data can
also be used to investigate scaling issues that arise from physical modeling. Scour monitoring
systems can also reduce operating costs of transportation departments as field inspections and

underwater inspections can be reduced. Over the years a number of studies have investigated
and developed scour monitoring systems (Ballio and Radice, 2003; Lu et al., 2008; Yu, 2009)

and the Hawaii DOT funded a sonar project which measured bridge scour in order to validate

scour equations included in the HEC-18 manual. Reliable scour monitoring technology is
essential for decision making and public safety: for instance, during 1 994 flooding in Georgia,

2100 bridges were monitored and 1000 were closed and no lives were lost as a result of bridge

failures (Schall and Price, 2004).
Not all bridge sites are optimal for installation of scour monitoring equipment. High

flow conditions, high sediment concentration, low bridge clearance, severe water temperature,

floating debris and ice accumulation cause interference and damage to monitoring equipment.

In a survey conducted by Hunt (2005) debris and ice flows were shown to be responsible for
26 and 13 percent of all damage to monitoring systems. The following is an outline of the
most successful and currently used scour monitoring technology.

174

SONAR: sound navigation and ranging (SONAR) technology works by sending an acoustic
pulse from a transmitter mounted to a bridge pier. The acoustic pulse travels through the water

and is reflected when it hits the river bed. The reflected pulse is captured by the receiver and
the time elapsed is calculated. The distance from the transmitter to the river bed can be
calculated by measuring the elapsed time and acoustic signal travel speed. ETI Instrument
Systems Inc. and Nortek are scientific instrumentation companies with offices in North

America and Europe. They currently offer a SONAR scour monitoring technology as
indicated in Figure 75.

Magnetic Sliding Collar: the magnetic sliding collar consists of a vertical pipe that is driven
into the river bed typically on the upstream side of the pier. A collar is installed around the
vertical pipe and rests on the riverbed. The collar moves down the vertical pipe in the event

of scour. Magnets are attached to the collar which create a magnetic field. Inside the pipe is

a magnetic sensor which can detect the position of the magnetic collar. The design is robust
and can survive flood conditions; however the main limitation in this technology is that the
collar becomes buried during deposition events.

175

Figure 75. SONAR scour monitoring equipment (NORTEK, 2014).

Float-Out Transmitter, a simple yet effective scour measuring device is the float-out

transmitter. The device consists of a buoyant container with an internal transmitter. The
container is buried at scour-prone locations around bridge piers. Should significant scour

occur, the device will float to the water surface and activate the internal transmitter which will
send a signal to a receiver that is mounted on the bridge. If significant scour occurs during ice
events the float-out device will be stuck under the ice cover and signal transmission may not

be successful.
Each bridge site can have a unique scour environment that must be considered prior to

selecting potential scour monitoring technology. Sonar systems are ideal for bridge sites that

have little debris while float-out transmitters and magnetic collars are better suited for debris

176

laden channels. In regards to a scour sensor that would be optimal for channel ice conditions
presented in this thesis, either a sonar-type system or sliding collar would be suitable for

measuring scour under ice conditions. A sonar-type system would have to be mounted on a
bridge pier at a suitable level to avoid ice interference and damage. A sliding collar would be
significantly less in cost than installing sonar technology and would avoid damage from

surface water ice. Installing a magnetic collar would however only be useful if it is not buried
by sediment deposition.

6.6 Evaluation of this study in the context of current day bridge design
As discussed, bridge design and construction has an extensive history involving both

academic and government institutions in Canada and the US. As researchers have investigated
pier scour for the past 50 years and published hundreds of journal articles, one must question

what further contributions this thesis can offer the field of scour research or is it merely an
‘exercise in academia’. Secondly, can information presented in this thesis be used or

incorporated into bridge design or form the groundwork for further studies relating river ice
and pier scour. The following is an overview of (a) knowledge gaps in TAC-GBH, (b) an

explanation of what this study contributes to the field of bridge design and (c) how this study

can be communicated to Engineers working in bridge design.
In reviewing the TAC-GBH, it is evident there are knowledge gaps regarding ice
induced pier scour. There are no equations in the Guide regarding how to calculate pier scour
under ice cover. The issues of floating ice, ice jamming and floating debris are reviewed over

pages 97-108, while specific attention is given to ice induced pier scour in the following
statements:

177

•

Ice jams and accumulations in the vicinity of bridges may cause or contribute to
general and local scour by inducing high velocities in the obstructed cross-section”
(p-97).

• "Jams may affect bridges. .. [by] their formation or release that may cause or aggravate
river-bed scour” (p.98).

• "Scour during breakup can occur as a result of high velocities during an ice run, or
because the toe of a jam forms in the vicinity of a bridge. A worst case scenario would
be a grounded jam, with all flow passing under the toe within the scour depression.”
(pl 03).

• “Accumulation of debris against... piers...can significantly increase the potential for
local scour, as the stream flow is forced downwards under the accumulation or as the
accumulation causes the flow to impinge obliquely on submerged pier shafts and
foundations” (p. 105)
While the TAC-GBH states ice cover and jamming may result in greater scour, no literature
or data are referenced to support these statements. The Guide suggests the above statements

are issues to consider during pier design. In considering debris accumulation around piers, the
Guide does provide equation 6.8 in order to calculate a larger pier width due to debris. The

TAC-GBH also suggests that in order to protect bridge piers from increased ice induced scour,
“rip-rap should [be placed and] have a diameter at least equal to the ice thickness” (p.1 03);
again, no literature is referenced to support this suggestion. While no literature is referenced

in the ice section of the TAC-GBH, it is assumed that currently bridge engineers review the
site specific channel hydraulics and design piers taking in account the historical evidence of

178

ice jam occurrence and install rip-rap as appropriate. In section 4.4.8, the TAC-GBH states
that due to the “inherent uncertainty of scour estimates and the high risk of public safety usually

associated with foundation undermining, bridge foundations should be designed or upgraded
to provide an adequate margin of safety against scour failure. It is difficult to give general

guidance because scour estimates interact with various design features and other

considerations” (p. 93). In addressing scour uncertainty the Guide states that extra attention
should be given to extreme flows exceedance probabilities, seriousness of consequences of
total or partial pier failure and experience of the designer in comparable situations (p.94).
While the TAC-GBH does acknowledge the uncertainty in channel hydraulics and the

literature on river ice processes is referenced (Ashton, 1986; Beltaos, 1995; Gerard, 1989;
USACE, 1982), there is still a large knowledge gap surrounding how river ice processes

interact with bridge infrastructure and the process of pier scour.
Taking into consideration the above mentioned knowledge gaps, the following points

are the main results of this study that contribute to the field of bridge design and address the
knowledge gaps in the TAC-GBH.

1. This study found that in non-uniform sediments, ice cover can produce a 20-37 percent
larger scour depth than compared to open channel flow (Figure 35-37, Chapter 5). The

variation in scour depth will depend upon the range in bed sediment size which will
influence the formation of an armour layer (p. 74, Chapter 5). The TAC-GBH does suggest

that ice cover leads to greater pier scour by inducing higher velocities; however there is no
reference quantifying or explaining this phenomenon. Chapter 5, Section 3.5 of this study
addresses this knowledge gap, confirming that ice cover does in fact increase pier scour.

179

2. The second main result of this study was that under ice cover, armour layer formation can
lead to a decrease in scour depth (Chapter 5, Section 5.7). Smaller particles in the armour
layer will provide less scour protection around the pier and result in a larger local scour

depth. This indicates that an armour layer will provide scour protection as long as the critical
bed shear velocity is not large enough to cause incipient motion of the armour layer. This
study also confirms that under ice cover, the greatest scour depth always occurs at the pier

face (Figure 42-47, Chapter 5); therefore implying that rip-rap size and placement must be
designed to withstand scour hole velocities at the pier face. The TAC-GBH has an entire
chapter on scour mitigation measures (Chapter 5, p. 1 15-145); however, all guidelines are

based upon open channel flow conditions. For example, to ensure incipient motion does
not occur when selecting rip-rap size around piers, the TAC-GBH states that the “local

design velocity V should generally be at least 50 percent greater than the velocity
approaching the pier” (p. 131 ). The TAC-GBH addresses ice cover once when stating that

the safety factor (S) of 1.1 is normally used when calculating rip-rap size, except “where
there is ice or debris impact” (p. 1 29), a higher value may be used.
3. The third main result of this study was that the downflow and turbulence within the scour

hole is greater under ice cover than open channel conditions. Scour hole velocity under
rough ice cover is 29-37 percent greater than compared to open channel scour hole velocity
(Figure 56-57, Chapter 5). These findings address the knowledge gap around calculating

rip-rap sizes for bridge piers under ice cover. Local design velocities are currently estimated

under ice cover conditions and rip-rap size is calculated based upon design velocities. This
study provides evidence that scour hole velocity under ice cover is 29-37 percent greater

180

than open channel conditions (Chapter 5, Section 5.8) which can provide supporting

evidence for greater certainty in rip-rap design.
As discussed, this study addresses current knowledge gaps in the pier scour literature

and also provides supporting evidence to address unknowns in the TAC-GBH. While this is
the first study to measure under ice scour depth and scour profiles with armour layer formation,
realistically further research is required before bridge engineering committees will find

confidence in study results. Additional research exploring a range of flow velocities and
depths as they relate to pier scour under ice cover is required. Also, pier scour under ice over
for a greater range in sediment D50 should also be examined.

In the United States and Canada, Bridge Design Codes are written by various
committees comprised of professionals whom specialize in a specific bridge construction area.
Bridge hydraulic committees are made up of engineers working in academia, private industry

and local governments. Annual meetings are held to discuss code revisions and ongoing

research.

The code and reference manuals typically have 2-3 leading authors.

Recommendations and standards are written based upon citing academic and professional

literature. For example, in both the Canadian and US bridge manuals, the CSU pier scour
equation is recommended as a scour equation based upon 20 years of research using the
equation. Whether or not academic work will be recognized by bridge committees depends

on (a) whether the academic work is titled effectively and found in a search of the literature
(b) whether the study is designed well and actually builds on knowledge gaps in the current

bridge manuals and (c) whether the study is repeatable and well respected in the field of

literature. Given that this study does contribute valuable evidence that scour processes around
bridge piers is different under ice cover, communicating study results to the engineering

181

community is important Publication plans are discussed in the conclusion section of this thesis

(Chapter 7), and are the avenue in which study results will be communicated to the bridge
design community.

6. 7 Conclusion and recommendations

The purpose of this chapter was to examine the various components that involve bridge
scour research, namely flume work, scour equation development and computational fluid
dynamics and assess where components of this thesis relate and are useful in current day
practices. Additionally, bridge standards in North America along with scour prevention
practices were reviewed in order to gain insight as to whether academic bridge research is used

and applicable to real world bridge design problems around pier scour. The following points

are the conclusions drawn from this critical assessment:

(1) While there is a long standing practice of using physical hydraulic models to investigate

sediment transport around bridge structures, there is currently no known study that successfully
models such phenomena without scale effects. For the experimental flume used in this study,

issues around dynamic similitude arise because flume bed material relative to pier size is larger

than the real world prototype. In addition, similitude of particle mobility and similitude of the
Froude number cannot occur simultaneously. Also, non-cohesive sediments are traditionally
used in flumes whereas that is not the case in natural river systems.

The design and

construction of the flume and experiments undertaken in this thesis were subject to scale

effects. The scale effects of flow depth to pier width were addressed by only changing one
parameter, that being channel cover, while all other parameters remained constant. The scale

182

effects of incipient motion were investigated at specific Froude numbers < 1 , and the scale

effects of sediment size could not be avoided.
(2) The success of developing and validating equations for pier scour depends entirely on how

representative the equations are of the real world system. Experimental flumes are subject to

scale effects and can over simplify river channel morphology and hydrology. As a result, pier

scour equations can underestimate or overestimate scour hole development.

The most

commonly used pier scour equation is the CSU equation. As found in numerous studies, for

open channel flow conditions, the CSU equation also overestimated scour depth as measured
in flume experiments conducted in this thesis. As flume studies are always subject to scale
effects, scour equations should be developed with a focus on replicating field data sets rather

than flume data sets. In order for this to occur, substantial funding and effort must be directed
towards developing field data sets for a variety of bridge environments. As discussed, scour
measurement and monitoring technology such as SONAR, is currently used, however during
construction, firms must partner with both private and academic bridge engineers in order for

planning and financial investment towards scour monitoring equipment.

(3) The use of the CFD program FLUENT was assessed and it was decided that FLUENT was
not appropriate for use in this thesis due to the inability to accurately model a floating ice

cover. In addition, current CFD practices are limited due to the inability to accurately solve
turbulent, unsteady and multi-phase flows.

183

(4) Bridge management in the United States is highly organized with nationwide construction

standards and substantial federal funding backing numerous research institutions. The DOT

manages a National Bridge Inventory which provides a snap shot of bridges in the United

States. Canada on the other hand downloads bridge responsibility to local municipalities who
typically contract out bridge construction and management to private companies. (In doing so
liability is also passed from the government to private companies). Canada also operates under

numerous bridge codes and data systems. Each province has its own bridge code (adapted
from the CAN/CSA-S6 code) and various computer management systems ranging from
provincial software programs such as Alberta's BEADS, to relying on pen and paper filing
systems. Having various bridge codes for each province may be appropriate considering

varying Canadian topography and hydrology; however, there is a need to develop sufficient
bridge management software that can provide an understanding of the current state of bridges

in Canada.

(5) In reviewing the Bridge Design Manuals, it is evident there is still a large knowledge gap

surrounding how river ice interacts with bridge infrastructure and the process of pier scour.

There are no equations regarding how to calculate pier scour under ice cover. When ice cover

or debris accumulations are part of the channel dynamics, the TAC-GBH recommended that
the factor of safety be greater than 1.1 when designing rip-rap sizes. This study addresses
current knowledge gaps by providing evidence that (a) pier scour is greater under ice cover
than open channel conditions, (b) under ice cover, armour layer formation can lead to a

decrease in scour depth, and (c) the downflow and turbulence within the scour hole is greater
under ice cover than open channel conditions.

184

7.0 CONCLUSIONS AND FUTURE RESEARCH
7. 1 Thesis conclusions
The purpose of this thesis was to gain a better understanding of how ice cover is related
to local scour around bridge piers. In doing so, 54 flume experiments were completed

investigating pier scour under open, smooth and rough ice cover conditions. In order to assess
the current state of bridge design, a critical analysis of bridge research was also conducted.
The following are the major conclusions of the experimental flume component of this thesis:

•

The average maximum velocity under rough ice cover ranged from 0.36-0.43 of the total
depth and for smooth ice cover was 0.41 of the total depth. Channel depth was measured
from the channel bed to water surface. This indicates that ice cover causes the maximum
flow velocity to move closer to the channel bed in comparison to open channel conditions.

• Turbulent intensities and Reynolds stresses are affected by the presence of simulated
smooth and rough ice cover. Both turbulent intensity and Reynolds stresses are greater
under ice cover then open channel conditions.

• The smallest sediment size, D50 = 0.47 mm yielded the largest pier scour depth under all
channel covers. The largest sediment size, D50 = 0.58 mm, yielded the smallest pier scour
depth under all channel covers. This indicates that less shear velocity is required for

incipient motion of smaller D50S.

•

The average maximum scour depth for the 22 cm pier was 29, 12 and 25 percent larger

than the 11 cm pier for rough, smooth and open channel conditions respectively.

185

• Scour depth is greater under rough ice cover compared to open channel conditions for 60
percent of experiments. Scour depth is greater under smooth ice cover compared to open

channel conditions for 53 percent of experiments. The average rough and smooth ice

cover scour depths are 37 and 20 percent greater than open channel scour depth.

•

The streamwise and downward velocities at the pier face and within the scour hole are
greater under ice cover than open channel flow. For all sediment Dso’s under the 22 cm

pier, the average rough cover velocity is 37 percent greater and the average smooth cover
velocity is 1 5 percent greater than the open channel velocity. For all sediment Dso’s under

the 11 cm pier, the average rough cover velocity is 29 percent greater and the average
smooth cover velocity is 26 percent greater than the open channel velocity.

•

Regardless of cover condition, the location of maximum scour depth is always at the pier

face. Scour depth decreases towards the downstream side of the pier. Downstream

sediment deposition is greater for the 22 cm pier under all channel covers.

•

Under all channel conditions there is a positive relationship between maximum scour
depth and Froude number. Under the same Froude number, maximum scour depth is
greater under rough ice cover conditions.

•

Relating the scour depth to the densimetric Froude number provides no guidance when

applied to non-uniform sediment since the densimetric Froude number does not
adequately represent sediment nonuniformity. When scour depth and the densimetric

Froude number are compared through linear correlation, the scatter is large and no
significant relationship is present.

186

• Under all sediment sizes, the dimensionless shear stress increases with an increasing shear
Reynolds number. For the same scour depth, less dimensionless shear stress is needed

under rough ice cover compared to open channel conditions.

•

As sediment D50 increases so does the D50 of the scour hole armour layer. Under all

channel covers, the scour depth decreases as the armour layer size increases.

•

Results of multiple regression analysis indicate that the most important factors to consider

when examining pier scour under ice cover are: the Froude number, sediment D50, and

armour layer D50.

The following are the major conclusions of the critical assessment component of this thesis:

• The major scaling issues present in this study are that of dynamic similitude because flume
bed material relative to pier size is larger than the real world prototype. In addition,
Reynolds similitude cannot be satisfied while meeting Froude similitude.

• Since flume studies are always subject to scale effects, in order for pier scour equations to
become more accurate, scour equations should be developed with a focus on replicating

field data sets rather than flume data sets.

•

CFD projects require significant expertise, training, finances and time. Current CFD

practices are limited due to the inability to accurately solve turbulent, unsteady and multi¬
phase flows all the while modeling a floating ice cover.

•

Bridge management in the United States is highly organized with nationwide construction

standards and a National Bridge Inventory. Bridge management in Canada is governed
by numerous bridge codes and software management systems.

187

• Whether or not this thesis work will be referenced in present day bridge design literature
is dependent upon the success of publishing this work in a well-respected engineering
journals and whether bridge committees come across this work in their search of the bridge

literature.

7.2 Study limitations
As with most hydraulic engineering research, the success of many laboratory

experiments is related to the design and financial support for a study. Hydraulic flumes are
located in university labs all over the world and are an important piece of infrastructure when
studying river and ocean processes. Hydraulic flumes are typically designed by an engineering

company and contracted out to a construction company or the facilities department of a
university. This study required building a hydraulic flume as use of a flume at another

institution was not possible. The design of the hydraulic flume for this study was completed
by using (a) information found in the literature, (b) calculations regarding hydraulic head and

flow velocities, and (c) trial and error in testing design flows. Construction of the hydraulic
flume was completed by graduate students, a member of the university facilities department
and hiring of local contractors for initial site excavation and concrete removal. Provincial and

federal approval was required as construction occurred in the riparian area of the Quesnel
River. No professional contractor or design support was available and existing infrastructure

was an old fish spawning channel; as a result, the constructed hydraulic flume, budgetary
provisions and remote location presented study limitations which are listed as follows:

188

• The flume velocity was regulated by gravity fed water valves that could only open by
quarter revolutions at a time. This limited the precision and range of experimental flow

velocity. Due to the inability to set specific approach velocities, parameters such as
incipient motion velocity for each sediment were not measured.

• The Sontek 10 MHz ADV was large and measured a flow volume 10 cm from the
measuring head. Alternatively, if funding permitted. Sontek’s 16 MHz micro ADV would
have been suitable as the flow volume is measured 5 cm from the instrument head allowing

for increased precision in flow measurements. It is also assumed that the ADV somewhat
interferes with the natural flow dynamics by simply being present in the water while under
natural circumstances it would not be. Using a smaller ADV would reduce impact to flow

dynamics. Turbulence, Reynolds stresses and the location of flow reversal within the

scour hole are all analyzed based upon the output of the ADV measurements. Using
Sontek’s 16 MHz micro ADV would enable additional measurements of the scour hole
flow profile and ultimately a greater understanding of scour hole flow dynamics.

•

Due to the size of the flume and finances available, sand from local quarries was used
which yielded a fairly small range in D50S for this study (D50S of 0.47 mm, 0.50 mm and

0.58 mm). The quarries in the area only offered a limited range in Dsos; using larger bed

material would have provided more information around how scour depth varied under a
larger armour layer. The sand itself was not expensive, but rather high costs were

associated with transporting the sand to the remote field location of Likely, BC, where the
hydraulic flume was located. Also, since the flume required an entire dump truck load of

189

sand to fill the experimental sandbox area, additional time and labour was needed to move

sand in and out of the flume.

• The accuracy and extent of scour hole and sediment ridge contouring was limited by using
manual measuring techniques. It is standard in the flume literature to use a bed profiler to

map and display changes in scour and bed morphology. Funding limitations did not allow
for such technology. Use of a bed profiler would increase the accuracy of scour depth
measurements and also allowed for measurement of dune and ripple formation around the

bridge pier.

7.3 Strengths of study and contributions to science

Even though the flume construction and operational budget presented study limitations,
aspects of the flume design mirrored natural river systems more closely in comparison to some

smaller laboratory flumes. The size of the flume experimental section (40m x 2m), was most
likely more representative of natural river systems in comparison to shorter flumes used in the

literature (16 m x 0.60 m). The experimental section of the flume prior to the first sandbox

was 11.3 meters; this also allowed for flow to fully develop prior to passing the bridge pier.
Also, non-uniform sediment was used which is more representative of natural river systems in
comparison to uniform sediment.

Overall, this study expanded on current knowledge of how ice cover influences scour
around bridge piers. In the current issue of the TAC-GBH there is limited guidance on how to

address ice cover in bridge design and construction. It is anticipated that this study will be

190

written up in a series of publications that can addresses the knowledge gaps in the TAC-GBH.

The following is a brief outline of publications plans.
(1) Title: The impacts of ice cover on local scour around bridge piers

— experimental
an

study.

Summary: The average maximum velocity under rough and smooth ice cover was
greater than open channel conditions. Ice cover causes the maximum flow velocity to

move closer to the channel bed, which leads to greater scour depths under ice cover
conditions. There is also a positive relationship between the Froude number and
maximum scour depth. Taking into account the ice cover velocity profile and Froude
number, the resulting scour hole geometry is greater compared to open channel

conditions.
(2) Investigation of 3D scour hole flow field around bridge piers under ice cover.

Summary: The scour hole flow field under ice cover is significantly different than that

under open channel conditions. The streamwise and downward velocities at the pier

face and within the scour hole area greater under ice cover than open channel

conditions. Both turbulent intensity and Reynolds stresses are greater under ice cover
then open channel conditions. The scour hole flow field under ice cover provides
evidence as to the processes that lead to greater scour depth under ice cover.
(3) Title: Investigation of armour layer development around bridge piers under ice cover.

Summary: It was found that armour layer development was different for all 3 channel

covers: open, smooth and rough ice cover. Armour layer development appears to be
related to the sediment Dso rather than the channel cover. As sediment D50 increases
so does the D50 of the scour hole armour layer. Under all channel covers, the scour

191

depth decreases as the armour layer size increases.

Dimensional analysis and

regression are used to determine the relationship and dependence between variables.

7.4 Future work

Recommendations regarding possible future work in relation to this thesis are as follows:

•

In this study all scour hole profiles were measured using a Sontek 10 MHz ADV flow
meter. Future work should measure the scour hole flow field under ice cover with a Sontek

16 MHz micro ADV. A micro ADV would impact the flow to a lesser extent and enable
measurements around the entire scour hole. In doing so a greater understanding of the

turbulent intensities and Reynolds stresses under ice cover could be investigated.

•

In future flume experiments investigating pier scour under ice, using a bed profiler to
measure changes in scour morphology would yield greater understanding of sediment
transport processes in and around the pier scour hole. Using a bed profiler would also

allow for measurement of dune and ripple formation around the bridge pier which may
provide useful information for rip-rap placement and scour countermeasure design.

•

Future flume experiments investigating pier scour under ice should use larger bed
sediment (D50) and explore a greater range in flow velocities.

Maintaining an

experimental ratio of b/h>5 (Melville and Coleman, 2000) will ensure that resulting scour
depth is independent of flow depth and only proportional to pier width.

•

Future flume experiments investigating pier scour under ice cover would benefit from

pump and water control infrastructure so that flow velocities can be adjusted in small

increments. Incremental adjustment of water velocity in addition to the installation of

192

underwater cameras would allow for the study of incipient motion of sediment under ice

cover conditions which would build on the previous study by Wang et al. (2008).
In the TAC-GBH, Melville’s (1997) curve is used, Figure 74, to calculate scour depth
based upon approach flow depth and pier size. Future flume experiments could expand

upon this study and address gaps in the TAC-GBH by investigating the relationship
between ds/b and H/b under ice cover; this would basically recreate Melville’s (1997)

relationship, except data would represent ice cover conditions. Flume experiments
investigating this relationship should be independent from sediment size effects with

b/Dso>5O.

193

8.0 REFERENCES
Aberle, Q. and Nikora, V. Statistical properties of armored gravel bed surfaces. Water
Resources Research, 42, 2006. DOI. 10.1029/2005WR004674
Acharya, A. 2011. Experimental study and numerical simulation of flow and sediment
transport around a series of spur dikes. PhD Dissertation. University of Arizona, Tucson,
Arizona.
Alabi, P. 2006. Time development of local scour at a bridge pier fitted with a collar. MSc
Thesis. University of Saskatchewan, Saskatoon, Saskatchewan.

Ali, K. and Karim, O. 2002. Simulation of flow around piers. Journal of Hydraulic Research,
40(2), 161-174.

Andrey, B. and Gareth, P. 2000. Flume study of the effect of relative depth on the incipient
motion of coarse uniform sediments. Journal of Water Resources Research, 36(2),
619-628.
Assaad, F., LaMoreaux, P. and Hughes, T. (Eds.) 2004. Field methods for geologists and
hydrogeologists. Springer, 377p.

Armstrong, J. 2004. Does an academic research paper contain useful knowledge?
Australasian Marketing Journal, 12(2), 62-63.
Ameson, L., Zevenbergen, L., Lagasse, P., and Clopper, P. 2012. Evaluating scour at bridges,
HEC-18, 5th Edition. FHWA-HIF Report 12-003, Federal Highway Administration U.S.
Department of Transportation, Washington, D.C

Ashton, G.D. (ed.) 1986. River and lake ice engineering. Water Resources Publications,
Littleton, Colorado.
Ataie-Ashtiani, B., and Beheshti, A.A. 2006. Experimental investigation of clear-water local
scour at pile groups. Journal of Hydraulic. Engineering, 132(10), 1100-1104.
Ataie-Ashtiani, B. and Aslani-Kordkandi, A. 2013. Flow field around single and tandem piers.
Flow Turbulence and Combustion, 90, 471-490.

Aziz, I. 1983. Dimensional analysis as applied to scouring around bridge piers in alluvial
rivers. Jemel Teknologi Billiten, 3, 59-66.

Babaeyan-Koopaei, K. and Valentine, E. 1999. Bridge pier scour in self-formed laboratory
channels. Proceedings from the 28th IAHR Congress, Graz, Austria, pp. 22-27.

194

Ballio, F. and Radice, A. 2003. A non-touch sensor for local scour measurements. Journal of
Hydraulic Research, 41(1), 105-108.
Barbhuiya A. and Dey, S. 2003. Vortex flow field in a scour hole around abutment.
International Journal of Sediment Research, 1 8(4), 310-325.

Batuca. D. and Dargahi. B. 1986. Some experimental results on local scour around cylindrical
piers for open and covered flows. Proceedings from the 3rd International Symposium of
River Sedimentation, University of Mississippi, pp. 1095-1104.
Beg, M. and Beg, S. 2013. Scour reduction around bridge piers. International Journal of
Engineering Inventions, 2(7), 7-15.
Beheshti, A. and Ataie, B. 2008. Analysis of threshold and incipient conditions for sediment
movement. Journal of Coastal Engineering, 55, 423-430.
Beltaos, S. (ed.) 1995. River ice jams. Water Resources Publications, Littleton, Colorado.

Beltaos, S. 2000. Advances in river ice hydrology. Hydrological Processes, 14, 1613-1625.

Bong, C. 2012. Incipient motion of sediment in open channels: a comparison between
laboratory data and site observation. Proceedings of 2nd International Conference on
Water Resources, Malaysia.
Breusers, H., Nicollet, G. and Shen, H. 1977. Local scour around cylindrical piers. Journal of
Hydraulic Resources, 15(3), 21 1-252.
Briaud, J., Brandimarte, L., Wang. J. and D’Odorico, P. 2007. Probability of scour depth
exceedance owing to hydrologic uncertainty. Georisk: Assessment and Management of
Risk for Engineered Systems and Geohazards, 1(2), 77-88.

Brunner, G. 2010. River Analysis System User’s Manual HEC-RAS, version 4.1. US Army
Corps of Engineers, California.
Buffington, J. and Montgomery, D. 1997. A systematic analysis of eight decades of incipient
motion studies, with special reference to gravel-bedded rivers. Water Resources
Research, 33(8), 1993-2029.

Bunte, K. and Abt, S.R. 2001. Sampling surface and subsurface particle-size distributions in
wadable gravel and cobble bed streams for analyses in sediment transport, hydraulics,
and streambed monitoring. Technical Report RMRS-GTR-74. Fort Collins, CO: U.S.
Department of Agriculture, Forest Service, Rocky Mountain Research Station. 428p.

195

Carey, K. 1966. Observed configuration and computed roughness of the underside of river ice,
St Croix River, Wisconsin. Professional paper 550-B. US Geological Survey:
Washington, USA, B192-B198.

Chang, H. 1992. Fluvial Processes in River Engineering. Krieger Publishing Company,
Florida, 432p.
Chiew, Y.M. and Melville, B.W. 1989. Local scour at bridge piers with non-uniform
sediments. Proceedings of the Institution of Civil Engineers, Part 2, 87, 215-224.
Church, M., Hassan M.A. and Wolcott, J.F. Stabilizing self-organized structures in gravel-bed
stream channels: field and experimental observations, Water Resources Research, 1998,
34(11), 3169-3179.

Clifford, N. and French, J. 1993. Monitoring and modelling turbulent flows: historical and
contemporary perspectives. In Turbulence: Perspectives on flow and sediment transport,
(Eds. Clifford N., French, J., and Hardistry, J). Wiley: Chichester, 1-34.
Crance, M. and Frothingham, K. 2008. The impacts of ice cover roughness on stream
hydrology. Proceedings from the 65th Eastern Snow Conference, Vermont, USA. 149165.
Davidson, D. 2002. The role of computational fluid dynamics in process industries. The
Bridge, 32(4), 9-14.

Deng, L. and Cai, C. 2010. Bridge Scour: prediction, modeling, monitoring and
countermeasures - review. Practice Periodical on Structural Design and Construction,
15(2), 125-134.
Dey, S. and Raikar, R. 2007. Clear-water scour at piers in sand beds with an armor layer of
gravels. Journal of Hydraulic Engineering, 133, 703-711.

Diplas, P. and. Fripp, J.B. 1992. Properties of various sediment sampling procedures. Journal
of Hydraulic Engineering, 118 (7), 955-970.
Diab, R., Zanke U. and Link, O. 2010. 3D turbulent flow field at square pier in a gravel scour
hole. International Conference on Fluvial Hydraulics “River Flow 2010” Braunschweig,
691-697.

Dongguang, G., Pasada, L. and Nordin, C. 1993. Pier scour equations used in the People’s
Republic of China. Report No. FHWA-SA-93-076. FHWA, Washington, DC.
Escauriaza, C. 2008. Three-dimensional unsteady modeling of clear-water scour in the vicinity
of hydraulic structures: Lanrangian and Eulerian Perspectives. PhD Dissertation,
University of Minnesota, Minneapolis, Minnesota.

196

Ettema, R. 1980. Scour at bridge piers. PhD Dissertation, Auckland University, Auckland,
New Zealand.
Ettema, R. 1984. Sampling armor-layer sediments. Journal of Hydraulic Engineering, 10(7),

992-996.

Ettema, R. Braileanu, F. and Muste, M. 2000. Method for estimating sediment transport in ice
covered channels, Journal of Cold Regions Engineering, 14(3), 130-144.
Ettema, R. Melville, B. and Barkdoll, B. 1998. Pier width and local-scour depth in Stream
stability and scour at highway bridges Compendium of stream stability and scour papers
ASCE Water Resources Engineering Conferences, (Eds. Richardson, E and Lagasse, P).
Nashville, Tennessee, 1991-1998.

Ettema, R. and Muste, M. 2004. Scale effects in flume experiments on flow around a spur
dike in flatbed channel. Journal of Hydraulic Engineering, 130(7), 635-646.

Federal Highway Administration. 1993. Evaluating scour at bridges. HEC-18, Report No.
FHWA-IP-90-017, Federal Highway Administration U.S. Department of
Transportation, Washington, D.C.

Faruque, Md. Abdullah Al. 2009. Smooth and rough wall open channel flow including effects
of seepage and ice cover. PhD Dissertation, University of Windsor, Windsor, Ontario.
Fluent Inc. 2006. FLUENT 6.3 users guide. Fluent Inc., Canonsburg, PA. p.2428.
Froehlich, D. 1988. Analysis of onsite measurements of scour at piers. Proceedings from
ASCE National Hydrology and Engineering Conference, New York, NY.
Froehlich, D. 1989. Local scour at bridge abutments. Proceedings from the 1989 National
Conference on Hydraulic Engineering, New York, 13-18.

Garcia, M. (ed.) 2008. Sediment Engineering - Processes, Measurements, Modeling and
Practice. ASCE Manuals and Reports on Engineering Practice No. 1 10.
Gerard, G. 1989. Ice jam floods. Chapter 10 in Hydrology of Floods in Canada: A guide to
planning and design, (Ed. Watt, W.E.) National Research Council, Ottawa.
Gharabaghi, B., Inkratas, C., Beltaos, S. and Krishnappan, B. 2007. Modelling of threedimensional flow velocities in a deep hole in the East Channel of the Mackenzie Delta,
Northwest Territories. Canadian Journal of Civil Engineering, 34, 1312-1323.

Graebel, W. 2001. Engineering Fluid Mechanics. CRC Press, 752p.

197

Graf, W. and Istiarto, W. 2002. Flow pattern in the scour hole around a cylinder. Journal of
Hydraulic Research, 40(1), 13-20.
Guo, J. 2012. Pier scour in clear water for sediment mixtures. Journal of Hydraulic Research,
50(1), 18-27.
Hager, W. 1999. Wastewater Hydraulics: Theory and Practice. Springer: Berlin, New York.

Hains D.B. and Zabilansky L. 2004. Laboratory test of scour under ice: data and preliminary
results. Cold Regions Research and Engineering Laboratory, ERDC/CRREL TR-04-09:
Hanover, NH, USA.
Hains, D. and Zabilansky, L. 2007. Scour under ice: Potential contributing factor in the
Schoharie Creek Bridge Collapse. From Current Practices in Cold Regions Engineering,
in Proceedings of the
International Conference on Cold Regions Engineering,
Maine, USA.
Hains, D. 2004. An experimental study of ice effects on scour at bridge piers. PhD Dissertation,
Lehigh University, Bethlehem, PA.
Heller, V. 201 1. Scale effects in physical hydraulic engineering models. Journal of Hydraulic
Research, 49(3), 293-306.
Hodi, B. 2009. Effect of blockage and densimetric Froude number on circular bridge pier local
scour. MSc Thesis. University of Windsor, Ontario, Canada.

Hoffmans, G. and Verheij, H. 1997. Scour manual. CRC Press, 205p.

Hoque, M. 2009. Hydraulic analysis of ice-covered river flow. Master’s Thesis, Concordia
University, Montreal, Canada.
Hunt, B.E. 2005. Practices for Monitoring Scour Critical Bridges. NCHRP Project 20-5 first
draft report, Transportation Research Board, National Research Council, National
Academy Press, Washington, D.C.
Inkratas, C. 2007. Stability analysis of a deep hole in the east channel of the Mackenzie Delta,
Northwest Territories. MSc Thesis, University of Guelph, Guelph, Ontario.
Inkratas, C., Gharabaghi, B., Beltaos, S. and Krishnappan, B. 2009. 3D modelling of icecovered flows in the vicinity of a deep hole in the East Channel of the Mackenzie Delta,
N.W.T. Canadian Journal of Civil Engineering, 36, 791-800.

Jain, S.C. and Fischer, E.E. 1979. Scour around bridge piers at high Froude numbers. Report.
No. FHWA-RD-79-104, Federal Highway Administration, Washington, D.C.

198

Jain, B. and Modi, P. 1986. Comparative study of various formulae on scour around bridge
piers. Journal of the Institution of Engineers (India), 67, Part 3.
Johnson, P.A. 1995. Comparison of pier-scour equations using field data. Journal of Hydraulic
Engineering, 121(8), 626-629.
Johnson, P.A., and Ayyub, B.M. 1996. Modelling uncertainty in prediction of pier scour.
Journal Hydraulic Engineering, 122(2), 66-72.
Katess, G. and Mao, L. 2011. Numerical simulation of armour layer development under
conditions of sediment starvation. Proceedings from the National Conference of the
Italian Association of Agricultural Engineering, Belgirate. 1-7.
Kanellopoulos, P. 1998. Incipient motion under shallow flow conditions. MSc Thesis. Virginia
Polytechnic Institute and State University, Blacksburg, Virginia.
Khanzada, K. 2012. State of bridge management in Canada. Master’s Thesis. North Dakota
State University, Fargo, North Dakota.
Khorsandi, Babak., Mydlarski, L. and Gaskin, S. 2012. Noise in turbulence measurements
using acoustic Doppler velocimetry. Journal of Hydraulic Engineering, 138(10), 829838.

Khwairakpam, P., Ray, S., Das, S., Das, R. and Mazumdar, A. 2012. Scour hole characteristics
around a vertical pier under Clearwater scour conditions. ARPN Journal of Engineering
and Applied Sciences, 7(6), 649-654.

Kumar, A. and Kothyari, U. 2012. Three-dimensional flow characteristics within the scour
hole and circular uniform and compound piers. Journal of Hydraulic Engineering,
138(5), 420-429

Lagasse, P. 1999. National Research Council, Research Results Digest: 1998 Scanning review
of European practices for bridge scour and stream stability countermeasures.
Transportation Research Board of the National Academics, Washington, D.C.
Lagasse, P., Zevenbergen, L., Schall, J. and Clopper, P. 2001. Bridge scour and stream
instability countermeasures: Experience, selection, and design guidelines. FHWA NHI
01-003: Federal Highway Administration, Hydraulic Engineering Circular No. 23, 2nd
ed., U.S. Department of Transportation, Washington, D.C.
Lagasse, P. Clopper, P. Zevenbergen, L. and Girard, L. 2007. Countermeasures to protect
bridge piers from scour. National Cooperative Highway Research Program (NCHRP)
Report No. 593, Transportation Research Board, Washington, D.C.

199

Landers, M. and Mueller, D. 1996. Evaluation of selected pier-scour equations using field data.
Transportation Research Record, 1523, 186-195.

Lau, L. and Knshnappan, B. 1985. Sediment transport under ice cover. Journal of Hydraulic
Engineering, 3(6), 934-951.
Lauchlan, C. 1999. Countermeasures for pier scour. PhD Thesis. University of Auckland,
Auckland, New Zealand.
Lauchlan, C. and Melville, B. 2001. Riprap protection at bridge piers. Journal of Hydraulic
Engineering, 127(5), 412-418.

Lee, S. and Sturm, T. 2009. Effect of sediment size scaling on physical modeling of bridge
pier scour. Journal of Hydraulic Engineering, 135(10), 793-802.
Li, S.S. 2012. Estimates of the Manning’s coefficient for ice covered rivers. Proceedings of
the Institution of Civil Engineers. Water Management, 165(9), 495-505.
Lin, Y., Lai, J., Chang, K., Chang, W., Lee, F. and Tan, Y. 2010. Using mems sensors in the
bridge scour monitoring system. Journal of the Chinese Institute of Engineers, 33( 1), 2535.
Link, O., Pfleger, F. and Zanke, U. 2008. Characteristics of developing scour holes at sand
embedded cylinder. International Journal of Sediment Research, 23, 258-266.

Lu, J., Hong, J., Su, C., Wang, C., and Lai, J. 2008. Field measurements and simulation of
bridge scour depth variation during floods. Journal Hydraulic Engineering, 134(6), 810821.
Mao L., Cooper, J. and Frostick, L. 2011. Grain size and topographical differences between
static and mobile armour layers. Earth surface Processes and Landforms, 36(10), 13211334.
Mays, L.W. 1999. Hydraulic Design Handbook. McGraw-Hill: New York, USA.
Melville, B.W. 1997. Pier and abutment scour: integrated approach. Journal of Hydraulic
Engineering, 123(2), 125-136.
Melville, B. 2008. The physics of local scour at bridge piers. Proceedings from the 4th
International Conference on Scour and Erosion, Tokyo, Japan.

Melville, B. and Chiew, Y. 1999. Time scale for local scour at bridge piers. Journal of
Hydraulic Engineering, 125(1): 59-65:

200

Melville, B. and Coleman, S. 2000. Bridge Scour. Water Resources Publications, LLC,
Highlands Ranch, Colorado, USA.
Melville, B. and Sutherland, A.J. 1988. Design method for local scour at bridge piers. Journal
Hydraulic Engineering, 11 4(10), 121 0-1 226.
Miranda, K. 2004. Effects of the upstream length of a fixed ice cover on local scour at a bridge
pier. MSc. Thesis. Lehigh University, Bethlehem, Pennsylvania.
Mohamed, T., Noor, M., Ghazali A. and Huat, B. 2005. Validation of some bridge pier scour
formulae using field and laboratory data. American Journal of Environmental Sciences,
1(2), 119-125.
Molinas, A. 2003. Bridge scour in non-uniform sediment mixtures and in cohesive materials:
synthesis report. Report No. FHWA-RD-03-083, Federal Highway Administration,
McLean, Virginia.
Molinas, A. and Wu, B. 1998. Effect of size gradation on transport of sediment mixtures.
Journal of Hydraulic Engineering, 24(8), 786-793.

Mostafa, E. 1 994. Scour around skewed bridge piers. PhD Dissertation. Alexandria University,
Alexandria, Egypt.

Mount, J. 1995. California rivers and streams: the conflict between fluvial process and land
use. University of California Press, 359p.
Mueller, D.S. 1996. Local scour at bridge piers in nonuniform sediment under dynamic
conditions. Ph D. Dissertation, Colorado State University, Fort Collins, Colorado.

Muste, M., Braileanu, F., and Ettema, R. 2000. Flow and sediment transport measurements in
a simulated ice-covered channel. Water Resources Research, 36(9), 2711-2720.
Muzzammil, M., Gangadharaiah, T. and Gupta, A. K. 2004. An experimental investigation of
a horseshoe vortex induced by a bridge pier. Water Management Journal, Proceedings
of the Institution of Civil Engineers, Thomas Telford Journals, London, 157 (2), 109119. Paper 13904.

Nowell, A. and Jumars, P. 1987. Flumes: theoretical and experimental considerations for
simulation of benthic environments. Oceanography and Marine Biology Annual Review,
25,91-112.
Oliveto, G. and Hager, W. 2002. Temporal evolution of clear-water pier and abutment scour.
Journal of Hydraulic Engineering, 128(9), 811-820.

201

Olsson, P. 2000. Influence of ice cover on local scour at circular bridge piers: an experimental
study. PhD Dissertation, Lulea University of Technology, Lulea, Sweden.
Olufayo, O., Otieno, F. and Ochieng, G. 2010. Sand water storage systems using coefficient
of uniformity as surrogate for optimal design: a laboratory study. International Journal
of the Physical Sciences, 5(6), 1227-1230.

Ou, Z. 2007. Numerical simulation of flow around vertical cylinders. PhD Dissertation,
University of Western Australia, Crawley, Australia.

Ozalp, M. 2013. Experimental investigation of local scour around bridge pier groups. MSc
Thesis. Middle East Technical University, Cankaya Ankara, Turkey.
Proffitt, G.T. 1980. Selective Transport and Armoring of Non-uniform Alluvial Sediments,
PhD Dissertation, University of Canterbury, Christchurch, New Zealand.
Raikar, R. and Dey, S. 2005. Clear-water scour at bridge piers in fine and medium gravel beds.
Canadian Journal of Civil Engineering, 32, 775-78 1 .

Raikar, R. and Dey, S. 2009. Maximum scour depth at piers in armour-beds. Journal of Civil
Engineering, 13(2), 137-142.
Raudkivi, A.J. and Ettema, R. 1983. Clear-water scour at cylindrical piers. Journal of
Hydraulic Engineering, 109(3), 339-350.
Raudkivi, A. J. and Ettema, R. 1985. Scour at cylindrical bridge piers in armored beds. Journal
of Hydraulic Engineering, 111,713-731

Richardson, E. and Davies, S. 1995. Evaluating scour at bridges. Rep. No. FHWAIP-90-017
(HEC-18), FHWA, U.S. Department of Transportation, Washington, D.C.
Richardson, E. and Davis, S. 2001. Evaluating scour at bridges, 4th edition. Publication No.
FHWA NHI 01-001. p380.

Richardson, E., Simons, D. and Julien, P. 1990. Highways in the river environment
publication. HI-90-016, Federal Highway Administration, McLean, Virginia.
Robert, A. and Tran, T. 2012. Mean and turbulent flow fields in a simulated ice-covered
channel with a gravel bed: some laboratory observations. Earth surface processes and
landforms, 37, 951-956.

Rynes, S., Bartunek, J. and Daft, R. 2001. Across the great divide: knowledge creation and
transfer between practitioners and academics. Academy of Management Journal, 44(2),
340-355.

202

Schall, J. and Price, G. 2004. Portable Scour Monitoring Equipment. NCHRP Report 515,
Transportation Research Board, National Research Council, National Academy Press,
Washington, DC.
Shen, H. and Schneider, V. 1969. Local scour around bridge piers. Journal of the Hydraulics
Division, Proceedings of the American Society of Civil Engineers, 95(6), 1919-1941.
Shen H., Schneider V.R., and Roper A.T. 1966. Analytical Approach to local scour.
Proceedings of the 12th Congress, IAHR, Vol. 3, Colorado State University.
Shen, H. and Wang, D. 1995. Under cover transport and accumulation of frazil granules.
Journal of Hydraulic Engineering, Vol. 121, No. 2, 184-195.

Sheppard, D., Odeh, M. and Glasser, T. 2004. Large scale clear-water local pier scour
experiments. Journal of Hydraulic Engineering, 130(10), 957-963.
Simons, D.B. and Senturk, F. 1992. Sediment Transport Technology. Water and Sediment
Dynamics. Water Resources Publications, Littleton, Colorado, 897p.
Smith, B. and Ettema, R. 1997. Flow resistance in ice covered alluvial channels. Journal of
Hydraulic Engineering, 123(7), 592-599.

SonTek. 1997. Pulse coherent Doppler processing and ADV correlation coefficient. SonTek
Technical Notes, www.sontek.com
Sontek. 2014. Website: <www.sontek.com> Accessed March 23, 2014.

Speziale, C. 1990. Analytical methods for the development of Reynolds stress closure in
turbulence. NASA Contractor Report 182017, 1CASE Report No. 90-26, Virginia.
Sui, J., Wang, D. and Kamey, B. 2000. Sediment concentration and deformation of riverbed
in a frazil jammed river reach. Canadian Journal of Civil Engineering, 27(6), 1120-11 29.
Sui, J., Wang, J., He, Y. and Krol, F. 2010. Velocity profiles and incipient motion of frazil
particles under ice cover. International Journal of Sediment Research, 25(1), 39-51.

TAC. 2004. Transportation Association of Canada, Guide to bridge hydraulics, 2nd Edition.
Transportation Association of Canada, Thomas Telford Publishing, 18 Ip.
Unger, J. and Hager, W. 2007. Down-flow and horseshoe vortex characteristics of sediment
embedded bridge piers. Experimental Fluids, 42, 1-19.
USACE. 1982. Ice engineering. Engineer Manual 1110-2-1612, U.S. Army Corps of
Engineers, Washington D.C.

203

Vollmer, A. and Kleinhans, M. 2007. Predicting incipient motion, including the effect of
turbulent pressure fluctuations in the bed. Water Resources Research, 43, 1-16.

Wang, J., Sui, J., and Kamey, B. 2008. Incipient motion of non-cohesive sediment under ice
cover - an experimental study. Journal of Hydrodynamics, 20(1), 117-124.
Watt, W.E. (Ed). 1989. Hydrology of Floods in Canada, A Guide to Planning and Design.
National Research Council of Canada, Ottawa.
Wu, P., Hirshfield, F., Sui, J., Wang, J., and Chen, P. 2014. Impacts of ice cover on local scour
around semi-circular bridge abutment. Journal of Hydrodynamics, 26(1), 10-18.

Wu, W., Wang, S., and Jia, Y. 2000. Non-uniform sediment transport in alluvial rivers. Journal
of Hydraulic Research, 38(6), 427-434.
Xu, H., Lu, J. and Liu, X. 2008. Non-uniform sediment incipient velocity. International
Journal of Sediment Research, 23, 69-75.

Yang C. 2003. Sediment transport, theory and practice. Krieger publishing company, Florida,
396p.

Yang, Q. 2005. Numerical investigations of scale effects on local scour around a bridge pier.
MSc. Thesis, Florida State University, Tallahassee, Florida.
Yanmaz, A. 2002. Dynamic reliability in bridge pier scouring. Turkish Journal of Engineering
and Environmental Sciences, 26, 367-375.

Yu, X. 2009. Experimental Study of an innovative bridge scour sensor. PhD Dissertation. Case
Western Reserve University, Cleveland, Ohio.
Zabilansky, L. 1996. Ice force and scour instrumentation for the White River, Vermont. US
Army CRREL Special Report, 59p, Hanover, USA.

Zabilansky, L., Hains, M. and Remus, J. 2006. Increased bed erosion due to ice. Proceedings
from the 13th International Conference on Cold Regions Engineering, Maine, USA.
Zevenbergen, L., Ameson, L., Hunt, J. and Miller, A. 2012. Hydraulic design of safe bridges.
FHWA-HIF Report 12-018-HDS 7.
Zhang, H. 2005. Study of flow and bed evolution in channels with spur dykes. PhD
Dissertation, Kyoto University, Japan.

Zhang, L. 2009. 3D numerical modeling of hydrodynamic flow, sediment deposition and
transport in stormwater ponds and alluvial channels. PhD Dissertation, Old Dominion
University, Norfolk, Virginia.

204

Zhao, M., Cheng, L. and Zaong, Z. 2010. Experimental and numerical investigation of local
scour around a submerged vertical circular cylinder in steady currents. Coastal
Engineering, 57, 709-721.

Zhuravlyov, M. 1978. New method of estimation of local scour due to bridge piers and its
substantiation. Transactions, Ministry of Transport Construction, State All Union
Scientific Research Institute on Roads, Moscow, Russia, 4-51.

205

APPENDIX A - FLUME DATA
The data in this appendix represents the measured experimental flume conditions along with
the values of calculated hydraulic parameters for all 54 flume experiments. Data are sorted by
cover type rather than run number as that is how is was analyzed and presented in this thesis.
Table 23. Measured experimental data for 54 flume runs.

Date
run
finished

Run
#

Sep-23 1
Sep-26 2
Sep-22 1
Sep-27 4
Sep-28 5
Sep-29 6
Oct-03 10
Oct-04 11
0ct-05 12
Oct-O2 9
Oct-Ol 8
Sep-30 7
Oct-07 14
0ct-08 15
0ct-06 13
Oct-09 16
Oct-10 17
Oct-11 18
Oct-19 20
0ct-20 21
Oct-18 19
Oct-21 22
Oct-23 24
Oct-22 23
Oct-27 28
Oct-28 29
Oct-29 30
Oct-26 27
Oct-25 26

Cover

open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice

Position in
flume
upstream
upstream
downstream
upstream
downstream
downstream
upstream
upstream
downstream
upstream
downstream
downstream
upstream
upstream
downstream
upstream
downstream
downstream
upstream
upstream
downstream
upstream
downstream
downstream
upstream
upstream
downstream
upstream
downstream

#
#
valves Tailgate

1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25

1
2
1
1
2
1
1

2
1
1
2
1
1
2
1
1
2
1
1
2
1
1

2
1
1

2
1
1
2

Dso

d90a1^

mm
m
0.58 0.37
0.58 0.37
0.58 0.37
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.47
0.47

0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47

0.47

0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.37
0.31
0.31
0.31
0.31
0.31
0.31
0.31
0.31
0.31
0.31
0.31

Pier
width
cm
11
11
11
22
22
22
11
11
11
22
22
22
11
11
11
22
22
22
11
11

11
22
22
22
11
11
11
22
22

206

Oct-24
Oct-31
Nov-01
Oct-30
Nov-02
Nov-04
Nov-03
Nov-08
Nov-09
Nov-07
Nov-10
Nov-12
Nov-11
Nov-15
Nov- 16
Nov- 18
Nov-17
Nov-14
Nov-13
Nov-20
Nov-21
Nov-19
Nov-22
Nov-23
Nov-24

25
32

33
31
34
36
35
38
39
37
40
42
41
46
47
48
45
44
43
51
50
49
52
53
54

smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice

downstream 1.25
upstream
upstream

downstream
upstream
downstream
downstream
upstream
upstream
downstream
upstream
downstream
downstream
upstream
upstream
downstream
upstream
downstream
downstream
upstream
upstream
downstream
upstream
downstream
downstream

1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25
1.25
2.25
1.25

1
1
2
1
1
2
1
1
2
1
1
2
1
1
2
1
1
2
1
1

2
1
1
2
1

0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50

0.31
0.31
0.31
0.31
0.31
0.31
0.31
0.36
0.36
0.36
0.36
0.36
0.36
0.36
0.36
0.36
0.36
0.50 0.36
0.50 0.36
0.50 0.36
0.50 0.36
0.50 0.36
0.50 0.36
0.50 0.36
0.50 0.36

22
11
11
11
22
22
22
11
11
11
22
22
22
11
11
11
22
22
22
11
11
11
22
22
22

207

Table 24. Measured experimental data for 54 flume runs. Note: for ADV Q data ‘nan’ indicates there was an
error in the data measurement.

Date run
finished
Sep-23
Sep-26
Sep-22
Sep-27
Sep-28
Sep-29
Oct-03
Oct-04
Oct-05
Oct-02
Oct-Ol
Sep-30
Oct-07
Oct-08
Oct-06
Oct-09
Oct-10
Oct-11
Oct-19
Oct-20
Oct-18
Oct-21
Oct-23
Oct-22
Oct-27
Oct-28
Oct-29
Oct-26
Oct-25
Oct-24
Oct-31
Nov-01
Oct-30
Nov-02

Run
#

1
2
1
4
5
6
10
11
12
9
8

7
14
15
13
16
17
18
20
21
19
22
24
23
28
29
30
27
26
25
32
33
31
34

Cover

open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice

Temp

Approach

Celsius

cm/s

Water
level
cm

11.145
10.33
11.313
10.792
10.343
10.647
10.388
10.267
10.435
10.441
10.361
10.553
10.443
10.358
10.507
10.412
10.437
10.418
10.22
10.351
10.258
10.143
10.341
10.152

23.2
23.7
24
23.2
22.5
26.9
23.3
19.55
22.6
23.3
25.5
21.8
25
26
18.3
25
21.6
18.2
23.5
27.11
19.4
23.5
22.8
22.6

9.5
21
13
9.7
24.4
13.3
9
21.6
13.1
9
25.6
13
9.8
20.5
13.5
10
25.5
13.4
9.8
21.4
13
9.8
24.3
13.4

10.121

23.3
23.3

9.7
21.5
12.9
9.6
25.8
13
12.8
22
13
11.2

10.319
10.129
10.1
10.391
10.124
9.979
10.306

10.004

velocity

20.07
23.3
24.6
22.9
25.08
28.02
21.6
25.08

ADV
Q

Q Discharge
(calculated)

m3/s

m3/s

0.05
0.10
0.06
0.04
0.11
0.06
0.04
nan
0.06
0.01
0.13

0.06
nan
0.09
0.05
nan
0.10
0.05
nan
nan
0.05
nan
0.11
0.05
0.03

nan
0.05
0.05
0.13
0.06
nan
nan
nan
nan

0.04
0.10
0.06
0.05
0.11
0.07
0.04
0.08
0.06
0.04
0.13
0.06
0.05
0.11
0.05
0.05
0.11
0.05
0.05
0.12
0.05
0.05
0.11
0.06
0.05
0.10
0.05
0.04
0.13
0.06
0.06
0.12
0.06
0.06

208

Nov-04
Nov-03
Nov-08
Nov-09
Nov-07
Nov-10
Nov-12
Nov-11
Nov-15
Nov-16
Nov-18
Nov-17
Nov-14
Nov-13
Nov-20
Nov-21
Nov-19
Nov-22
Nov-23
Nov-24

36
35
38
39
37
40
42

41
46

47
48
45
44

43
51
50
49
52
53
54

rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice

10.386
10.023
9.839
10.161
9.932
9.723
10.198
9.699
9.668
10.093
9.669
9.64
10.19
9.773
8.89
9.76
9.619
8.62
9.973
9.525

21
21.6

23.2
24.14
20.5
23.2
22.7
20.2
22
23.3
22.6
22
22.5
21.2
22.91
26
22.2
22.91
21.5
22.5

23.1
13.4

10
21.1
13.2
9.9
24.6
13.1
9.8
21.6
13.1
9.8
25.9
13.2
13
22
13
13
25.9
13

0.09
0.05
nan
nan
0.05
nan
0.11
0.05
0.02
nan
0.05
nan
0.12
0.06
nan

nan
0.06
nan
0.12
0.06

0.10
0.06
0.05
0.10

0.05
0.05
0.11
0.05
0.04
0.10
0.06
0.04
0.12
0.06
0.06
0.11
0.06
0.06
0.11
0.06

209

Table 25. Measured maximum scour depth and various calculated parameters associated with 54
experimental flume runs.

Date run
finished

dmai/
Run
#

Cover

max
scour
depth

R,

dmax/h

radius

cm
Sep-23
Sep-26
Sep-22
Sep-27
Sep-28
Sep-29

0ct-03
0ct-04
0ct-05
0ct-02
Oct-Ol
Sep-30
0ct-07
Oct-08
Oct-06
0ct-09
Oct-10
Oct-11
Oct-19
Oct-20
Oct-18
Oct-21
Oct-23
Oct-22
Oct-27
Oct-28
Oct-29
Oct-26
Oct-25
Oct-24
Oct-31
Nov-01
0ct-30

1
2
1
4
5
6
10
11
12
9
8
7
14
15
13
16
17
18
20
21
19
22
24
23
28
29
30
27
26
25
32

33
31

open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice

5
2.5
2
5.5
1
3
4.5
4.5
1.5
5.5
2
1.5
5
4
1
6
2
1.5
8
6
4
11
4
5
8.5
9
5
12
4.5
5
6
7.5
7

hydraulic

0.53
0.12
0.15
0.57
0.04

0.23
0.50
0.21
0.11
0.61
0.08
0.12
0.51
0.20
0.07
0.60
0.08
0.11
0.82
0.28
0.31
1.12
0.16
0.37
0.88
0.42
0.39
1.25
0.17
0.38
0.47
0.34
0.54

m
0.09
0.17
0.12
0.09
0.20
0.12
0.04
0.10
0.06
0.04
0.11
0.06
0.05
0.09
0.06
0.05
0.11
0.06
0.09
0.18
0.12
0.09
0.20
0.12
0.05
0.10
0.06
0.05
0.11
0.06
0.06
0.10
0.06

Fr,
Froude
number

Fo,
densimetric
froude
number

u.,
shear
velocity

m/s
0.24
0.17
0.21
0.24
0.15

0.24
0.25
0.13
0.20
0.25
0.16
0.19
0.25
0.18
0.16
0.25
0.14
0.16
0.24
0.19
0.17
0.24
0.15
0.20
0.24
0.16
0.18
0.24
0.15
0.20
0.22
0.19
0.19

3.21
3.28
3.32
3.21
3.11
3.72
3.22
2.70
3.12
3.22
3.53
3.01
3.46
3.59
2.53
3.46
2.99
2.52
3.61
4.16
2.98
3.61
3.50
3.47
3.58
3.58
3.08
3.58
3.78
3.52
3.85
4.30
3.32

0.13
0.18
0.15
0.13
0.20
0.15
0.09
0.14
0.11
0.09
0.15
0.11
0.10
0.14
0.11
0.10
0.15
0.11

0.13
0.19
0.15
0.13
0.20
0.15
0.10
0.14
0.11
0.09
0.15
0.11
0.11
0.14
0.11

210

Nov-02
Nov-04
Nov-03
Nov-08
Nov-09
Nov-07
Nov-10
Nov-12
Nov-11
Nov-15
Nov-16
Nov-18
Nov-17
Nov-14
Nov-13
Nov-20
Nov-21
Nov-19
Nov-22
Nov-23
Nov-24

34
36
35
38
39
37
40
42
41
46
47
48
45
44
43
51
50
49
52
53
54

rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice

9
8
9
4.5
4.5
2
6
3
2
6
6
2.25
6
2
3
6
5
3
7
3
5

0.80
0.35
0.67
0.45
0.21
0.15
0.61
0.12
0.15
0.61
0.28
0.17
0.61
0.08
0.23
0.46
0.23
0.23
0.54
0.12
0.38

0.05
0.10
0.06
0.09
0.17
0.12
0.09
0.20
0.12
0.05
0.10
0.06
0.05
0.11
0.06
0.06
0.10
0.06
0.06
0.11
0.06

0.24
0.14
0.19
0.23
0.17
0.18

0.24
0.15
0.18
0.22
0.16
0.20
0.22
0.14
0.19
0.20
0.18
0.20
0.20
0.13
0.20

3.85
3.23
3.32
3.45
3.59
3.05
3.45
3.38
3.01
3.28
3.47
3.37
3.28
3.35
3.16
3.41
3.87
3.31
3.41
3.20
3.35

0.10
0.14
0.11
0.13
0.18
0.15
0.13
0.20
0.15
0.10
0.14
0.11
0.10
0.15
0.11
0.11
0.14
0.11
0.11
0.15
0.11

211

Table 26. Calculated hydraulic parameters for 54 flume experiments.

U.c
Date run
finished

Run#

Cover

Critical
shear
velocity

Re*
Shear
Reynolds
number

TDimensionless

Manning's n

shear stress,

m/s
Sep-23
Sep-26
Sep-22
Sep-27
Sep-28
Sep-29
Oct-03
Oct-04
Oct-05
Oct-02
Oct-Ol
Sep-30
Oct-07
Oct-08
Oct-06
Oct-09
Oct-lO
Oct-11
Oct-19
Oct-20
Oct-18
Oct-21
Oct-23
Oct-22
Oct-27
Oct-28
Oct-29
Oct-26
Oct-25
Oct-24
Oct-31
Nov-01
Oct-30

1
2
1
4
5
6
10
11
12
9
8
7
14
15
13
16
17
18
20
21
19
22
24
23
28
29
30
27
26
25
32
33

31

open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice

0.012
0.012
0.012
0.012

0.011
0.014
0.014
0.010
0.013
0.014
0.013
0.012
0.015
0.014
0.010
0.014
0.011
0.010
0.012
0.013
0.010
0.012
0.011
0.012
0.014
0.012
0.011
0.014
0.013
0.013
0.014
0.015
0.012

5.49
5.13
5.47
5.48
4.80
6.12
6.08
4.56
5.61
6.08
5.83
5.41
6.45
6.10
4.52
6.43
4.94
4.50
4.49
4.75
3.59
4.49
3.95
4.16
4.88
4.40
4.04
4.88
4.55
4.61
5.06
5.28
4.35

0.029
0.026
0.029
0.029
0.022
0.036
0.036
0.020
0.030
0.036
0.033
0.028
0.040
0.036
0.020
0.040
0.024
0.020
0.037
0.041
0.023
0.037
0.028
0.032
0.043
0.035
0.030
0.043
0.038
0.039
0.047
0.051
0.034

0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.17
0.16
0.16
0.16
0.16
0.16
0.16
0.16

0.16
0.16
0.16
0.16
0.16
0.16
0.16
0.16

212

Nov-02
Nov-04
Nov-03
Nov-08
Nov-09
Nov-07
Nov-10
Nov-12
Nov-11
Nov-15
Nov-16
Nov-18
Nov-17

34

Nov-14
Nov-13
Nov-20
Nov-21
Nov-19
Nov-22
Nov-23
Nov-24

44

36
35

38
39
37
40
42
41
46
47
48
45
43
51
50
49
52
53
54

rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice

0.014
0.011
0.012
0.012
0.012
0.011
0.012
0.011
0.010
0.013
0.012
0.013
0.013
0.012
0.012
0.013
0.014
0.012
0.013
0.011
0.013

5.15

0.048

3.93
4.33
4.70
4.51
4.02
4.71
4.17
3.97
4.89

0.028
0.034
0.033
0.031
0.024

4.68

4.83
4.89
4.43
4.53
4.90
5.21
4.75
4.90
4.23
4.82

0.034
0.026
0.024
0.036
0.033
0.035
0.036
0.030
0.031
0.036
0.041
0.034
0.036
0.027
0.035

0.16
0.16
0.16
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24

0.24
0.24
0.24
0.24
0.24

213

APPENDIX B - SCOUR HOLE DATA
Table 27. Scour depth values associated with Figure 29, and calculated percentages for open channel flow.

D50 = 0.47 mm

Open run #

1
2
3
4

5
6
average scour depth
% higher than Dso=O.58 mm

D50 = 0.58 mm

D50 = 0.50 mm

8
6
4
11
4
5
6.33
50.00

scour depth (cm)
4.5
4.5
2
6
3
2
3.66
13.64

5
2.5
2
5.5
1
3
3.16

Table 28. Scour depth values associated with Figure 30, and calculated percentages for smooth ice cover
conditions.

Dso = 0.47 mm

Smooth run #

1
2
3
4
5
6

D50 = 0.50 mm

8.5
9
5

12
4.5

5
7.33

average scour depth
% higher than Dso=O.58 mm

55.68

D50 = 0.58 mm

scour depth (cm)
6
6
2.25
6
2
3
4.20
22.77

4.5
4.5
1.5
5.5
2
1.5
3.25

Table 29. Scour depth values associated with Figure 31, and calculated percentages for rough ice cover
conditions.

D50 = 0.47 mm

Rough run #

1
2
3
4
5
6
average scour depth
% higher than Dso=O.58 mm

6
7.5
7
9
8
9
7.75

58.06

D50 - 0.50 mm
scour depth (cm)
6
5
3
7
3
5
4.83
32.75

D50 = 0.58 mm
5
4
1
6
2
1.5
3.25

214

APPENDIX C - ARMOUR LAYER DATA

The data in this appendix represents the measured armour layer size for all pier scour holes for
all 54 flume experiments.
Table 30. Median size of armour layer in scour hole along with D50 of channel bed.

Date run Run
#
finished
Sep-23
Sep-26
Sep-22
Sep-27
Sep-28
Sep-29

0ct-03
0ct-04
0ct-05
Oct-02
Oct-Ol
Sep-30
Oct-07
0ct-08
0ct-06
Oct-09
Oct-10
Oct-11
Oct- 19

0ct-20
Oct-18
Oct-21
Oct-23
Oct-22
Oct-27
Oct-28
Oct-29
Oct-26
Oct-25
Oct-24

1
2
1
4
5
6
10
11
12
9
8
7
14
15
13
16
17
18
20
21
19
22
24
23
28
29
30
27
26
25

D50
Cover

open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice

channel
bed
mm
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47
0.47

D50 armour
layer

mm
0.6
0.8
1.05
0.62
0.65
0.78
1.25
0.87
0.85
1.81
0.83
0.94
1.25
1.8
1.42
1.6
0.85
1.57
0.48
0.56
0.5
0.59
0.43
0.45
0.4
0.5
0.49
0.55
0.3
0.53

Geometric
standard
deviation
dimensionless
3.06
2.87
2.98
3.12
2.81
3.04
3.00
3.06
2.99
2.98
2.98
3.11
3.22
3.14
2.79
2.67
2.95
2.54
2.93
2.91
1.74
3.78
2.16
2.17
1.84
2.46
1.97
3.23
2.91
2.48

215

Oct-31
Nov-01
0ct-30
Nov-02
Nov-04
Nov-03
Nov-08
Nov-09
Nov-07
Nov-10
Nov-12
Nov-11
Nov-15
Nov-16
Nov-18
Nov-17
Nov-14
Nov-13
Nov-20
Nov-21
Nov-19
Nov-22
Nov-23
Nov-24

32
33
31
34
36
35
38
39
37
40
42
41
46
47
48
45
44
43
51

50
49
52
53
54

rough ice
rough ice
rough ice
rough ice
rough ice
rough ice
open channel
open channel
open channel
open channel
open channel
open channel
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
smooth ice
rough ice
rough ice
rough ice
rough ice
rough ice
rough ice

0.47
0.47
0.47
0.47
0.47
0.47
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50

0.55
0.45
0.45
0.57
0.49
0.48
0.6
1.13
0.8
1.43
0.6
0.68
1.4

1.15
0.59
0.68
0.79
0.65

0.75
1.18
0.73
0.6
0.59
0.73

2.83
2.52
3.01
3.53
1.99
1.84
3.06
3.42
3.21
3.40
3.00
3.26
3.45
3.21
3.21
3.42
3.18
3.29
3.32
2.49
3.47
3.16
2.85
3.17

216

APPENDIX D - REGRESSION ANALYSIS
The data in this appendix presents the various parameter combinations that were used when

investigating the best regression model. The graphs below were not included in the main body
of the thesis for one of the following reasons:
(a) the regression models produced regression coefficients that did not support previous

findings in this study. For example, the regression coefficient for sediment D50 would

be positive, when in fact findings under section 5.1, indicate that sediment D50 actually
decreases with increasing scour depth or
(b) the regression model produced supported findings of this study however a stronger

correlation was found using various other parameters. For example, it was found that

the parameters D50/H and D50/B both typically produced correct regression models,
however for the most part, D50/B would produce a stronger correlation.

1.0 Open channel conditions:

Figure 76. Variation of maximum scour depth for the 11 cm pier under open channel conditions.

217

Figure 77. Variation of maximum scour depth for the 22 cm pier under open channel conditions

2.0 Ice covered conditions:

Figure 78. Variation of maximum scour depth for the 1 1 cm pier under ice covered conditions.

218

Figure 79. Variation of maximum scour depth for the 22 cm pier under open channel conditions

Figure 80. Variation of maximum scour depth under ice covered conditions.

219

3.0 Armour layer open channel:

X=
"

/ U V”
\4gHJ v B )

,b^n \/D50Armour\~015
/Ds^-401 \b)
)
H

Figure 81 . Variation of maximum scour depth under open channel conditions.

3.1 Armour layer ice cover, 1 1 cm pier:

Figure 82. Variation of maximum scour depth for the 11 cm pier under ice covered conditions. Note the
regression coefficient for ice cover roughness not reflect results from this study.

220

11 cm pier, smooth + rough ice
1.00

Figure 83. Variation of maximum scour depth for the 1 1 cm pier under ice covered conditions. Note
the regression coefficient for D50 armour does not reflect results from this study.

Figure 84. Variation of maximum scour depth for the 11 cm pier under ice covered conditions.

221

3.2 Armour layer ice cover, 22 cm pier:

Figure 85. Variation of maximum scour depth for the 22 cm pier under ice covered conditions. Note the D50/H
regression coefficient does not reflect results from this study.

Figure 86. Variation of maximum scour depth for the 22 cm pier under ice covered conditions. Note the D50
armour regression coefficient does not reflect results from this study.

222

Figure 87. Variation of maximum scour depth for the 22 cm pier under ice covered conditions. Note the D5q/H
regression coefficient does not reflect results from this study.

3.3 Armour layer all experiments:
1.4

dm„/H = 25.575 X- 0.003
R2 = 0.6394

1.2

1
X

08

| 0.6
0.4

°

0.2

-<59'

0
0

0.005

o

—
0.01

°
0.015

0.02

0.025

0.03

0.035

0.04

Figure 88. Variation of maximum scour depth under ice covered conditions. Note the D50/H and b/B regression
coefficients do not reflect results from this study.

223

1.4

d^/H* 8E-17X- 0.0193

1.2

R1 = 0.7777

1 -

0.8 0.6 -

0.2 -

0.4

0 1.5E+16

1E+16

X-

U
V38 B
f\JgH/

M012
\nJ V

' '

Z^\°05
) \b)
24

H

Figure 89. Variation of maximum scour depth under ice covered conditions. Note the Dso armour regression
coefficient does not reflect results of this study.
1.4

dm«/H = 24.298 X- 0.0015

1.2

R2 = 0.634

1

Po

£ 0.8
CO

E

o

0.6
0.4

'

o

o

° °-" o'

,-a'

Q--

o

o

O

o

°no

°"68L--o°o'°
o

0.2

° °

0
0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

54
u
fDS0armourx-° 49 fb\~° 02
(Dsox~° 29 /n,
\3
\Dsochannel) \BJ
H
'
'
Figure 90. Variation of maximum scour depth under ice covered conditions. Note the b/B regression

/

coefficient does not reflect results from this study.

224

1.4

IE-15 X - 0.02
R2 = 0.7708

1.2

o

1 -

o
0.8

o

E

oO

0.6

O °8 .
O-'

0.4

o
o

0.2

o

Q.-''O

o

"o

o
o

o

0

0

2E+14
2 70

/D
x = u I Ik B50

4E+14

nb

6E+14

Dso armour
D50channel

8E+14
0.05
0.21

1E+15

b_

Figure 91 . Variation of maximum scour depth under ice covered conditions. Note the D$o armour regression
coefficient does not reflect results from this study.

225