CIVIL ENGINEERING STUDIES

Hydraulic Engineering Series No. 56

Navigation - Induced Bed Shear Stresses:

Laboratory Measurements, Data Analysis, and Field Application

Marcelo H. Garcia
David M. Admiraal
Jose Rodriguez
Fabian Lopez

Sponsored by:

U.S. Army Corps of Engineers

Waterways Experiment Station
(Contract DACW39-96- K-005)

DEPARTMENT OF CWIL ENGINEERING

UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN
URBANA, ILLINOIS

April 1998
 Table of Contents

1. Introduction ............................................................................. 1

1.1. Motivation .................................................................... 1

1.2. Purpose of Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2. Shear Stress Sensors .................................................................. 3

2.1. Operation ..................................................................... 3

2.2. Calibration ..................................................................... 3

.
3 Shear Stress Footprints of the Barge Tow Model ............................... 7

3.1. Introduction .................................................................. 7

3.2. Experimental conditions .................................................. 7

3 3
3 .3 . Ensemble averaged shear stress distributions ........................... 10

3.4. Dimensional analysis ....................................................... 20

3.5. Bow dimensional analysis ................................................. 21

3.6. Stern dimensional analysis ................................................ 26

3.7. Results from 1996 experiments ......................................... 30

3.8. Comparison of measured velocity profiles and shear stresses ........ 33

3 . 9. Conclusions .................................................................. 34

4. Probability Density Functions ......................................................

4.1. Introduction ..................................................................

4.2. Ensemble statistics ..........................................................

4.3 . WES experiments .........................................................

4.3.1. Description of the experiments at WES ......................

4.3.2. Results ............................................................

4.4. UrCTC experiments .........................................................

4.4.1. Introduction ......................................................

4.4.2. Facility at HSL ...................................................

 4.4.3. Description of the experiments ................................ 54

4.4.4. Results ............................................................ 54

4.5. Conclusions .................................................................. 55

5 . Field Application ..................................................................... 62

5.1. Introduction .................................................................. 62

5.2. Sediment entrainment in unsteady flow .................................. 62

5.2.1. Background ....................................................... 62

5.2.2. The stochastic approach ........................................ 63

5.2.3. Computation of sediment entrainment into suspension ... 65

5.2.4. Practical considerations ......................................... 69

5.3. Application 1: computation of sediment entrainment induced by

navigation traffic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

5.4. Application 2: computation of the scour produced by a tow barge

passage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

6. References .........................,
,,..........................................
, 85

Appendix A . Stern shear stress measurements .................................... 87

 List of Figures

Figure 2.1 Shear stress calibration duct 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 2.2 Shear stress calibration duct 2 .........................................

Figure 2.3 Calibration of Sensor 1. second data set.

operating resistance 4.97 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.1 Schematic of shear stress measurement locations . Shear

stress was measured along lines 1 - 6 for all of the normal tests . Shear
stress was also measured along lines 7 and 8 for the Upbound D test .
Shear stress was only measured along line 3 for the PDF test . Positions
1 through 8 correspond to -1.3, 5.1, 11.4, 17.8, 24.1, 30.5, 43.2, and
55.9 cm from the barge centerline, respectively . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.2 Isometric drawing showing where shear stress has been
Measured . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.3 Measured shear stress distribution. downbound run A .............

Figure 3.4 Measured shear stress distribution. downbound run B ............

Figure 3.5 Measured shear stress distribution. downbound run C ...........

Figure 3.6 Measured shear stress distribution. downbound run D .............

Figure 3.7 Measured shear stress distribution. upbound run A .................

Figure 3.8 Measured shear stress distribution. upbound run B ................

Figure 3.9 Measured shear stress distribution. upbound run C ................

Figure 3.10 Measured shear stress distribution. upbound run D ...............

Figure 3.11 Variables used in dimensional analysis of barge shear

stress footprints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.12 Variables used to non-dimensionalize the bow peak ...............

Figure 3.13 Dimensionless bow shear stress footprint for downbound

Tows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.14 Dimensionless bow shear stress footprint for upbound tows ......
Figure 3 15 Downbound run C dimensionless stern shear stress footprint ..... 27

Figure 3.16 Downbound run D dimensionless stern shear stress footprint .....

Figure 3.17 Upbound run C dimensionless stern shear stress footprint ........

Figure 3.18 Comparison of measured velocity and log-law velocity profile ...

Figure 4.1 Estimates of <z>. z', :and as a hnction of the number of

samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 4.2 Estimates of S and F as a function of the number of samples. . . . . . . . .

Figure 4.3 Ensemble averages of sensors 1. 2 and 3 in the front part . . . . . . . . . . . . . .

Figure 4.4 Ensemble averages of sensors 1. 2 and 3 in the propeller part .........

Figure 4.5 Variation of r' with time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 4.6 Vzi.tim of S with time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 4.7 Variation of F with time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 4.8 variation of z' with time (front part) ......................................

Figure 4.9 Variation of S with time (front part) ......................................

Figure 4.10 Variation of F with time (front part) .....................................

Figure 4.11 Variation of z' with time (propeller part) ...............................

Figure 4.12 Variation of S with time (propeller part) ................................

Figure 4.13 Variation of F with time (propeller part) ................................

Figure 4.14 Probability density functions at characteristic points . . . . . . . . . . . . . . . . . . .

r lgure 4.15 Resuspension facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

l7:

Figure 4.16 Experimental setup for the HSL experiments . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 4.17 Estimates of <z>..'T S and F corresponding to test # 1 of the HSL

experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 4.18 Estimates of <z>. z'. S and F corresponding to test #2 of the HSL
experlment s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Figure 4.19 Estimates of <T>. T'. S and F corresponding to test #3 of the HSL

experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

Figure 4.20 Estimates of <T>. T ~ S. and F corresponding to test #4 of the HSL

experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

Figure 4.2 1 Estimates of <T>. T+. S and F corresponding to test #5 of the HSL

experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

Figure 5.1 Methodology used to predict entrainment ............................ 66

Figure 5.2 11. 12. 13. L7and 1 5 values as a function of ri . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Figure 5.3 Entrainment relations for steady flow ..................................... 70

Figure 5.4 I1.J17and J2 values as a fbnction of s' .................................. 71

Figure 5.5 Power law fit to I1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

Figure 5.6 Entrainment computations .................................................. 75

Figure 5.7 Wall shear stress and entrainment produced by a tow-barge passage 76

Figure 5.8 Scour produced by the propellers for Ds = 0.5 mm ..................... 83

Figure 5.9 Scour produced by the propellers for Ds = 0.1 mm .................... 84

 List of Tables

Table 2.1 Calibration parameters. first data set ..................................

Table 2.2 Calibration parameters. second data set ............................

Table 3.1 Summary of experimental conditions ..............................

Table 3.2 Model parameters. downbound barge ...............................

Table 3.3 Model parameters. upbound barge .................................

Table 3.4 Peak shear stress used to non-dimensionalize stern footprints ....

Table 3.5 Measured parameters, downbound barge. 1996 dataset ...........

Table 3.6 Measured parameters. upbound barge. 1996 dataset ..............

Table 4.1 Definition of ensemble statistics and PDF . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Table 4.2 Experimental conditions for PDF experiments . . . . . . . . . . . . . . . . . . . . . . .

Table 4.3 Characteristics of the shear stress signal for the WES experiment: . 38

Table 4.4 Characteristics of the shear stress signal for the HSL experiments . 54

vii
 1. INTRODUCTION

1.1. Motivation

River morphology is determined by the entrainment and deposition of sediment.

fivers like the Mississippi and Illinois are continuously changing because of erosion and
deposition. In the past, most of the erosion occurred during floods in regions of locally
high flow velocity, and a majority of the deposition occurred in the floodplain where flow
velocity is low. Sediment transport induced by navigation can happen throughout the
year, and deposition of sediment eroded by barge traffic can not occur in the floodplain
when river stages are normal. Instead, the sediment is deposited in other locations of low
velocity, these locations may include fish spawning areas and riparian wetlands.
Bed shear stresses underneath barge tows are locally much higher than bed shear
stresses during floods. Consequently, the riverbed underneath a barge tow may be
scoured more deeply than it would be scoured by a flood. Agricultural and industrial
chemicals that would normally remain in the bed may be entrained by the passage of
navigation traffic, contaminating the river water. In addition, since barge traffic occurs
more regularly than flooding, the water may remain turbid throughout the year.
Continual turbidity may adversely affect a variety of aquatic species, and the life cycles
of some species may require the river water to be clean at certain periods during the year.
Three questions need to be answered in order to predict the damage caused by
navigation induced sediment entrainment. First, what is the amount of sediment
entrained by a barge tow? Second, where is the entrained sediment transported by the
flow field of the barge tow? And finally, where is the suspended sediment deposited?
The information presented in this report is intended to help answer the first question.

1.2. Purpose of study

A number of studies have been conducted to determine the flow field underneath
barge tows, and the role barge tows play in the resuspension and transport of suspended
sediment (Bhowmik et al, 1996; Mazumder et al, 1993; Maynord, 1990). These studies
have primarily focused on velocity distributions and field concentration measurements.
The primary objective of this study is to characterize the bed shear stress pattern
produced by a barge tow for a wide range of test conditions. The bed shear stress
patterns are intended for use with sediment entrainment models for estimating the overall
suspended sediment load produced by a passing barge tow, and also for predicting the
riverbed scour caused by the tow.
The shear stress footprints were gathered for a 1:25 scale model of an actual barge
tow in the Navigation Effects Flume at Waterways Experiment Station in Vicksburg,
Mississippi. Shear stresses were gathered using hot-film shear stress sensors. Operation
and calibration of these sensors is discussed in Chapter 2 Results of the experiments are
presented in Chapter 3. Several realizations were performed for each test condition, and
ensemble averages of these realizations are presented. In addition, an attempt is made to
non-dimensionalize the shear stress footprints, both spatially and temporally. In Chapter
4, statistical characteristics of the shear stress distributions are analyzed. Finally, in
Chapter 5 examples of field application of the shear stress distributions are given.
 2.- SHEAR STRESS SENSORS

2.1. Shear stress sensor operation

Water flowing over a hot film shear stress sensor convects heat from the sensor to
the thermal boundary layer above the sensor. The heat convected to the thermal
boundary layer is related to the momentum transferred to the momentum boundary layer
(Hanratty and Campbell, 1996; Menendez and Ramaprian, 198 5; B ellhouse and Schultz,
1966). Thus, the wall shear stress can be calculated from measurements of heat transfer
fi-om the sensor. The amount of heat convected is equal to the electrical power provided
to the sensor; this power is easily measured using a bridge circuit. Eq. 2.1 is commonly
used for converting the voltage measured across the bridge to shear stress:

Where z is the wall shear stress, Ebis the bridge voltage, AT, is the difference in
temperature between the sensor and the fiee stream, and A, B, m, and AT, are constants.
AT, has been added to Eq. 2.1 for convenience and has no effect on the results as long as
it is held constant. For our experiments AT, was taken to be the operating temperature
minus 20" C. The dependence of B on AT, has been ignored in Eq. 2.1. Often, m is set
equal to 3, but according to Bruun (1995) m increases with an increase in thermal
conductivity between the fluid and the substrate on which the sensor is mounted.

2.2. Shear stress sensor calibration

Prior to use in the experimental facility at WES, the hot-film shear stress sensors had
to be calibrated. The constants A, B, and m needed to be determined so that Eq. 2.1
could be used to calculzte the shear stress from the gutput of the shear stress sensors. In
szfu calibration would have been preferable, but this was not possible and the sensors had
to be calibrated in another facility. In order to minimize changes in the calibration
curves, care was taken when moving the sensors from the calibration facility to the
experimental facility. Two separate ducts were used to calibrate the sensors.
The first duct, shown in Figure 2.1, was designed and constructed for the calibration
of shear stress sensors. The design consists of a rectangular duct with pressure taps
located along its length and a constant head at both its inlet and outlet. The aspect ratio
of the duct is approximately 30:1 so it is assumed that the flow in the duct is two-
dimensional.

Flow straightener Shear stress sensor

Figure 2.1 Shear stress calibration duct I

The pressure gradient in the duct is directly related to the wall shear stress as shown
by Eq. 2.2.
a
T=-R,-(~+~h) (2.2)
ax
In which, & is the hydraulic radius of the duct, and P + yh is the potential head of the
fluid. Shear stress calibrations were also done in a second duct, shown in Figure 2.2.
The second duct is similar to the first duct, but has a height of 10 cm (3.94 in.) and a
width of 30 cm (1 1.81 in.). This aspect ratio of 3 : 1 is not enough to guarantee two
dimensional flow, but the flow within the duct is close enough to two dimensional to
provide reliable measurements of the shear stress. The shear stress is measured the same
way as it was measured in the first duct. The second duct has a larger hydraulic radius
than the first duct. Thus, according to Eq. 2.2, the pressure gradient will be lower in the
second duct than in the first duct for the same shear stress. In addition, the Reynolds
number of the duct will be higher for the same shear stress.
 Figure 2.2 Shear stress calibration duct 2

Even with special mounting plates for the sensors there was some concern that
moving the sensors from the calibration ducts to the experimental facility would change
the calibration of the sensors. In order to address this concern, some sensor calibrations
were performed in both the first and second ducts and the calibration curves were
compared. The two calibration curves were closely matched (Admiraal, 1997),
increasing confidence in the calibration procedure.
Two sets of data were gathered at WES. The set gathered on the first trip was
incomplete because of technical problems, so a second set was also gathered. Calibration
of the shear stress sensors was performed in duct 1 for the set gathered on the first trip,
and in duct 2 for the set gathered on the second trip. A typical calibration curve is shown
in Figure 2.3. For the second data set a value of 5 was used for m so that the curve was
nearly linear when 21f5was plotted against ~ b ~ .
 Figure 2.3 Calibration of Sensor 1, second data set, operating resistance 4.9 7

A, B, and m are given for the four sensors in Tables 2.1 and 2.2 for the first and
second sets of data. Note that the sensors used are not the same for both data sets.
Replacement and repair was necessary for several of the sensors because of damage to
the sensors during use. In addition, multiple operating resistances were used for the
second dataset since some bubbling problems were encountered.
Table 2.1 Calibration parameters, first h t a set
I Sensor / T S I S e r i a l l Operating I Operating / A / B IMI
No. No. Resist. (51) Temperature ("C)
1 968178 5.49 66.7 0.1 158 -2.3894 3

Table 2.2 Calibrationparameters, second data set

Sensor 1 TSI Serial 1Operating I Operating I A B I M ~
No. No. I
Resist. (a) / Temperature ("C) / 1
 3. SHEAR STRESS FOOTPRINTS OF THE BARGE TOW MODEL

3.1. Introduction
In order to determine the amount of sediment moved by the passage of a barge tow it
is necessary to determine the bed shear stress distributions caused by the tow. In this
section the shear stress footprints of barges are given for a number of different flow
conditions and barge velocities. The footprints were measured for a 1:25 scale model of
a 3 by 5 barge tow in a basin with a hydraulically smooth bed. An attempt is made to
make the characteristics of the footprints dimensionless so that they may be used for
many different conditions.

3.2. Experimental conditions

The data presented in this report were gathered at the USACE Waterways
Experiment Station in Vicksburg, Mississippi. The experiments were conducted in the
Navigation Effects Flume with a 3 by 5 barge model having a draft of 9.75 cm,
corresponding to a prototype with a draft of 2.43 m (8 ft). Two sets of data are presented;
the second, more complete set of data is presented first, then observations are made using
the first, incomplete set of data. The second dataset includes four upbound tests, four
downbound tests, and one additional upbound test with a large number of runs (for PDF
analysis). A summary of the experimental conditions is given in Table 3.1.
Table 3.1 Summav of experimental conditions
Hp (m) H (m) U ( d s ) V ( d s ) Flume Pumps
4.3 (14 R) 0.17 0.610 0.11 Pumps3&4
3.4 (1 1 ft) 0.13 0.450 0.09 Pump9
7.0 (23 ft) 0.28 0.715 0.12 Pumps 3-6
DN D / Downbound 5.6 (18.5 ft) 0.23 0.715 0.11 Pumps5-6,3-441%
UPA / Upbound 4.3 (14 R) 0.17 0.305 0.11 Pumps3 & 4
3.4(llR) 0.13 0.305 0.09 Pump9
7.0 (23 R) 0.28 0.420 0.12 Pumps 3-6
UP D I Upbound 5.6 (18.5 R) 0.23 0.420 0.11 Pumps 5-6, 3-4 41%
PDF I Upbound 4.3 (14 R) 0.17 0.305 0.11 Pumps3&4
In which, Hp is the water depth of the prototype river, H is the water depth at the center of

the model, U is the velocity of the model barge tow, and V is the velocity of the model

river (measured with an ADV).

For each of the runs in the eight normal tests, the spanwise distribution of shear

stresses underneath the barge tow was measured with three shear stress sensors (The

fourth, outermost shear stress sensor did not operate properly). Positioning of the sensors

is shown in Figures 3.1 and 3.2. The shear stress probes used to gather the data are one-

dimensional and only give the shear stress in the streamwise direction.

Figure 3.1 Schematic of shear stress measurement locations. Shear stress was
measured along lines 1 - 6for all ofthe normal tests. Shear stress was also measured
along lines 7 and 8for the Upbound D test. Shear stress was only measured along line 3
for the PDF test. Positions 1 through 8 correspond to -1.3, 5.1, 11.4, 17.8, 24.1, 30.5,
43.2, and 55.9 cm @om the barge centerline, respectively.
 Figure 3.2 Isometric drawing showing where shear stress has been measured

Shear stresses were measured on one side of the barge tow centerline only. This was

done under the assumption that the shear stress distribution can be mirrored over the tow

centerline. Since two barge tow alignments were used for each of the eight tests, there

are six locations for which the shear stress has been measured over the time of the barge

passage. The primary locations where shear stress has been measured are -1.3, 5.1, 11.4,

17.8, 24.1, and 30.5 cm from the centerline of the tow. The sensor reading taken at 114

cm from the centerline is located directly under the axis of one of the propellers. The last

upbound test contains one additional set of shear stresses at distances of 3 0.5, 43.2, and

55.9 cm from the tow centerline.

During the eight normal tests, the sensors are always located 10.2, 22.9, and 35.6 cm

from the thalweg of the river model; only the boat alignment is changed to get the

spanwise shear stress distribution described above. The test used to compute the

probability density function (PDF) has all four sensors aligned in the streamwise

direction. The PDF test data was gathered with the axis of the port propeller passing over
all four sensors. This was done in order to provide insight into the nature of the shear

stress fluctuations about the mean.

In addition to the shear stress measurements the water velocity above Sensor 1 was

measured at the three largest depths. Velocities were not measured for the smallest depth

because of limited model boat clearance. The velocity was measured in the streamwise

and spanwise directions with a two-dimensional ADV.

3.3. Ensemble averaged shear stress distributions

Six or more runs were performed for each set of sensor positions. Since most of the

tests had two sets of sensor positions this required twelve or more runs for each set of test

conditions. The result was a minimum of six shear stress distributions for each of the six

sensor locations and for each of the test conditions. One of the ways to extract the mean

shear stress distribution from the test runs is to ensemble average the runs. This was

done for all of the sensor locations and all of the test conditions. Figures 3.3 through 3.10

show the ensemble averaged shear stress distributions for each of the sensor locations in

all of the normal tests.

Figures 3.3 through 3.10 demonstrate several important footprint characteristics.

First, the shear stress distribution caused by the bow of the barge tow is nearly constant

across the entire bow. There is variation in the shear stress footprints observed across the

front of the bow, but this is almost entirely caused by differences in sensor calibration.

This is quite apparent since the shear stress footprint of the bow measured at different

spanwise locations does not change as long as the same sensor is used for all the

measurements. Test W D has one spanwise location where the shear stress was
 I 5

Ti- PI 0

I I , I I
Ti- P4 0
measured with two different sensors. There was a significant difference between the bow

shear stress measured with the two different sensors that can be entirely attributed to

calibration differences between the two sensors. Each of the two sensors used for the

redundant measurement is used to measure shear stress at two other locations for the

same test conditions. The shear stresses measured by the sensors are independent of

location.

The footprints caused by the stern are more complicated than those caused by the

bow. The stern footprints for the two deepest downbound tests and the deepest upbound

test do not vary significantly in the spanwise direction. Tests with shallower depths do

vary significantly in the spanwise direction; For these tests the shear stress is highest

immediately behind the propellers and tapers off to the outside edge of the barge. The

width of the footprint following the stern does not appear to be very large. Apparently,

there are two effects causing the bed shear stress at the stern. The first is the shape of the

stern, and the second is the propellers. For the largest depth the propeller jets attach to

the water surface and do not significantly affect the bed shear stress.

A second thing that is shown in Figures 3.3 through 3.10 is that the bow shear stress

distribution is quite smooth. There is remarkably little variation between the shear

stresses measured at the bow for different test runs, and the ensemble average standard

deviation of the bow wave is quite small for most of the runs. On the other hand, the

region after the passage of the stem is extremely turbulent. In order to get a smooth

ensemble average for the portion of the shear stress footprint that occurs after passage of

the stern a much larger number of test runs would have been necessary.
3.4. Dimensional analysis

In this section, the ensemble averaged test results are used for a dimensional

analysis. Figure 3.1 1 shows the variables that should be considered for dimensional

analysis of the shear stress footprints. Not all of the variables are relevant for the current

set of conditions. Herein, the important variables will be determined, and an attempt will

be made to non-dimensionalize the shear stress footprints. The variables that are

important for dimensional analysis of the bow wave are not the same as the ones that are

important for dimensional analysis of the stern shear stress. Consequently, these two

parts of the shear stress footprint must be analyzed separately.

4 L b

Figure 3.11 Variables used in dimensional analysis of barge shear stressfootprints

The variables important for dimensional analysis are:

g gravitational acceleration

h draft of barge tow

H water depth

L length of tow

t time

u boat velocity

v water velocity

w boat width

X axial location (distance from bow)

Y transverse location (distance fiom centerline)

v
water viscosity

P water density

Z bed shear stress

R% propeller Reynolds number

geometry geometry characteristics of barge model.

 For both the bow and the stern dimensional analysis, differences between the model

and prototype geometry are ignored.

3.5. Bow dimensional analysis

In order to non-dimensionalize the bow shear stress distribution we note that:

= f(x, Y, h, H, W, L, U, V, P, v, R%, t, g) (3.1)

It has been observed that the shear stress distribution caused by the bow of the boat

does not vary across the front of the tow for each of the test conditions. The flow is

nearly two-dimensional across the width of the barge. This makes the width of the barge

and y irrelevant. The propeller Reynolds number is also irrelevant. For the bow, the

distance x can be entirely represented by time (t) and velocity (U), so x can be neglected.

Some of the variables did not change throughout the entire set of experiments. g, v, and L

were constant for all the experiments and are ignored in the analysis. If any of these

variables are changed their effect on the current results should be analyzed. The final

variable dependence is given by:

7 = f(H, h, U, V, t, P) (3.2)

Rearranging the variables leads to the following relation:

Here, p is constant for all the experiments so it can be absorbed into a constant. h, U, V,

and H are constant for all time during an experiment so the relationship between t and H

can be determined by plotting Zplil h for all of the runs, where,

vs. - zPkl is the
(u-v)~ H
maximum shear stress for the bow footprint for each of the runs. zPklis given along with
H, ~ ~ (The
k zmaximum shear stress at the stern), 6t (the time elapsed between the two

maximum shear stresses), and several other variables in Tables 3.2 and 3.3 for
,,-
L~kl
downbound barges and upbound barges, respectively. is given as a hnction of
(u v ) ~
-

in Figure 3.12.

(cm from H Sensor =pk, 1 Zpk2 6t

Test center) (m) ID Number (pa) (pa) 6)
DNA -1.3 0.17 1 0.275 0.68 20.2

DNB -1.3 0.13 1 0.210 1.14 27.7

5.1 1 0.199 3.13 27.4
11.4 3 0.231 5.94 26.6

30.5 2 0.194 0.64 27.3

DNC -1.3 0.28 1 0.097 0.197 19.3
5.1 1 0.074 0.12 22.3

24.1 2 0.106 0.076 18.4

30.5 2 0.084 0.082 17.3
DND -1.3 0.23 1 0.123 0.308 20.9
 Table 3.3 Model ~arametevs,upbound bavae
Sensor
Positions
(cmfrom H Sensor Zp41 Zp42 6t
Test
center) (m) ID Number (pa) (Pa) (s)
U P A
-1.3 0.17 1 0.36 1.21 41.5
5.1 1 0.39 1.18 41.7

UP B

UPC

UP D
 Figure 3.12 Variables used to non-dimensionalize the bow peak

The curve fits shown in Figure 3.12 are given by Eqs. 3 4 and 3.5 for the downbound

and upbound runs, respectively. These two equations can be used to predict the peak

shear stress for the bow footprints of the downbound and upbound runs.

rCpk,l

(u- v)'
Figures 3.13 and 3.14 show the ensemble averaged bow shear stress footprints

calculated for all of the tests. Although the shape of the bow footprint varies somewhat

for the different tests, the dimensionless footprints are very similar. Prior to the arrival of

the peak used to non-dimensionalize the downbound bow shear stress footprint there is a

region that is not so well behaved. The region shown by th :dimensionless time -5.00 to

-1 .OO is quite turbulent and may scale differently than the le: s turbulent region of the

bow footprint. However, analysis of this region is limited sinc :there is a large amount of

variation within this region even for the same test conditions.
Figure 3. 13 Dimensionless bow shear shess footprint for downbound tows

Figure 3.14 Dimensionless bow shear stress footprintfor upbound tows

3.6. Stern dimensional analysis

The same set of variables used for the bow also govern the shear stress distribution

after the stern.

.r = f(x, Y, h, H, W, L, U, V, P, v, Rep, t, g) (3.6)

x, W, h, p, v, L, and g are ignored. R% and y, however, can no longer be neglected.

The new relation is:

= f(y, H, U, V, Rep, t) (3.7)

The stern shear stress footprints are affected by two things, the propeller jets, and the

wake of the stern. For the two deepest downbound tests, and the deepest upbound test,

the propeller jet does not seem to have an effect on the bottom shear stress. In addition,

the shear stress imposed on the bed by the stern wake appears to be independent of y, and

V does not have a significant effect on the shear stress. In these three cases Eq. 3.7 can

be reduced as shown by Eq. 3.8.

x/xpk2 = f(tUH) (3.8)

Figures 3.15 through 3.17 show the dimensionless curves for Downbound runs C

and D and Upbound run C. A new variable is used instead of H in these figures; H,, is

the difference between the depth, H, and the propeller radius. The prototype propeller

diameter is 2.74 m. Care must be taken in the choice of xpk2 used to non-dimensionalize

the shear stress since the tests do not have enough runs to give a smooth ensemble

average. The magnitudes of xpk2 used in Figures 3.15 through 3.17 are given in Table

3.4.
 Table 3.4 Peak shear stress used to non-dimensionalize stern foopl-ints

Test Figure (Pa)

xpk~
Downbound C 3.15 0.104
DownboundD 3.16 0.179
Upbound C 3.17 0.083

For the rest of the stem shear stress footprints, the large number of variables given

by Eq. 3.7 makes dimensional analysis difficult, especially considering the limited

number of test conditions and sensor locations. In addition, the scatter of the shear stress

measurements used to calculate the ensemble averages is much greater after the stern

than at the bow. Consequently, obtaining a good non-dimensional form of the stem wave

is not possible with the limited number of test runs available.

Appendix A gives all of the ensemble averaged stern shear stress footprints gathered

at WES.

3.7. Results from 1996 experiments

Prior to the dataset presented above, another dataset was gathered. A number of

problems were encountered when the dataset was being analyzed, requiring the

acquisition of the second dataset. The first problem encountered was a software problem.

Each dataset was cut off after 80 seconds of data was gathered. Because of this software

glitch most of the runs only showed the arrival of the bow. The second problem occurred

because of bubbling, upon analyzing the data it was noted that not all of the runs

gathered behaved the same for a given ensemble.

The primary cause of this erratic behavior is suspected to be contamination of the

sensors by vapor bubbles. Later it became apparent that some of the calibration curves
used for the first dataset contained data points where bubbling occurred. Eliminating the

runs where bubbling occurred has been attempted, but the shear stresses measured

(particularly in the upbound runs) are still somewhat higher than those gathered in the

current dataset. Tables 3.5 and 3.6 give the peak shear stress measurements for the

downbound and upbound runs, respectively. It should be noted that even in the 1997

dataset a few of the runs read significantly higher than the others, but for the most part

the shear stress readings are well behaved.

Table 3.5 Measured ~auameters.downbound bauae. 1996 dataset

A w '

Test Sensor
LOG. HP H U Z ~ k l Zpk2 6t
(cm) ( 4 (m) ( d s ) S (Pa) (Pa) (Pa) (s) Notes
DNA -68.6 4.3 (14 ft) 0.17 0.61 1 0.21 None None Pumps 3 & 4
-43.2 2 0.39 0.27 0.76 7.3
-17.8 3 0.21 0.13 1.35 0.2
0 1 0.36 0.19 1.86 0.1
7.6 4 0.34 0.15 2.14 0.2
7.6 1 0.27 0.21 1.17 0.5
25.4 2 0.30 0.33 0.87 9.6
33.O 2 0.36 0.29 1.12 8.0
50.8 3 0.32 0.22 0.28 8.0
58.4 3 0.20 0.35 0.21 7.6
76.2 4 None None -
83.8 4 None None None
DN B -25.4 5.6 (18.5 ft) 0.23 0.79 1 0.29 0.14 0.32 5.5 Pumps 3-4 41%
-16.5 1 0.32 0.12 0.41 7.4 Pumps 5-6
0 2 0.39 0.17 0.43 5.6
8.9 2 0.32 0.10 0.35 5.3
25.4 3 0.29 0.12 0.38 5.6
34.3 3 0.15 0.13 0.41 5.9
50.8 4 0.20 0.10 0.33 5.8
59.7 4 0.16 0.11 0.12 6.5
Note: Z ~ is~ given
A since the maximum bow shear stress is not always the shear stress
that occurs on the smooth part of the footprint. The maximum shear stress in the less
turbulent region of the bow footprint is r,kl.
 Table 3.6 Measured parameters, upbound barge, 1996 dataset
%,

Sensor
LOC. *P H U Sensor ~ ~ 1 1 2 6t
Test (cm) ( 4 (m) ( d s ) ID (Pa) (Pa) 6 ) Notes
UP A -67.3 4.3 (14 ft) 0.17 0.31 1 0.67 Pumps 3 & 4
-41.9 2 0.65 0.58 6.9
-25.4 1 0.78
-16.5 3 0.86 2.03 1.4
0 2 0.88
8.9 4 0.69 1.19 1.0
25.4 3 0.74
50.8 4 0.68
VP B -25.4 5.6 (18.5 ft) 0.23 0.42 1 0.70 - Pumps 3-4 41%
-8.9 1 0.74 - Pumps 5-6
0 2 0.81 -
16.5 2 0.71 -
25.4 3 0.67 -
41.9 3 0.58 -
50.8 4 0.75 -
67.3 4 0.24 -
UP C -33.0 7.0 (23 ft) 0.28 0.42 1 0.35 - Pumps 3-6
-7.6 2 0.64 -
17.8 3 0.41 -
43.2 4 0.33 -

The first dataset did show a few things that the second dataset could not. While the

magnitude of the shear stresses measured is not entirely trusted, some of the spatial

aspects of the shear stress distribution are usehl. For the second dataset the sensors were

spaced 12.7 cm apart. This allowed higher resolution, but limited the width of the field

of shear stress that could be measured (due to time constraints). In the first dataset the

sensors were spaced 25.4 cm apart, allowing the measurement of a wider shear stress

field (with less resolution, of course). One of the interesting things that the wider

measurement field showed was that though the bow peak shear stress is fairly constant

across the front of the barge tow, the influence of the peak quickly drops outside the
edges of the barge. This behavior was observed in the first downbound run where no

bow peak was observed outside the 1.27 m wide barge and also in the second upbound

run where the bow peak was greatly reduced just outside the edge of the tow. The

remainder of the experiments did not have sensors located far enough outside of the

barges' edges. A nice thing about this behavior is that it allows the shear stresses

produced by the stern to be modeled as a constant footprint over the entire bow, and

outside of the edges of the tow the influence of the shear stresses can be neglected.

3.8. Comparison of measured velocity profiles and shear stresses

At the time the first dataset was gathered a comparison was made of the shear

velocity computed from three different measurements. The shear velocity was computed

fiom shear stress measurements, a measured velocity profile, and depths measured at

three locations in the model basin. The comparison was made for a very high velocity

flow. Although the flow was non-uniform (since the bed of the river model is

horizontal), the results of the comparison provide an estimate of the accuracy of the

sensors, and some insight into the facility at WES.

The shear stress measurements gave shear velocities of between 0.29 and 0.34 d s ,

the log-law curve fit of the velocity profile gave a shear velocity of 0.38 d s and the

depth measurements gave a shear velocity of 0.32 d s . The three shear velocity

measurements are similar and indicate that the shear stress sensors are operating
1- - - - - - - 1 - - - 1 --- 1 - ' "-
conectiy. -
1I ilt: siiciil -:A.-- -A.
~ e u depth measurements is only approximate
VCIULILY ~ a l ~ u l a livm

since the flow was non-uniform. The shear velocity measured with the shear stress

sensors may be less than the shear velocity computed from the velocity profile because of
the location of the sensors. The sensors are mounted in a large, smooth sheet of

plexiglass. The ADV used to measure the velocity profile was mounted above a sheet of

marine plywood. The marine plywood is probably a little rougher than the plexiglass.

On the other hand, the velocity points gathered with the ADV and the corresponding log-

law velocity profile shown in Figure 3.18 demonstrate that the bed is hydraulically

smooth.

0 10 20 30 40 50 60
Velocity ( c d s )
Figure 3.18 Comparison of measured velocity and log-law velocity profile

3.9. Conclusions

The conclusions that can be drawn from this model study are the following:

1. The shear stress footprint caused by the bow of a barge tow is given by Figure

3.13 for a downbound tow and Figure 3.14 for an upbound tow. The bow

footprint of the downbound tow is similar in shape to that of the upbound tow.

2. The magnitudes of the shear stress footprint at the bow can be found by

calculating the peak bo;v sheax stress using Eqs. 3.4 and 3 .5. The resulting she=

stresses should be multiplied by 25 to get the prototype shear stresses.

3 . The stem shear stress distributions are given by Figures 3.15 through 3.17 and

Table 3.4 for Downbound runs C and D, and Upbound run C. The rest of the
 stern footprints are too complicated to reduce, and shear stresses must be

interpolated from the graphs given in Appendix A.

4. The bow shear stress footprint is constant across the entire bow of the vessel and

rapidly becomes negligible outside the path of the boat.

5. Velocity profiles gathered with an ADV show that the surface of the model basin

is hydraulically smooth, and that shear stress sensors should behave

appropriately in the channel.

 4. PROBABILITY DENSITY FUNCTIONS

4.1. Introduction
Being the wall shear stress a random variable, it is usefui to describe it in terms of
its statistical properties, namely PDF and statistical moments. Although there is still some
discrepancy concerning the general shape of the PDF of the bed shear stress, there is
strong experimental evidence that, for stationary conditions, it approaches a log-normal
distribution (Grass, 1970; Lopez 1994). Typical values of the variation coefficient,
skewness (S) and flatness (F) are about 0.4, 1, and 5 respectively (Alfredsson et al., 1988;
Lopez 1994). This characteristic shape may be affected by unsteadiness, with important
implications for sediment transport; thus, it needs to be investigated.
Two sets of experiments are presented in this section: the first one was obtained at WES
as part of the data gathered in 1997; the second consists of laboratory measurements
carried out at the Hydrosystems Laboratory (HSL), University of Illinois at Urbana-
Champaign.

4.2. Ensemble Statistics

Most of the data measurement and analysis techniques have been developed for
the case in which the data is random and stationary (i.e. the statistical properties do not
vary with time). The extension of these methodologies to unsteady (nonstationary) data is
not straightforward and some assumptions like ergodicity do not apply any more. The
immediate consequence is that the statistics are now time dependent and defined in terms
of ensemble values, unlike the stationary statistics that use time means. This requires the
repetition of the experiment under statistically similar conditions, each one constituting a
realization of the ensemble. Calling N the number of realizations, the time dependent
statistical estimates and PDF are defined for the shear stress series in Table 4.1 :
 Table 4. I Definition of ensemble statistics and PDF
*
Ensemble Statistic 1 PDF Formula
Average I "
< r(t) >= - z,(t) (4.1)
N 1=1
Variance 1 "
2;' (t) = - [r, (t)- a ( t ) >I' (4-2)
N 1=1
Variation Coefficient 5'- (t)
T+ (t) = (4.3)
< r(t) >
N

Skewness S(t) = 1
C[r,(9- < ~ ( t>I3
- I='
)
(4.4)
N z ' k (t)
N

Flatness 1 C[r,(t)- <r(t) >I4

~ ( t=) - I=' (4.5)
N , : IX (t)
Probability Density Function N, (t)
PDF(r,, t) = - (4.6)
NW

In the PDF formula, W is the size of an interval centered at Tk and Nk is the

fraction of the N points that falls in the interval rk?W/2.

4.3. WES Experiments

4.3.1. Description of the Experiments at WES

As already described, the situation selected for the experiments corresponded to

the settings surnmarized in Table 4.2:

 Table 4.2 Experimental conditionsfor PDF experiments

Variable Model Prototype
Flow velocity 0.20 m/s (0.66 fps) 1 d s (3.3 fps)
Boat velocity Upbound0.30m/s(lfps) Upbound1.5m/s(5fps)
Flow depth 0.17 m (0.56 ft) 4.26 m (14 ft)

The shear stress sensor array (sensors equi-spaced 2.54 cm or 10 in) was aligned
in the streamwise direction. The traverse mechanism placed the barge-tow so that the
centerline of one propeller would pass right over the sensors.
As already pointed out, in order to compute unsteady statistics an ensemble of
realizations is required for each experiment, usually resulting in a tedious time-
consuming task. The setup selected for the sensors reduced the required amount of
realizations by a factor of four, since for each passage of the barge tow four independent
shear stress records (one for each sensor) could be gathered. A total of 18 runs were
carried out, resulting in 53 useful realizations after discarding the records in which
problems with the sensors were observed. The bad data included all the records of sensor
4, which did not operate properly.

4.3.2. Results
As mentioned earlier, the tow barge produces two basic effects on the flow
underneath (and hence on the wall shear stress): a wake effect due to the movement of the
barges and a propeller effect. The analysis will focus in regions where these effects can
be observed. The most relevant characteristics of the signal are summarized in Table 4.3.

Table 4.3 Characteristics ofthe shear shess signal for the WES experiments.
Region 1~ Part Acceleration Max. Veloc. 2ndPart Acceleration
Front 0.8 seconds 0.3 m/s2 0.25ds 35seconds -0.01m/s2
Propeller 1 second 0.6 m/s2 0.35 d s 50 seconds - 0.0025 m/s2
Statistical Estimates:
After the data selection, a total of 53 realizations were available for the
computation of ensemble statistics. In order to investigate whether or not this number is
enough to compute reliable estimates, the values of <r>, ,:z' z', S and F at one time are
plotted in Figures 4.1 and 4.2 as a function of the number of samples used in the
computations. It is apparent that as the order of the statistical moment increases, more
samples are required to achieve the same degree of accuracy. While the available number
of samples produces acceptable results for <r> and r', the values of S and F must be
considered with some caution.

number of samples

Figure 4.1 Estimates of < z>, z',tand r'as afunction of the number of samples.
 -5 I I I 1 I I I

0 10 20 30 40 50 60
number of samples

Figure 4.2 Estimates of S and F as afnnctzon of the number of samples.

The three sensors used did not give the exact same value of the shear stress.
Figures 4.3 (front part) and 4.4 (propeller part) show the ensemble average values of the
three sensors, and it can be seen that there is approximately 20% difference between the
maximum and minimum values measured. Considering the fact that the sensors were not
calibrated in situ, the difference was not surprising. This shortcoming was solved to some
extent by multiplying the values of sensors 2 and 3 by the ratio between the ensemble
average of sensor 1 and the ensemble average of the corresponding sensor. This
procedure represents an effective reduction in the number of samples used for the
computation of the ensemble average of the whole series to only the sensor 1 records, i.e.
18 samples.
Figures 4.5,4.6 and 4.7 show how the statistics vary with time. Figures 4.8, 4.9
and 4.10 zoom into the front part and 4.11,4.12 and 4.13 into the propeller part. Due to
the limited amount of samples the statistics looked somehow noisy and a 115 of a second
moving average was performed in order to smooth out the series. The moving average
period was chosen by looking at the original shear stress series and verifying that, during
this lapse of time, no important changes in the statistics occurred.
 sensor I

-. -. - .-sensor 2

. . . . sensor 3

65 70 ' 75 80 85

time (sec)

Figure 4.3 Ensemble averages of sensors 1, 2 and 3 in the fiontpart.

sensor I

- - - -sensor 2

. . . . . . sensor 3

150 200

time (sec)

Figure 4.4 Ensemble averages of sensors 1, 2 and 3 in the propeller part.

Figure 4.5 Variation of r+with time.

time (sec)

Figure 4.6 Variation of S with time.

100 150

time (sec)

Figure 4.7 Variation o f F with time.

 time (sec)

Figure 4.8 Variation of z+with time Cfrontpart).

65 65.5 66 66.5 67 67.5 68

time (sec)

Figure 4.9 Variation of S with time fiontpart).

time (sec)

Figure 4.10 Van'atzon of F with time Cfront part).

Figure 4.11 Variation of rf with time (propellerpartj .

105 107 109 II I 113 115

time (sec)

Figure 4.12 Variation of S with time (propellerpartj.

0 I I I I I / -1.6
105 107 109 111 113 115
time (sec)

Figure 4.13 Variation of F with time (propellerpar9 .

 The figures show that, before and after the barge passes, the statistics tend to
values similar to their predicted steady state values (0.4, 1, and 5 for T', S and F,
respectively). The pattern during the passage of the barge train is complicated, and some
portions of the signal must be considered with caution. For example, the overshoot in the
statistics observed just before the front of the barge and before the propellers is likely due
to a very low value of <T>, situation were the sensors do not work reliably. The most
distinctive feature of the front part is a decrease of the values of T+ to 0.25, and a more
erratic behavior of S and F due to the limited number of samples. However, it can be
observed that S tends to be lower than 1 and F grows in the decelerating part. For the
propeller part, all the statistics show first a small decrease and then an increase over the
steady values, with a gradual return to normal conditions.

Probability Density Function:

In order to compute the PDF, a value of the window width W must be chosen.
The estimate will have a bias and a random error, and a measure of the total error can be
obtained with the normalized mean square error (Bendat and Piersol, 1986):

where PDF" is the second derivative of the PDF with respect to T. The first term
expresses the influence of the random error, whereas the second is related to the bias
error. Conflicting requirements arise when choosing the value of W, since a large value
reduces the random error but increases the bias and vice versa. An idea of the appropriate
value of W can be obtained considering a standardized normal PDF and minimizing Eq.
4.7 at ? = 0 withN = 48:
which is satisfied for W = 1.5 . A normal PDF spans from around -4 to 4, so this value of
W gives only 5 points to describe the whole PDF. It was decided then, to use a value of
W = 1 and, in this way, get approximately 8 points for each PDF.
Figure 4.14 presents the sampled PDFs computed at some 24 characteristics
points in time. The shear stress values have been standardized (they have zero mean and
unit variance) and the normal distribution has been added to the graphs for comparison.
For approximately steady flow conditions, the PDFs approach a log-normal
distribution, in accordance with the findings of other researchers already commented
earlier. However, in the unsteady part of the signal the PDFs deviate from the log-normal
type, sometimes towards a more Gaussian shape.

Analytical Expression for the PDF:

The results already presented show that, the general shape of the PDF of the bed
shear stress for stationary conditions approaches a log-normal distribution with typical
values of z', skewness (S) and flatness (F) are near 0.4, 1, and 5, respectively. It has also
been shown that this PDF changes for unsteady conditions, and so do its statistical
moments. In order to obtain a general expression for the PDF valid for unsteady
situations, we can express it in terms of its moments that are in turn functions of time.
This can be done using the characteristic function cP, defined as (Monin and Yaglom,
1971):

I
+m

@(k) = PDF(z) e-jTkdz

-CO
(4.9)
where k is the frequency and j = fi. Note that @(k) is the Fourier transform of the

PDF(z) and its Taylor series expansion is related to the moments of the PDF(z) by:

with the moment of order i, Mi , defined as:

 So the moments Mi , define the PDF(r) uniquely. In turn, the low order moments
can be related to the variance, skewness and flatness. However, the procedure described
requires the knowledge of all the moments of the distribution, since the truncation of the
series in Eq. 4.10 introduces considerable error. Thls is due to the fact that as the order
increases, the moments get greater (Monin and Yaglom, 1971). An alternative approach
consists in the expansion of the logarithm of cP(k), which is expressed in terms of the
cumulants of the distribution, Ci, defined as:
8'

ci = (-j)l -
dk
ln@(k)lk = o

It can be shown (Monin and Yaglom, 1971) that the cumulants, unlike the
moments, tend to zero as the order increases. This allows neglecting higher order
cumulants and still getting a good description of the PDF. Neglecting cumulants of order
higher than four, Nakagawa and Nezu (1977) obtained the following expression for the
PDF:

where ? is a variable with zero mean and unit variance, G(? ) is the Gaussian distribution
of the variable, and the cumulants can be computed from:

TI
J. ne- n nF
rvr, /-\ -- - I-
an o e
- -1
votained from Eq. 4.13 using the following transformation
rule (Bain and Engelhardt, 1987):

- n m
rur, (T) =
~ ' m -
-
1 ---
rur;
,Z-
(
< Z >
1
I dxl T' m 7' m
where 't = (z- < z >) /z', and subindices have been used to distinguish between the two
PDFs.
Combining Eqs.4.13, 4.14 and 4.15, an expression for the PDF, ( 7 ) as a function
of its mean, standard deviation, skewness and flatness is given by:

PDF, (T) =
-
(z- < z >)z

t' m
2zlm2

G
I[

Note that S and F are the same for z as well as for 'r . The above equation allows
for the computation of the PDF, (z) in terms of its first four moments. This statistics are
believed to describe properly the shape of the distribution, at least for steady and
moderate unsteady flow conditions. If the typical steady state values of z', S and F are
used in (16) the resulting PDF approaches a log-normal distribution.
The approach is also promising when extended to unsteady flow situations, since
in this case the PDF, (z) changes and deviates from the log-normal distribution. In Figure
4.14, the sampled PDFs are shown together with the analytical expression obtained in
Eq.4.16 Despite some dispersion expected due to the limited number of realizations, the
predicted PDFs follow the general trend of the sampled PDFs reasonably well.
 / t=60 sec
-I

.'.O ".

t=66.5 sec
V

0.2

Figure 4.14 Probability density functions at characteristic points.

o computed -Gaussian - ---analytical
_
w
I
0.4 i t=68 sec

0 -2
O '

0.6
t=75 sec lt=80 sec

0.6
1 t=103 sec

Figure 4.14 (cont.) Probability density functions at characteristic points.

o computed -- Gaussian - - ---analytical
 t=105 sec

0.2
I

0.6 .
I
.
\

0.4
,, t=108 sec : ',
IC3

0.2
--.

Figure 4. I4 (cont.) Probability density functions at characteristicpoints.

o computed -Gaussian -- - - analytical
4.4. UIUC Experiments

4.4.1. Introduction
After reading the previous section, it is clear that the conditions at WES were not
ideal for the operation of the sensors. The limited amount of realizations due to time
constraints and the calibration not done in situ were factors that would have limited any
conclusions based on those experiments alone. In other words, measurements under more
controlled conditions were necessary. The measurements carried out at Hydrosystems
Laboratory (HSL) of the University of Illinois, did not attempt to reproduce the exact
conditions of the WES experiments but to simulate some characteristics considered of
importance to the phenomenon under study, and in particular the effect of unsteadiness
on the shear stresses.

4.4.2. Facility at HSL

The unsteady experiments were carried out in the facility illustrated in Figure
4.15, which was also used for calibration purposes (see section 2). This apparatus was
designed to study sediment resuspension and essentially consisted of a rectangular water
tunnel made of plexiglass with a head tank at one end and a pneumatic valve at the other
end. The water delivered through the tunnel is collected in a downstream tank. The
facility can be operated either as an open system using outside alimentation to the head
tank, or as a close system by means of a pump that recirculates the water from the
collection tank back to the head tank. Manipulating the valve, the water level in the head
tank and the flow rate of the pump, both steady and unsteady flows can be generated in
the facility.The cross section of the tunnel has a height of 0. l m and an aspect ratio of 3 : 1.

tank

test section

I valve

Figure 4.15 Resuspensionfacility.

 This design is a result of a compromise between two requirements. The first one
is to have an aspect ratio as large as possible in order to obtain two-dimensional flow in
the center part of the cross section, and the other is to have a height big enough to make
accurate observations. Although the aspect ratio 3 : 1 is not enough to guarantee complete
two-dimensionality, secondary currents were found to have an insignificant effect in
preliminary experiments. The data was obtained near the downstream end of the tunnel,
which is 6 m long. In this region, the complete development of the flow was tested
comparing consecutive velocity profiles obtained with a Pitot tube.
The facility has pressure taps on the upper wall of the tunnel to monitor pressure
gradients. Several ports allow the positioning of different instruments, i.e. Acoustic
Doppler Velocimeter (ADV), Pitot tubes, Hot Film sensors, Acoustic Concentration
Profiler (ACP) and concentration sampling tubes.
The experimental setup used during these experiments is presented in Figure 4.16
It includes two shear stress sensors (TSI 1237W) flush mounted to the upper wall of the
tunnel and a velocity sensor (ADV) located at the centerline of a downstream section,
which can be moved up and down inside the tunnel if so desired.

HOT FILM VELOCITY

SENSOR^-7 SENSOR

Figure 4.16 Experimental setup for the HSL experiments.

4.4.3. Description of the Experiments
Five different sets of experiments were carried out. In the first t lree cases, the
velocity inside the tunnel was varied linearly by imposing a triangular iI7aveinput to the
valve, and in the last two the velocity was linearly increased and then it leld constant.
The rate at which the velocity was increased varied from set to set. The characteristics are
summarized in Table 4.4:

Table 4.4 Characteristics of the shear stress signalfor the HSL exprriments.
Exp # l a Part Acceleration Max. Veloc. 2ndPart Acceleration
1 2 seconds 0.7 m/s2 1.4 m/s 2 seconds - 1.7 m/s2
2 4 seconds 0.4 m/s2 1.6 m/s 4 seconds - 4 m/s2
L

3 8 seconds 0.2 m/s2 1.6 m/s 8 seconds - 0.2 d s 2

4 4 seconds 0.35 m/s2 1.4 m/s 50 seconds 0 mls2
5 8 seconds 0.17 m/s2 1.5 m/s 44 seconds 0 m/s2

As it can be seen, the characteristic values are similar to those measured at WES.
The two shear stress sensors operated at 500 Hz and were calibrated in situ. Velocity
measurements were also gathered during these experiments. An Acoustic Doppler
Velocimeter (ADV) measured instantaneous velocities at 25 Hz in the centerline of the
channel and at 0.5 cm from the bottom. The velocity and shear stress records were
synchronized by means of an external trigger.
The runs consisted in periodic repetitions of the signal spaced by a time lapse of
no velocity. Later, during the data analysis, all the periods were collapsed in a common
system of reference and treated as independent realizations. A total of 60 realizations
were gathered for each situation.

4.4.4. Results
The experiments conducted at HSL were more precise and detailed, with around
60 realizations for each experiment, a higher measuring resolution (500 Hz sampling
frequency), and two independent measurements recorded by two sensors. This provided a
reference set of data for comparison with the WES experiments. Since the statistical
estimates of <z>, zt, S and F give essentially the same information that the PDF, the
computation of the probability distributions was not performed for these experiments.

Statistical Estimates:
Both sensors gave basically the same readings, with a difference on the order of
5%. However, it was decided to use only data from one sensor to compute the estimates
instead of mixing the records of the two of them since, strictly speaking, they belong to
different populations. The second sensor was used only for verification purposes.
As done with the WES experiments, estimates were computed for <z>, z', S and
F, and are presented in Figures 4.17 to 4.21 for experiments number 1 to 5, respectively.
A moving average of 1/25 seconds was applied to the series for smoothing purposes.
In general, the trend is similar to that observed for the WES experiments, with 6,
S and F
changing to 0.2, 0, and 3 as a result of the unsteadiness. For the triangular shape
experiments (test #1 to #3), the statistics return to the original values once the unsteady
part finishes, whereas for the plateau shape (runs #4 and #5) they remain at the new
levels.

4.5. Conclusions
Detailed measurements of wall shear stress have been taken under unsteady flow
conditions. Two sets of experiments were carried out, the first one simulating the effects
of the passage of a tow barge, and the second featuring simplified patterns with a better
time resolution and more realizations. After analyzing the results of these tests, the
following conclusions can be drawn:
Before and after the barge passes, the statistics tend to values similar to those
predicted for steady state conditions which are 0.4, 1, and 5 for z', S and F, respectively
The corresponding PDF approaches a log-normal distribution.
The wall shear stress and its statistics are strongly modified in the vicinity of the
front of the barge and by the passage of the propellers.
The change in the statistics in the front part, is characterized by a decrease of the
value of z' to 0.25, and a more erratic behavior of S and F due to the limited number of
samples. Later experiments carried out at HSL confirm the trend in r' and show that S
and F tend to their Gaussian values of 0 and 3, respectively.
For the propeller part the pattern is more complicated, with all the statistics
showing first a small decrease and then an increase over the steady values, with a gradual
return to normal conditions after the passage of the barge train.
Concerning the PDF of the shear stress, it has been found that a fourth order
cumulant expansion provides a reasonable approximation to the sampled PDF in terms of
<T>, T+,S and F.
 4.5 5 5.5 6 6.5 7 7.5 8

tim e ( s e c )

4.5 5 5.5 6 6.5 7 7.5 8

tim e(sec)

Figure 4.17: Estimates of <z>, zC,S and F corresponding to test # I ofthe HSL
experiments.
 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5
ti me (sec)

6,S and F corresponding to test # 2 of the HSL

Figuve 4.18: Estimates of a>,
experiments.
Figure 4.19: Estimates of <T>, T+, S and F corresponding to test # 3 of the HSL
experiments.
Figure 4.20: Estimates of <z>, z,' Sand F corresponding to test #4 of the HSL
experiments.
 time (sec )

Figure 4.2 1: Estimates of <r>, r+,S and F corresponding to test #5 of the HSL
experiments.
 5. FIELD APPLICATION

5.1. Introduction

One of the driving objectives of the present work was the quantification of the
amount of sediment that gets entrained during the passage of a tow barge and, in a more
general framework, the consequences of navigation on sediment transport. This chapter
shows a methodology that incorporates the unsteadiness effect in the computation of
suspended sediment transport. Under very high shear stresses most of the transport occurs
as suspended load; besides, suspended sediment is more likely to be affected by inertia
(unsteadiness) forces than the bed load.
The two applications that close this chapter are intimately related with the analysis
carried out in previous chapters. The first one is a computation of the amount of sediment
entrained by a barge tow from a 0.5mm unifo- sand bed. The second is a computation
of the scour produced by the propeller of the barge tow on two different uniform-size
beds, i.e. 0.5 and 0.1 mm.

5.2. Sediment Entrainment in Unsteady Flow

5.2.1. Background
The depth-integrated forrn of the transport equation for uniform size suspended
sediment considering a nearly horizontal plane bed, simplifies to (Garcia, 1997):

where C is the mean concentration in the vertical, h is the flow depth, U, is the mean
strearnwise velocity, v, is the fail velocity of the sediment, Es is the entrainment and < is
the near bed concentration averaged over turbulence. The right hand side of (5.1) is the
net upward vertical flux of sediment at the bed and can be interpreted as the balance
between two terms, vs Es accounting for the resuspension and v, E,, representing the

deposition. In order to solve (5.1) a specification of Es is required, and considerable

research effort has been dedicated to obtain this parameter as a function of other flow
variables and sediment characteristics.
Up to date, all of the available relations are of the type (Garcia and Parker, 1991):

where the generic functions fi and f2 are nonlinear in T or u* with different powers
depending on the formulation chosen, and D is a measure of the sediment diameter
(usually D jo) . All of them have been developed for equilibrium conditions, which means
that no net erosion or deposition occurs. Under this assumption, the right hand side of
(5.1) is zero, and thus Es =Eb (in fact, experimentally Sb is used instead of Es to obtain fi

or fi for equilibrium conditions). This is unrealistic in unsteady flow, and although some
researchers have extended the use of such relations to nonequilibrium situations (van
Rijn, 1987) their applicability has never been demonstrated.
In order to avoid the shortcomings mentioned, an alternative approach is presented
herein. The present analysis recognizes the interacting variables of (5.2) as stochastic,
and provides relations between them in terms of their statistics (PDFs and low order
moments) which, in turn, are able to capture the unsteadiness (or nonequilibrium).

5.2.2. The Stochastic Approach

The usual approach to the problem of entrainment is to consider it as a deterministic
quantity. Consider for example the Garcia and Parker (199 1) relationship:

where A is a constant with value 1 . 3 1~0-7and Z, is defined as:

with u* and v, being the friction velocity and the fall velocity of the sediment,
respectively. The particle Reynolds number is expressed as:
where g is the gravitational acceleration, R is the submerged specific gravity, D5()is the
particle mean diameter and v is the kinematic viscosity.
If we consider a particular particle size and fluid, the entrainment would be only a
function of u., which is in turn related to the bed shear stress r, so we can write:

E, = f, (r) = f, (u*)

where fi and f2 express now one to one relationships. In a turbulent flow the bed shear
stress is a stochastic variable, and so should be the entrainment of sediment into
suspension. As is common with turbulent quantities, the instantaneous values will be
expressed as the sum of an ensemble mean and a fluctuation, i.e. r=<r>+r', Es =< Es >+
E,' . Accordingly, the ensemble average value of the entrainment is defined as:

< E, >= JE, PDFE(E,) dE,

where PDFE strn& for the probability density function of Es. Since (5.6) is a one to one
relation, we can express the PDFE as a function of the PDF of the shear stress PDF,:

PDF, (E,) = PDF, (7) (5 -8)

Changing variables in (5.7) and considering only positive values for Es and r:
 Graphically, (5.9) can be considered as the area below the curve resulting from the
multiplication of the entrainment and the probability density function o f t (see figure
5.1).
The proposed approach allows for the computation of not only ensemble averaged
values of entrainment, but also higher order moments (i.e. variance, skewness, flatness)
of its probability distribution. The information required is the shear stress probability
distribution in a form suitable for the integration, like (5.1 1) in the previous chapter. A
very similar approach has been used to assess bedload transport by unsteady flows (Grass
and Ayoub, 1982).

5.2.3. Computation of Sediment Entrainment into Suspension

Equation (5.3) has been derived based on observations of mean values and, since Es
is a random variable, it should be written as:

where the denominator of (5.3) has been omitted, since it is an ad hoe term that accounts
for a higher limit of the entrainment (Garcia and Parker, 1991) and can be replaced by a
condition < Es > 1 0.3. Assuming that the instantaneous relation between entrainment
and wall shear stress obeys a similar equation:

Combining (4.1 6), (5.9) and (5.1 1) the equation for the entrainment takes the forrn:
 7 (Pas)

Figure 5.1: Methodology used to predict entrainment.

After integration:

where the following notation has been used:

and II ...I5 are functions of 'r only, and are defined by:

In (5.1 5) Lz (x) are generalized Laguerre polynomials. The values of I1.. .I for different

r+ can be obtained from figure 5.2.

1jo uo.~j~unJa
S D sanlan SI pua +I 'CI 'ZI '11 :z'ga ~ n z ! ~
 Equation (5.13) is valid for both steady and unsteady flow, since the effect of
unsteadiness is embedded in the probability density function. Its performance can be
tested using typical steady flow values for the statistical parameters, i.e. r+=0.4, S=l and
F=5. For this particular set of the parameters, (5.13) takes the fonn:

or, in terms of Z,:

which is very close to the original equation (5.10), as it can be seen in figure 5.3, where
both equations have been plotted in the <Es>, Z, plane. The line resulting from using
<r>+r',, in (5.10) has also been included in the graph.

5.2.4. Practical Considerations

Equation (5.12) requires the evaluation of the five expressions II. . .Is in order to
calculate the entrainment, which limits its applicability from a practical point of view.
Moreover, the first four moments of the probability distribution must be computed to
evaluate r+, S and F, which is also a problem since, for the same number of samples, the
error in the estimates of the statistical moments increases with the order of the moment. It
is of interest, then, to explore the performance of a probability density function defined
by the lesser number of moments possible. With this in mind, the terms in (5.12) can be
evaluated in order to quantify the relative importance of S and F. Grouping terms, (5.12)
can be rewritten as:
Figure 5.3: Entrainment relations for steady flow.

70

In (5.18) J1and Jz are functions of T+ only, given by:

In order to compare the order of magnitude of the terrns in (5.18), I*;JI and J2 can be
plotted as a function of T'. This is shown in figure 5.4, where absolute values have been
taken on the functions J1and J2 in order to facilitate the plot in a logarithmic scale; this is
valid since the analysis is concerned with magnitude, not sign.

A r nrS
I . C-VL

Figure 5.4: 11,Jl and J2 values as a function of z+.

From the figure above it is clear that, except for very low values of ?+(which are
rarely found in turbulent flows) I1 is greater than J1 and J2 by at least an order of
magnitude. Even considering unusually high values of S and F (i.e. 5 and 20,
respectively) the first term on the right hand side of (5.18) is much more important than
the other two. If this is the case, the contribution of the terns involving S and F can be
conveniently neglected, resulting:

which corresponds to a Gaussian distribution, since (5.20) can be readily obtained from
(5.18) using S=O and F=3. Further simplification can be obtained fitting a power law to I1
for the range of r' from 0.2 to 1 (see figure 5.5) and writing:

or in terms of Z,, which is now expressed as a function of <r>:

The steady state (~+=0.4)

result of (5.22) reduces to:

which is very close to (5.17).

5.3. Application 1: Computation of Sediment Entrainment Induced by Navigation

Traffic

The methodology presented to compute sediment entrainment was used to evaluate

the resuspension of bed material that occurs in a river during the passage of a barge. The
wall shear stress measurements gathered during the WES experiments were scaled up to
the prototype scale used in the ca!sul.tions. After scaling ~ pthe
, characteristic vdues
for the simulated situation are the following:
Boat direction: upbound

Flow velocity: 1 m/s

Boat velocity: 1.5 m/s

 Boat length (barges+tow-boat): 3 10 m

Boat width: 32 m
Flow depth: 8.89 m
Draft: 2.74 m
Maximum shear stress: 46 Pa
Instantaneous entrainment was computed for some characteristic points both using
(5.13) and the traditional methodology, i.e. using (5.3) with <r> as a descriptor of the
instantaneous shear stress. A sediment size of 0.5 mm (v,=7.9 cm/sec) was considered,
which can be taken as typical of portions of the Upper Mississippi River (Pool 8 to be
more precise). Figure 5.6 shows the results plotted in the Zu-<Es> plane (note that 2, is
always computed using <r>).
The entrainment computed using (5.22) is plotted in figure 5.6 also. As it can be
seen, it provides a good approximation of the complete formulation. Using (5.22),
instantaneous entrainment along the centerline of the propeller was computed and is
plotted in figure 5.7, together with the one calculated using the traditional approach.
The total sediment entrained by the passage of the barge can be computed using the
proposed methodology and some assumptions. First of all, a Lagrangian system of
reference moving with the barge at a constant velocity will be adopted. In this way, the
time coordinate will be replaced by a longitudinal coordinate (x), with origin at the front
of the barge. The transverse coordinate (y) will have its origin at the centerline of the
barge. Since there is no available information on the behavior of the PDFs away from the
line of measurement (centerline of one propeller) it will be assumed that,'t S and F are
constant in y.
Figure 5.5: Power law fit to I/.
Figure 5.6: Entrainment computations.
 The whole shear stress signal will be analyzed separating the effect of the bow and
hull of the barge from the effect of the stem. The mean flow effect on the entrainment
will be considered negligible.
Based on information obtained during the mentioned experiments, the shear stress
produced by the front and hull was considered constant in the region underneath the
barge. For the outer region a linear decay with zero value at a distance equal to the barge
width was assumed. The total sediment entrained by the front and hull will result from
the integration of the value of the concentration cb=< Es (x,y)>v, over x and y, as follows:

where w is the width of the barge, in our case 32 m, and 310 m has been chosen as the
distance on the x axis of influence of the front and hull.
The effect of the propellers was considered assuming that they behave as a jet in the
far field. Following this idea, a Gaussian decay was proposed for the shear stress, of the
form:

<r (x,O)> is the maximum value of the shear stress and is located on the axis of the jet.
Based on observations during the experiments, it was considered that the signal recorded
at the centerline of the propeller was a good estimate of the signal at the centerline of the
barge (y=O).
Since it was assumed that T+ does not vary with y, r'=<z>z+ also follows (5.25). In
consequence:

The total sediment entrained by the propellers is given by:

where it has been considered that the propellers entrain sediment betweeii x=3 10 m and
x=1700 m. Using (5.27) and integrating first in y gives:

Numerical integration yields:

In this way, the total sediment entrained by the barge will be:

If instead of a Lagrangian frame of reference an Eulerian one is used, the results is:

Although the present results are approximate, they give an idea of the huge amount
of sediment that is resuspended during barge passages. The order of magnitude of the
computed values of resuspended sediment seem to agree with the ones reported by Sutton
(1996) and corresponding to ship maneuvering effects.
It must be pointed out that the situation considered corresponds to an upbound-
moving barge in a very shallow water body, which is the worst scenario in terms of flow
dynamics. However, the results can be even more dramatic if we consider a smaller
sediment size, not only because of the larger volume expected but also due to the fact that
the finer material can remain suspended into the water column, unlike the 0.5 rnm
material considered herein that re-deposits almost immediately. The 0.5 mm diameter
sediment can be regarded as typical of the Upper Mississippi River, a very traveled
waterway where this type of resuspension events due to barge passage frequently occurs.
Another heavily trafficked river, the Illinois, is more in the range of the 0.1 rnm, so the
expected amount of material entrained is even greater.
5.4. Application 2: Computation of the scour produced by a tow barge passage

The variation of the bed level can be computed using the Exner equation:

where h, is the porosity, q is the bed position and qb is the bedload transport per unit
width. qb can be estimated using, for example, the Engelund and Fredsoe relation:

where r,' = 0.05 is a critical Shield's stress, Ds is the sediment diameter (usually Djo ),
and the dimensionless variables are defined by:

The entrainment can be estimated using the Garcia and Parker relationship (5.3), and
for < we can use the expression:

where ro is a shape factor, which for the case of a Roussean concentration profile takes
the foim (Parker, 1982):

To close the problem, C also needs to be found. The suspended sediment

conservation equation can be written in the form:
where advection and lateral diffusion of suspended sediment have been neglected. Using
again the shape factor for E, , (5.38) now reads:

which can be solved for C for a given time series of shear velocities u*.Notice that h is
the mean flow depth and is taken to be constant for the computation of C. Once C is
obtained, replacing back in (5.36) gives cb.

Solution:
Basically, the problem consists of solving two differential equations, namely (5.39)
and (5.32). An important simplification that can be made in this problem is that x and t
are related through the velocity of the boat Ub. This reduces the problem to only one
variable, i.e., time.
Equation (5.39) can be solved for C numerically. For example, a simple forward
difference, explicit scheme would give:

where the subindex 1 corresponds to the values of the variable computed in a previous
time step. To solve the equation an initial value for C, i.e., Co = 0 is needed.
Once C is obtained, the next step is to proceed with the Exner equation. The relation
between time and distance is given by:

where Ub is the boat velocity. Replacing in the Exner equation yields:

which has a simple solution:

4 + hC = constant
q(1- h p )+
Ub

The constant represents the state of the system before the barge passes and can be set
equal to zero. Finally, the expression for q reads:

Sample computations are attached in figures 5.8 and 5.9 for a sediment diameter of
0.5 mm and 0.1 mm, respectively. A value of 0.2 was selected for the porosity. The time
step should be selected carefully, because the simple explicit scheme used in (5.40) will
become unstable otherwise. By inspection of (5.40), it is possible to develop a simple
criterion to select the time step At by requiring that

For the attached examples, Ub = 1.5 rn/s and h = 1 m, so from (5.45) a time step,

1.5m
At - 0.1 seconds
1.5 m/s

is obtained. The above criterion ensures that in (5.40), the parameter -

A t vSis quite small
h
and the computations remain stable.
The results show the big difference in the scour-hole produced for the two situations.
As already commented, the coarser material deposits almost immediately and 100
seconds before the propellers passed the scour hole has been completely filled up. In the
case of the lighter sediment, it takes ten times that time to fill the scour. The scours also
differ a lot. In the first case only 0.05 m have been reshaped, whereas in the second this
depth reaches as much as 0.2m.
Lopez, F. (1994) Near-wall coherent structures and their role on sediment transport in
smooth-bed open channel flows, Masters thesis. University of Illinois at Urbana-
Champaign.
Maynord, S.T. (1990) "Velocities induced by commercial navigation," Waterways
Experiment Station Technical Report HL-90- 15, Vicksburg, Mississippi.
Mazumder, B. S., Bhowmik, N.G., and Soong, T.W. (1 993) "Turbulence in rivers due to
navigation traffic," Journal of Hydraulic Engineering, 119(5), pp. 58 1-597.
Menendez, A.N., and Ramaprian, B.R. (1985) "The use of flush-mounted hot-film gauges
to measure skin friction in unsteady boundary layers," Journal of Fluid Mechanics,
161, pp. 139-159.
Monin, A. S. and Yaglom, A.M. (197 1) Statistical fluid mechanics. Vol 1, MIT Press.
Nakagawa, H. and Nezu, I. (1977) "Prediction of the contribution to the Reynolds stress
from bursting events in open-channel flows," J. Fluid Mech., 80(1), pp. 99- 128.
Parker, G. (1982) "Conditions for the ignition of catastrophically erosive turbidity
currents," Marine Geology, 46, pp. 3 07-327.
Sutton, D.W. (1996) "Mass loading and subsequent baywide transport of sediment
resuspension during tug assisted ship movements at the Naval Station, San Diego.
Part I: Field study and deterministic computer model simulation," Proc. AGU Fall
Meeting, San Francisco, California.
Van Rijn, L.C. (1997) Mathematical modeling of morphological processes in the case of
sediment transport, Thesis approved by the Deift University of Technology. Delft
n y d i . Communication No 3 82.
T T-
Appendix A. Stern shear stress measurements

Figure A. 1 Stern shear stress~foo

print 1.2 7 cm from the barge centerline, downbound run A
Figure A. 3 Stern shear stress footprint 11.4 cm @om barge centerline, downbound run A
Figure A. 4 Stern shear stressfoodprint 17.8 cm from barge centerline, downbound run A
Figure A. 5 Stern shear stressfootprint ;?4.I cm from barge centerline, downbound run A
Figure A. 6 Stern shear stress distribution 30.5 cm jkom barge centerline, downbound run A
-50 -40 -30 -20 -10 0 10 20 30 40 50 601 70 80 90 100 110 120 130 140 150
turn,
Figure A.8 Stern shear stress footprint 5.08 cmfi.om barge centerline, downbound run B
Figure A. 9 Stern shear stress distribution 11.4 cm @om barge centerline, downbound run B
-50 -40 -30 -20 -10 0 110 20 30 40 50 60 70 80 90 100 110 120 130 140 150
turn,
Figure A. I0 Stem shear stressfootprint 17.8 cm @om barge centerline, downbound run B
 ?

--

--

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

,P
% 0.6 -

'-t

-50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
tU/H,
irr~t?,downbozmd run B
Figure A. 11 Stern shear stress footprint 24.1 cm from barge c~*?t;,
-50 -a40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
tU/H,
Figire A. 12 Stern shear stress footprint 30.5 cm fion~barge centerline, downbound run B
Figure A. 13 Stern shear stress footprint 1.2 7 cm .from barge centerline, downbound run C
Figure A. 14 Stern shear stress footprint 5.08 cm from barge centerline, downbound run C
Figure A. 15 Stern shew stvess footprint 11.4 cmfi.om barge centerline, downbound run C
-50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
turn,
Figure A. 16 Stern shear stress footprint 17.8 crn from barge centerline, downbound run C
Figure A. 18 Stern shear stress footprint 30.5 crn fiorn barge centerline, dowrzboz,nd run (7
<t-'% 0.6

Figure A. 19 Stern shear stressfootprint 1.27 ern @om the barge centerline, downbound run D
Figure A.20 Stern shear stress footyrint 5.08 cmfi.orn barge centerline, downbound run D
-50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
turn,
Figure A.21 Stern shear stress footprint 11.4 cm from barge centerline, downbound run D
Figure A. 22 Stern shear stressfootprint 17.8 cm @om barge centerline, downbound run D
Figure A.23 Stern shear stress foolprint 24.1 cm,from barge centerline, downbound run D
figure A. 24 Stern shear stress footprint 30.5 cm from barge centerline, downbound run D
Figure A. 2 7 Stern shear stressfootprint 11.4 cm from barge cerrterline, upbound run A
Figure A. 31 Stern shear stressfootprint 1.27 cm @om barge centerline, upbound run B
Figure A. 32 Stern shear stressfootprint 5.08 cm from barge centerline, upbound run B
Figure A. 33 Stern shear stressfootprint 11.4 cm from barge centerline, upbound run B
-30 -20 -10 O 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
tU/H,
Figure A. 43 Stern shear s fressfoo print 1.2 7 cm from barge centerline, upbound run D
-30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
turn,
Figure A. 44 Stern shear stress footprint 5.08 cmJiom barge centerline, upbound run D
Figure A. 45 Stern shear stress footprint 11.4 cnz from barge centerline, uphound run D
-30 -20 -10 O 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
turn,
Figure A. 46 Stern shear stress footprint 17.8 cm from barge centerline, upbound run D
Figure A. 47 Stern shear stress footprint 24.1 cm from barge centerline, upbound run D