Sie sind auf Seite 1von 17

Click

Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, C12015, doi:10.1029/2009JC005356, 2009

for

Full
Article

Ripple morphology under oscillatory flow:


2. Experiments
F. Pedocchi1,2 and M. H. Garca1
Received 3 March 2009; revised 28 July 2009; accepted 17 August 2009; published 11 December 2009.

[1] Recent large-scale laboratory experiments on the formation of ripples under

oscillatory flow are presented. The experiments were performed in the Large Oscillatory
Water-Sediment Tunnel (LOWST) at University of Illinois at Urbana-Champaign, using
250 mm silica sand as sediment. The dimensions of the ripples formed under a wide range
of flow conditions are compared with some of the existing predictors and with a new
predictor presented in a companion paper. For a given near-bed water excursion the size of
the ripples is observed to initially decrease with the increase of the maximum orbital
velocity, as has been suggested before. However, an abrupt change of the ripple size and
the transition to large round-crested ripples is observed when the maximum orbital
velocity becomes larger than 0.5 m/s. Above this value the size of these round-crested
ripples continuously increased with the increase of the maximum orbital velocity.
Additionally, anorbital ripples were never formed despite the long water excursions used
in several of our experiments, confirming that anorbital ripples are only formed in fine
sands. Finally, the performance of the existing planform geometry predictors and a newly
proposed predictor is evaluated using our new experimental data. The results confirm that
the bed planform geometry is controlled by the wave Reynolds number and the
particle size. The comparison or the new data with previous results from narrow facilities
shows that the facility width can restrict the development of bed form threedimensionality.
Citation: Pedocchi, F., and M. H. Garca (2009), Ripple morphology under oscillatory flow: 2. Experiments, J. Geophys. Res., 114,
C12015, doi:10.1029/2009JC005356.

1. Introduction
[2] When a fluid moves over an erodible sediment bed, it
may interact with the sediment particles, transporting and
redistributing them along the bottom. Subject to this
transport, an initially flat bed may become unstable and
give rise to the formation of wavy features on the bed.
These bed features or bed forms in turn affect the water
flow, which results in a strong coupling between the fluid
motion, the sediment transport, and the bed morphology. In
coastal and continental shelf areas the surface waves induce
oscillatory water motions in the vicinity of the seabed. In
the presence of a sandy bed these oscillatory motions
produce bed forms, called ripples. These ripples can either
scale with the near-bed water excursion or not, and therefore
they are known as orbital and anorbital ripples, respectively.
Similarly, ripples can either present a two-dimensional or
three-dimensional planform geometry, depending if their
crests are long and straight or short and wavy.
1
Ven Te Chow Hydrosystems Laboratory, Department of Civil and
Environmental Engineering, University of Illinois at Urbana-Champaign,
Urbana, Illinois, USA.
2
Now at Instituto de Mecanica de los Fluidos e Ingeniera Ambiental,
Faculatad de Ingeniera, Universidad de la Republica, Montevideo,
Uruguay.

Copyright 2009 by the American Geophysical Union.


0148-0227/09/2009JC005356$09.00

[3] Ripples modify the seabed roughness, which affects


the wave height as the waves propagate toward the
coast. This has a profound impact on large-scale coastal
morphodynamics. The presence of ripples also changes the
near-bed turbulence level, influencing the sediment
transport and the exchange of substances between the
seabed and the water column. On another aspect,
understanding the environmental conditions that lead to
the formation of a particular bed configuration is an
essential tool for interpretation of the sedimentary record.
[4] Oscillatory flow ripples were first systematically
studied in the field by Inman [1957], followed by Dingler
[1974] and Miller and Komar [1980b]. These early surveys
were performed by divers, limiting the observations to fair
weather conditions. Recent field efforts have included the
deployment of measuring equipment for several weeks,
allowing for the study of bed evolution under different
wave conditions [e.g., Traykovski et al., 1999; Hanes et al.,
2001; Xu, 2005]. In spite of great advances made on the
quality and quantity of data obtained in field campaigns, the
cost and time involved still limit the number of conditions
that can be actually studied. In this regard, laboratory
studies are significantly less expensive and the experimental
conditions can be fully controlled for long periods of time,
allowing for a more detailed observation of the processes
involved.

C12015

1 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

[5] In the past, most studies of ripples in the laboratory


mainly produced data for short oscillation periods and water
excursions. This was the consequence of the small size of
the experimental facilities used at the time: oscillating beds
and small wave flumes [e.g., Bagnold, 1946; Kennedy and
Falcon, 1965]. Ripples formed in oscillatory bed experiments have been found to significantly differ from those
produced in wave tanks [Miller and Komar, 1980a].
Similarly, the periods of the waves that can be obtained in
wave flumes are generally limited to a maximum of 6 s
[Miller and Komar, 1980a], even in the case of extremely
large facilities used more recently [Williams et al., 2004].
Oscillatory water tunnels are an alternative to these laboratory facilities in which these limitations are significantly
diminished. The water motion produced inside an oscillatory water tunnel is essentially a current that reverses its
direction periodically. This is not the exact flow condition
found under progressive waves, where the water motion at a
given time is not the same at every cross section. Despite
this difference, bed configurations generated in water tunnels tend to be in good agreement with the ones obtained in
wave flumes [Mogridge and Kamphuis, 1972], and water
tunnels have been extensively used in bed morphology
studies [e.g., Carstens et al., 1969; Mogridge and
Kamphuis, 1972; Lofquist, 1978; Southard et al., 1990;
Ribberink and Al-Salem, 1994; ODonoghue and Clubb,
2001; Dumas et al., 2005; ODonoghue et al., 2006].
[6] The size of oscillatory water tunnels and the type of
conditions that can be simulated have dramatically
increased since the pioneering work of Carstens et al.
[1969]. Newer tunnels have allowed for the study of long
oscillation periods, long water excursions, and high orbital
velocities that mimic the conditions found in the field
during storms [Southard et al., 1990; Ribberink and AlSalem, 1994; ODonoghue and Clubb, 2001; Dumas et al.,
2005; ODonoghue et al., 2006]. However, with few exceptions, tunnels have tended to be relatively narrow, thus
constraining the development of ripple three-dimensionality
and limiting the study of ripple planform geometry.
[7] The experiments described herein were performed
in the Large Oscillatory Water-Sediment Tunnel (LOWST)
at the Ven Te Chow Hydrosystems Laboratory at the
University of Illinois at Urbana-Champaign. The test
section of this facility is 12.5 m long, 0.8 m wide, and
1.2 m high. The sediment bed is 0.60 m deep, leaving a
0.60 m water column throughout the test section. This
allows for the study of ripple morphodynamics over a wide
range of flow conditions. Oscillations with long periods,
high orbital velocities, and long water excursions can be
generated within the tunnel covering a wide range of fieldlike oscillatory flows.
[8] Two main objectives were pursued with the experiments presented in this article: First, to identify the conditions that lead to the formation of two-dimensional or
three-dimensional ripples. The planform geometry of
ripples affects the near-bed hydrodynamics changing the
hydraulic roughness felt by the flow. ODonoghue et al.
[2006] recognized the importance of adequately predicting
the planform geometry of ripples and noted that the existing
planform geometry predictors were unable to predict the
bed configurations observed in their facility, which was
0.3 m wide. Second, to study the combined effect of long

C12015

near-bed water excursions and orbital velocities on the


wavelength and height of equilibrium ripples. There is
general agreement in the research community on the
importance of the water excursion for defining the size of
oscillatory flow ripples. And all of the commonly used
ripple-size predictors [e.g., Nielsen, 1981; Mogridge et al.,
1994; Wiberg and Harris, 1994] take the near-bed excursion
into account. Although it is clear that a second variable is
necessary to fully characterize a sinusoidal oscillation, there
has been significant debate on how this second variable
should be included in the predictors [ODonoghue and
Clubb, 2001]. The need for a second variable to characterize
the flow becomes evident under long water excursions,
where both orbital and anorbital ripples have been observed
[e.g., Inman, 1957; Traykovski et al., 1999; ODonoghue
and Clubb, 2001].
[9] Our experiments showed that the ripple formation
process can be extremely complex. In some tests, the bed
presented several pseudoequilibrium stages before a final
bed configuration was reached. For the weakest flows, the
final bed configuration was sometimes achieved only after
more than a hundred hours of running an experiment. On
the other hand, for intense flows the bed evolved rapidly
and reached equilibrium in less than an hour. These recent
experiments are analyzed in detail using an alternative
dimensionless framework presented in the companion paper
[Pedocchi and Garca, 2009b]. The terminology and
notation used herein are the ones defined in the companion
article.

2. Preliminary Considerations
2.1. Dimensionless Framework
[10] Previous studies have addressed the need for a
dimensionless set of variables to characterize the sediment
transport under oscillatory flows [Yalin and Russell, 1963;
Carstens et al., 1969; Mogridge and Kamphuis, 1972;
Dingler, 1974]. An extended revision was given in the
companion paper, and just a brief summary is included
herein. A sediment transport phenomenon x (e.g., ripple
wavelength) under oscillatory flow should be completely
defined by the sediment and fluid properties, the flow
condition, and the external forces. For the experiments
described here the sediment was quartz sand with a density
of rs = 2650 kg/m3. Its size distribution is given in Figure 1
from which D50 = 250 mm, D10 = 185 mm, and D90 = 373 mm
were estimated. As usual with natural sands, its submerged
angle of repose was 8 = 32. The fluid was fresh water, with
density r and kinematic viscosity n, both functions of the
measured temperature. The imposed flows were sinusoidal
oscillations, defined by a period T and a maximum orbital
velocity Umax. Note that for a sinusoidal flow, the near-bed
water excursion d verifies d = 2A = UmaxT/p. Finally, the
acceleration of gravity g must be included.
[11] Applying the Buckingham Pi Theorem, and defining
the submerged specific gravity R = rs/r1, the following
expression for the dimensionless form Px of x can be
derived:

2 of 17

p

T n Umax D50 gRD50 D50
;
;
;
R;
8
:
n
n
D250


Px f

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

Figure 1. Size distribution of the sand used in the present


experiments in the Large Oscillatory Water-Sediment
Tunnel (LOWST).
In our experiments both R and 8 were constants. The
third parameter
the right-hand side of
pon

equation (1) gRD50 D50/n is called Rep [Garca, 2008],


and for our experiments its variation was associated with
the change in water viscosity with the temperature. Since
the temperature range in our experiments was limited, the
range of variation of Rep was also fairly small. Therefore
our experiments can be properly presented and analyzed
using just two variables Tn/D250 and UmaxD50/n [Pedocchi
and Garca, 2009b].
[12] The dimensionless numbers Tn/D250 and UmaxD50/n
can be considered as a dimensionless period and maximum
orbital velocity, respectively. Combining them, the dimensionless water excursion A/D50 and the wave Reynolds
number Rew = UmaxA/n are obtained. For oscillatory boundary layer flows over a flat surface covered with particles
of size D50, the hydraulic roughness k can be approximated as k = 2.5D50 [Jensen, 1989]. The wave friction
factor fw = 2(u*max/Umax)2 relates the maximum shear
velocity at the bed over the cycle u*max with the maximum orbital velocity Umax and is a function of A/D50 and
Rew. The friction factor expression used here is the one
by Pedocchi and Garca [2009a] for smooth, transition,
and rough flow conditions (a summary is included in
Appendix A). Once the wall shear velocity is known, the wall
shear velocity Reynolds number Re* = u*maxD50/n can be
defined, which controls the boundary layer hydraulic regime
and defines smooth or rough flow conditions. In summary,
the oscillatory flow hydraulics over a flat sediment bed can be
fully characterized on a plane with Tn/D250 and UmaxD50/n as
main dimensionless variables. Furthermore, if the
third dimensionless number Rep is known, the Shields
dimensionless parameter q = u2*max/(gRD50) can be
computed.
2.1.1. Initiation of Motion
[13] As an example of the possible use of the described
dimensionless framework and to evaluate the performance

C12015

of the proposed expression for the friction factor, initiation


of motion experiments were performed. The results
obtained at three different water temperatures, 16C,
17C, and 26C are summarized on Table 1. The methodology
used to define the initiation of motion in these experiments
was naked-eye observation, which was highly subjective
but still fulfilled the objective of obtaining an approximated
value of the critical shear stress for our experiments. It also
allowed to observe the effect of the oscillation period on the
critical velocity that would initiate sediment motion.
Starting from a flat bed for each oscillation period, the
velocity was gradually increased until particle motion was
observed. This was done for a sequence of increasing
oscillation periods and then for a sequence of decreasing
oscillation periods. No significant difference between the
values obtained both ways was found, and the values
reported in Table 1 correspond to the average of both.
[14] The results are shown in Figure 2, where they are
plotted on the (Tn/D250, UmaxD50/n) plane. Since the changes
in viscosity (temperature) affect the value of the third
dimensionless parameter Rep, not included in the plane
representation, data from the different experiments do not
collapse. At 16C, the water kinematic viscosity is n =
1.11*106 m2/s and Rep = 14.34; the critical Shields stress
that best approximates the experimental data is qc  0.063.
At 17C, n = 1.08*106 m2/s, Rep = 14.72, and the critical
Shields that approximates the experimental data best is qc 
0.070. For 26C, n = 0.87*106 m2/s, Rep = 18.25, and
qc  0.072. The trends in the experimental data follow the
constant Shields stress contours very well, showing the slope
change when the boundary layer transitions from smoothturbulent to laminar at Rew  6.6*104 [Pedocchi and
Garca, 2009a].
[15] Several modifications to the traditional Shields
diagram have been proposed in the literature, both for
unidirectional and oscillatory flows. For oscillatory flows
some authors [e.g., Soulsby and Whitehouse, 1997; Hanson
and Camenen, 2007] suggest that the critical Shield
parameter does not increase for small values of Rep but
instead tends to a constant value. In particular, [Hanson and
Camenen, 2007] suggest values of qc very close to 0.07,
which are in good agreement with our results. It is important
to note that the oscillatory flow friction factor fw is involved
in the determination of qc. There are several expressions in
the literature for the friction factor in oscillatory flows, but

Table 1. Summary of Experiments Performed to Study Initiation


of Motion
Umax (m/s)
T (s)

16C

17C

26C

2
3
4
5
6
7
8
9
10
12
15
20

0.13
0.16
0.22
0.21
0.23
0.24
0.25
0.27
0.27
0.27
0.28
0.28

0.12
0.19
0.20
0.23
0.27
0.25
0.30
0.30
0.30
0.30
-

0.17
0.22
0.26
0.25
0.29
0.29
0.30
0.30
0.30
0.30
-

3 of 17

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

C12015

Figure 2. Initiation of motion experiments for different water temperatures (16C, 17C, 26C), shown
on the (Tn/D250, UmaxD50/n) plane. The constant Shield curves that best fit the data are shown, with the
Shields parameter value indicated. The laminar-turbulent transition Rew = 6.6*104 and smooth-rough
transition Re* = 6 are also shown. Note the break on the data trend when the boundary layer transitions
from laminar to turbulent. Data from Table 1.
the results presented here were accurately captured with our
friction factor expression (see Appendix A).

3. Equipment
3.1. Large Oscillatory Water-Sediment Tunnel
[ 16 ] The Large Oscillatory Water-Sediment Tunnel
(LOWST) was designed to study sediment transport and
related phenomena under controlled wave current boundary
layer flows similar to those found in the continental shelf.
The LOWST was built with support from the DURIP
Program of the U.S. Office of Naval Research, and it was
constructed by Engineering Laboratory Design Inc. and
MTS Systems Corporation.
[17] The test section of the facility is 12.5 m long with a
0.8 m wide by 1.2 m high internal cross section (Figures 3
and 4). Half of the tunnel height (0.6 m) is full of uniform
size silica sand. Before each experiment, the sediment bed
can be flattened with the help of a cart that redistributes the
wet sand as it is pulled along the tunnel. The oscillatory
motion of the water is driven by three pistons that run inside
0.78 m diameter cylinders with a maximum nominal stroke
of 2.1 m. At the opposite end of the tunnel, a 1.0 m by 2.0 m
holding tank open to the atmosphere acts as a passive
receiver for the water displaced by the pistons. Three servo
motors, controlled by a computer, are in charge of imparting
the motion to the pistons through a screw gear system. The
system was designed for the pistons to drive the water in
both directions producing a symmetric periodic motion. The
pistons are able to displace up to 0.8 m3/s of water which
gives a maximum nominal velocity inside the tunnel of 2 m/s.
Finally, the maximum nominal piston acceleration is 2.1 m/s2.
During the start (stop) transients at the beginning (end) of a
run, the pistons increase (decrease) their motion amplitude
in about 5 to 10 oscillations to avoid potential water
hammer effects. The facility also has two centrifugal pumps
that allow for the superposition of a unidirectional current to
the oscillatory motion. A 0.36 m diameter PVC pipe

connected to the pumps directs the water and sediment


mixture from the reception chamber and discharges it under
the pistons through a diffuser (Figure 4). Sediment traps
placed at both ends of the test section collect the sediment
that is transported as bed load.
3.2. Pencil Beam Sonar
[18] Surveying bed bathymetry inside a close conduit like
the LOWST is challenging when compared to performing
the same task in more conventional flumes, where the
presence of the free water surface gives an easy access for

Figure 3. Detail showing an imaginary cross section of


the LOWST and its central part with the main window.
Dimensions in meters.

4 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

Figure 4. General view of the LOWST showing the recirculation system.

instrumentation [e.g., Catano Lopera and Garca, 2007].


After considering several possibilities, it was decided to
pursue the design of our own system based on a Imagenex
pencil beam sonar and a Velmex Bislide positioning system.
[19] The selected sonar is an L-shaped Imagenex 881L
Digital MultiFrequency Profiling Sonar, produced by
Imagenex Technology Corp., Canada. The positioning system is a Velmex B4800TS Motorized Rotatory Table with
Velmex VXM controller, produced by Velmex Inc., USA.
We designed the mechanical system to combine both units
and the system was built by the Civil and Environmental
Engineering Machine Shop at the University of Illinois at
Urbana-Champaign. The necessary software to control the
motion and acquisition of the backscatter data was developed
in collaboration with Nils Oberg, Software Engineer working
at the Ven Te Chow Hydrosystems Laboratory. The system
has two axes of rotation, a vertical one controlled by the
rotatory table and a horizontal one controlled by a stepper
motor inside the sonar head. As the ultrasound beam is
rotated over the horizontal axis, it covers a fan-shaped region
contained in a vertical plane. This plane crosses the bed over a
line. By rotating the vertical plane around the vertical axis,
several crossing lines can be acquired to give a complete
survey of the bed over an area.

4. Experiments
4.1. General Description
[20] The final bed equilibrium configuration of the
experiments performed in the LOWST are summarized in
Table 2. For most of the tests the initial bed configuration
was a flatbed in order to prevent any effect of the initial bed
conditions. However, for some particular cases the bed
configuration left by a previous experiment was used as
the initial bed condition. This is indicated in the second
column of Table 2. The flow conditions were selected in
order to cover a wide range of periods and orbital velocities,

to study the two-dimensional to three-dimensional transition


of the planform geometry, and to explore the effect of the
orbital velocity on the ripple size. The selection was also
based on the conditions previously reported by other
researchers from both filed measurements and laboratory
experiments. As is shown in Appendix B, most of the
simulated conditions could be generated by linear waves
in 5 m of water depth.
[21] The experiments were run for as long as it was
necessary to assure that the bed had reached its final
configuration. This was done in the following way: Once
the bed was thought to have reached equilibrium, the
experiment was continued for a reasonable additional time.
This additional time was between 10% and 20% of the
originally run time. If the general aspect of the bed remained
unchanged, the experiment was stopped. If instead changes
where found, the previous process was successively repeated
until a stable configuration was obtained. The total
running time is reported in the last column of Table 2.
When two-dimensional ripples were the final configuration,
no net ripple migration was observed, indirectly showing
that the water motion inside the tunnel was completely
symmetric. For the tests producing quasi two-dimensional
and three-dimensional ripples as a final configuration, the
bed never reached a truly stationary state. Instead, quasi
two-dimensional and three-dimensional ripples were found
to continuously move and rework the sediment bed, but
with their average size remaining constant. For experiments
lasting more than 10 h, the experiments were stopped at the
end of the day and restarted the following morning. The
oscillatory motions in the tunnel were slowly started and
stopped, and no effect of these transients on the final
configuration of the bed was noticed.
[22] The ripple development was found to be a complex
process, in some cases presenting several pseudoequilibrium
stages before the final bed configuration was reached. Once
the bed was clearly in equilibrium, the experiment was

5 of 17

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

C12015

Table 2. Summary of Experiments Performed to Study Ripple Formation and Final Equilibrium Configuration
Experimenta

Initialb
Condition

T (s)

dmax
(m)

Umax
(m/s)

l
(m)

h
(m)

h/l

l/d

l/D50

2-D/3-Dc

l/wd

Temperature
(C)

Duratione
(hours)

01
02
03L
03S
04
05
06
07L
07S
08
09
10
11
12L
12M
12S
13
14
15
16
17
18L
18S
19
20L
20S
21L
21S
22L
22S
23L
23M
23S

00
01
02
02
00
04
05
00
00
00
00
09
00
00
00
00
00
00
14
00
16
00
00
00
00
00
00
00
00
00
00
00
00

5.0
5.0
15.0
15.0
2.0
5.0
5.0
25.0
25.0
8.0
6.0
6.0
6.0
12.0
12.0
12.0
8.0
3.5
8.0
8.0
5.0
7.0
7.0
10.0
15.0
15.0
15.0
15.0
10.0
10.0
18.0
18.0
18.0

0.48
0.32
0.95
0.95
0.16
0.40
0.40
1.99
1.99
1.53
0.57
1.34
1.91
1.91
1.91
1.91
1.91
0.45
0.51
1.27
0.80
0.78
0.78
1.91
1.91
1.91
1.67
1.67
0.64
0.64
2.86
2.86
2.86

0.30
0.20
0.20
0.20
0.25
0.25
0.25
0.25
0.25
0.60
0.30
0.70
1.00
0.50
0.50
0.50
0.75
0.40
0.20
0.50
0.50
0.35
0.35
0.60
0.40
0.40
0.35
0.35
0.20
0.20
0.50
0.50
0.50

0.300
0.200
0.600
0.050
0.115
0.260
0.260
1.130
0.106
0.600
0.300
0.400
1.120
0.500
0.200
0.150
0.800
0.230
0.350
0.400
0.350
0.350
0.070
0.600
0.500
0.060
1.000
0.060
0.450
0.050
1.800
0.250
0.080

0.045
0.035
0.070
0.006
0.021
0.045
0.045
0.170
0.014
0.100
0.040
0.130
0.190
0.100
0.050
0.030
0.150
0.045
0.065
0.060
0.050
0.065
0.010
0.110
0.100
0.015
0.150
0.015
0.065
0.015
0.150
0.040
0.010

0.150
0.175
0.117
0.120
0.183
0.173
0.173
0.150
0.132
0.167
0.133
0.325
0.170
0.200
0.400
0.020
0.188
0.196
0.186
0.150
0.143
0.186
0.143
0.183
0.200
0.250
0.150
0.250
0.144
0.300
0.083
0.160
0.125

0.628
0.628
0.628
0.052
0.723
0.653
0.653
0.568
0.053
0.393
0.524
0.299
0.586
0.262
0.105
0.079
0.419
0.516
0.687
0.314
0.440
0.449
0.090
0.314
0.262
0.031
0.598
0.036
0.707
0.079
0.628
0.087
0.028

1200
800
2400
200
460
1040
1040
4520
424
2400
1200
1600
4480
2000
800
600
3200
920
1400
1600
1400
1400
280
2400
2000
240
4000
240
1800
200
7200
1000
320

3
2
2.5
2.5
2
2
2
2
2.5
3
3
3
2
3
3
3
2.5
3
2
3
3
3
2.5
2.5
3
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5

0.13
0.25
0.75
0.06
0.14
0.33
0.33
1.41
0.13
0.75
0.38
0.50
1.40
0.63
0.25
0.19
1.00
0.29
0.44
0.50
0.44
0.44
0.09
0.75
0.63
0.08
1.25
0.08
0.56
0.06
2.25
0.31
0.10

20
20
23
23
16
19
27
27
27
22
19
19
21
18
18
18
14
17
20
17
20
18
18
25
21
21
25
25
27
27
18
18
18

0.5
18.0
18.0
50.0
0.5
19.0
5.0
150.0
150.0
2.8
1.7
1.0
0.6
15.0
15.0
15.0
1.0
2.0
20.0
2.3
1.5
7.0
7.0
3.7
7.6
7.6
12.5
12.5
70.0
70.0
8.7
8.7
8.7

a
Experiment: L, M, and S at the end of the name indicate large, medium, and small bed forms. Small bed forms were observed superimposed on the
larger ones. Medium bed forms are used to describe complex morphologies. For the experiments in boldface at the bottom of the table, significant wall
effects were observed.
b
Initial condition of the sediment bed at the beginning of the experiment: 00 indicates flat bed, the other numbers indicate the number of the experiment
ran before, which description can be found also in this table.
c
Here 2 indicates two-dimensional bed forms, 3 indicates three-dimensional bed forms, 2.5 indicates bed configurations that shown three-dimensional
bed forms together with some two-dimensional ones, or bed forms with wavy crests.
d
Bed form wavelength to tunnel width ratio. If the bed form is reported as 2-D but l/w is larger than 1 the bed form cannot be considered truly twodimensional.
e
Duration of the experiment. This time was longer than the necessary time for the formation of a stable bed configuration.

stopped and the general features of the bed recorded. This


included taking photographs of the bed through the tunnel
windows and measuring the wavelength and height of
several ripples to obtain the mean ripple sizes reported in
Table 2. Finally, once the sonar system was operative, the
complete width of the tunnel over a length of 1.5 m was
surveyed.
[23] The definition of a mean ripple size was not always
an easy task. Particularly, for the quasi two-dimensional and
three-dimensional bed configurations. For these cases the
sediment bed presented an extremely complex pattern with
several ripple sizes coexisting. To describe these cases, the
selected approach was to summarize the mean features of
the bed by reporting more than one ripple size when
necessary. This is indicated after the experiment number
with the letters L, M, and S, for large, medium, and small,
respectively. The ripple crests were usually sharp. However,
for experiments that involved long excursions and high
velocities (Experiments 10, 11, and 13) the ripple crests
were round.
[24] As noted in the companion paper, the facility width
may affect the free development of the bed planform

geometry, either constraining or triggering bed threedimensionality. The ripple wavelength to the tunnel width
ratio l/w, which can be considered as a measure of the
restriction imposed by the tunnel width to the bed threedimensionality, is displayed in Table 2. For five experiments
(07, 11, 13, 21, and 23) this ratio is above 1. For experiments 07, 11, and 13 no significant wall effects were
observed, other than the fact that the tunnel width could
have limited the development of further ripple threedimensionality. On the other hand, for experiments 21 and
23, as well as for experiment 22, for which l/w = 0.56, the
effect of the wall presence was clearly noticeable on the
final bed configuration. For these experiments the large bed
forms reported in Table 2 lay alternately over the tunnel
walls, leaving a sinuous path between them. For these
experiments the large ripple wavelength reported corresponds to the distance between two crests lying over the
same tunnel wall. If the ripple wavelength is defined in this
way, these ripples fall in the orbital class in agreement with
their other geometric characteristics.
[25] The water temperature was recorded during the
experiments and its mean value is reported in Table 2. In

6 of 17

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

Table 3. Long Water Excursion Experiments Sorted by Maximum


Orbital Velocity
Experiment
d (m)
Umax (m/s)
l (m)
h (m)
q

07

20

12

19

13

11

1.99
0.25
1.13
0.17
0.04

1.91
0.40
0.50
0.10
0.11

1.91
0.50
0.50
0.10
0.16

1.91
0.60
0.60
0.11
0.22

1.91
0.75
0.80
0.15
0.35

1.91
1.00
1.12
0.19
0.62

all cases the variation of the water temperature for the


duration of the experiment was less than 2C. The effect
of temperature on water density can be considered as minor
for the phenomenon under study. However, changes in
temperature can produce significant variations in the water
viscosity. Controlling the temperature of the water inside the
LOWST was not possible at this time. The room where the
LOWST is located has no temperature control and the water
temperature inside the tunnel would adjust to the ambient
temperature, which varies with both weather and seasons.
Experiment 05 ended during the spring and the water
temperature was 19C, during the following weeks the
summer season started and the water temperature rose to
27C. This opportunity was used to observe possible water
viscosity effects on the particular bed configuration present
at that time. However, no significant changes where
observed as it is reported in Experiment 06. It is possible
that our sediment was not fine enough for viscous effects to
be easily noticeable. The role of viscosity on bed configuration has been repeatedly reported in the unidirectional
flow literature [e.g., Vanoni, 1974; Shen et al., 1978;
Southard and Boguchwal, 1990; Garca, 2008]. Since the
basic process governing the sediment transport should be
the same regardless of the unidirectional or oscillatory nature
of the flow, the role of viscosity on oscillatory flow bed forms
deserves further research.
4.2. Long Water Excursion Experiments
[26] Some of the experiments reported in Table 2 were
designed to cover a wide range of oscillatory flow conditions in order to identify the ones leading to the formation
of either two-dimensional or three-dimensional ripples. This
was not the only goal of our work and another set of
experiments was specifically designed to study the maximum orbital velocity effect on long water excursion experiments. These long excursion experiments are specifically
reported in Table 3, and for all of them the near-bed water
excursion was kept constant at 2 m. For four of these
experiments a detailed description supported by several
figures and Animations 1 4 is provided.
4.2.1. Experiment 07
[27] For this experiment, T = 25 s, Umax = 0.25 m/s, and d =
1.99 m. The experiment started from a flat bed. First,
small ripples grew from the sidewalls of the tunnel and
propagated into the central part of the sediment bed,
completely covering it. The wavelength of those small
ripples was 0.06 0.07 m, their height was approximately
0.015 m, and they were straight crested. We classified them
as rolling grain ripples [Bagnold, 1946], since no significant
flow separation occurred over their crests, and no suspended
sediment was observed either [Faraci and Foti, 2001]. Soon

C12015

after these ripples had covered the entire bed a doubling


process started. The doubling process originated at the
sidewalls of the tunnel and consisted in the disappearance
of every other ripple crest and the growth of the remaining
ones. No merging of ripples was observed and these larger
ripples quickly replaced the initial ones over the entire bed.
After this, the doubling process started again, but this time
small perturbations consisting of short trains of small ripples
moving along the bed, were observed at several locations.
However, aside from the presence of these perturbations, the
doubling process was the only growth mechanism observed
until the final bed configuration was reached, which
happened after 150 h of running the experiment. The
doubling process is a good example of water flow bed
morphology coupling. The initial small ripples perturbed the
flow generating circulation cells that induce sediment transport, which in turn was reflected on the growth of some
ripples and the depletion of others, leading to the increase of
the overall ripple size. A fast-forward video of this process
is included as Movie S1 in the auxiliary material.1
[28] The final configuration consisted of large sharp
crested ripples with l = 1.13 m and h = 0.17 m (Figure 5).
These ripples were clearly orbital ripples, since l/d  0.6
and h/l  0.15 [Clifton, 1976], despite the fact that the
dimensionless excursion d/D50  7600 was clearly above
previously defined limits for the occurrence of orbital
ripples [Clifton and Dingler, 1984; Wiberg and Harris,
1994]. Small superimposed ripples were observed migrating
on top of the larger ones. These superimposed ripples
traveled from the troughs of the large ripples toward their
crests, where they were washed out as they were exposed to
higher shear stresses. The superimposed ripples had short
sharp crests, and their wavelength and height were 0.11 and
0.014 m, respectively. They looked very much like
unidirectional flow ripples, which was basically the local
flow they were exposed to. At a particular phase of the
oscillation cycle the local flow over the face of the large
ripple facing the flow (stoss side) was a current flowing
toward the large ripple crest. On the opposite side of the
large ripple (lee side), the flow detached and a recirculation
cell formed. Therefore the local flow on the lee side was
also toward the large ripple crest.
4.2.2. Experiment 11
[29] For this experiment, T = 6 s, Umax = 1.0 m/s, and d =
1.91 m. The experiment started from a flat bed. As the
experiment began, the bed was mobilized in a layer 5 to
10 sand grains thick. There was no development of smallscale bed forms and no suspended sediment was observed.
This remained unchanged for the first 5 min in the central
area of the tunnel. However at the same time, long wavy
perturbations were traveling from both tunnel ends toward
the central region. Once these long wave perturbations had
covered the entire sediment bed they rapidly increased their
height. The experiment was stopped after 35 min, when
some of the bed forms had grown enough to cover a
significant portion of the tunnel cross section. For this
experiment the Shields parameter was q = 0.62, close to
common criteria for the existence of sheet flow conditions
[Allen and Leeder, 1980; Sumer et al., 1996]. The absence
1
Auxiliary materials are available in the HTML. doi:10.1029/
2009JC005356.

7 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

Figure 5. Final bed configuration for Experiment 07


(T = 25 s, Umax = 0.25 m/s, d = 1.99 m). (top) Sonar scan,
dimensions in centimeters. (bottom) Photography showing
the general aspect of the bed, the window on the back is
60 cm wide. The resolution of the sonar data was not
enough to capture the small superimposed ripples. The
sonar scan and the photo correspond to two different
locations along the bed.
of initial ripples is another clear indication that sheet flow
regime had been initially established. A fast-forward video
of the process is included as Movie S2 as auxiliary material.
[30] The final ripple height was irregular, and the final
average ripple wavelength and height were 1.20 m and 0.19
m, respectively. All the ripple crests were round (Figure 6)
and the flow separation occurred over the lee side of the bed
form, instead of occurring at the ripple crest, as it was the
case in experiments with milder flow conditions. At the
final stage significant amounts of sediment were transported
in suspension but the bed load transport also seemed to be
important, specially over the stoss side of the ripple. No
superimposed bed forms were observed at any time during
the experiment.
4.2.3. Experiment 12
[31] For this experiment, T = 12 s, Umax = 0.5 m/s, and d =
1.91 m. For this experiment an intermediate orbital velocity,
between Experiments 07 and 11, was selected. As in the
case of Experiment 07, it started from a flat bed, and small
rolling grain ripples were observed to propagate from the
sidewalls rapidly covering the entire sediment bed. These
small ripples had straight crests, and their wavelength and
height were approximately 0.10 m and 0.01 m, respectively.
Again starting from the tunnel walls perturbations started to
grow adding three-dimensionality to the bed and increasing
the average ripple size. A fast-forward video of this process
is included as Movie S3 as auxiliary material, where several
perturbations can be seen traveling on top of the larger
features and reworking the bed. After 15 h of experiment,
the final bed configuration was highly complex and threedimensional, with the ripples continuously changing and
moving over the bed. The crests of both large and superimposed ripples were short and sharp (Figure 7). The
wavelength and height of the larger features was estimated

C12015

Figure 6. Final bed configuration for Experiment 11


(T = 6 s, Umax = 1.00 m/s, d = 1.91 m). (top) Sonar scan,
dimensions in centimeters. (bottom) Photography showing
the general aspect of the bed, the window on the back is 60 cm
wide. The sonar scan and the photo correspond to two
different locations along the bed.
to be 0.5 m and 0.1 m, respectively, putting these ripples in
the suborbital range according to the classification of
Wiberg and Harris [1994].
4.2.4. Experiment 13
[32] For this experiment, T = 8 s, Umax = 0.75 m/s, and d =
1.91 m. Initially, it was expected that these conditions
would generate smaller ripples than Experiment 12, but this
was not the case. The experiment started from a flat bed,
and very shallow ripples spontaneously appeared over the
whole sediment bed without any observable influence of the
sidewalls. The wavelength and height of these ripples were

Figure 7. Final bed configuration for Experiment 12


(T = 12 s, Umax = 0.50 m/s, d = 1.91 m). (top) Sonar scan,
dimensions in centimeters. (bottom) Photography showing
the general aspect of the bed, the window on the back is 60 cm
wide. The sonar scan and the photo correspond to two
different locations along the bed.

8 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

the experiment with velocities above 0.5 m/s, with the size
of the round-crested ripples increasing, not decreasing, as
the velocity increases. Third, the size of round-crested
ripples was very irregular along the tunnel, but all the crests
where round even the smallest ones for which the flow
restriction was minimum. Finally, bed forms with similar
characteristics have been observed by other researchers in
the laboratory [e.g., Southard et al., 1990; Arnott and
Southard, 1990; Ribberink and Al-Salem, 1994; Dumas et
al., 2005; Cummings et al., 2009] and in the field [e.g., Yang
et al., 2006], suggesting that our observations were not
artificially generated by our laboratory setup.

5. Ripple Size and Cross-Section Geometry

Figure 8. Final bed configuration for Experiment 13


(T = 8 s, Umax = 0.75 m/s, d = 1.99 m). (top) Sonar scan,
dimensions in centimeters. (bottom) Photography showing
the general aspect of the bed, the window on the back is 60 cm
wide. The sonar scan and the photo correspond to two
different locations along the bed.
0.15 m and less than 0.01 m, respectively. Their straight
crests moved half a wavelength during the oscillation cycle,
and as they increased their size their crests became more
wavy, but without any noticeable influence of the tunnel
walls. The ripples continued to grow until they reached their
final size, l = 0.8 m and h = 0.15 m. A fast-forward video of
this process is included as Movie S4 in the auxiliary
material. Smaller perturbations moving over the bed were
not observed during the experiment. The crest of the final
ripples appeared rounded similarly to Experiment 11
(Figure 8). For this case the tunnel entrances were not
found to play any role on the ripple formation.
[33] Given the heights of some of the ripples formed
under long water excursions, particularly the round-crested
ripples generated in Experiment 11, a note should be made
on the possible restrictions imposed by the cross-section
height of the LOWST to the free development of bed forms.
For example, in Experiment 11 the heights of the ripples
significantly reduced the available cross section for the
water to flow, therefore increasing the velocities over the
ripple crests and reducing them over the ripple troughs.
These changes in the water velocity have the potential to
induce variations on the shear stresses over the bed and
affect the resulting bed morphology. Similarly, the spatial
variation of the velocity could have induced groundwater
flows from the trough to the crest of the ripple, where the
pressures were lower, eventually helping to destabilize the
sediment at the crest of the ripple. Despite all these
undesired effects, which are unavoidable even in a large
facility like the LOWST, the consistency on the trends
displayed by the resulting morphology suggests that they
role was limited. First, ripples with round crests were
observed in early stages of the bed evolution before the
ripples had time to grow to their final hight. Second, the
transition to round-crested ripples can be observed for all

5.1. Performance of Existing Predictors


[34] The performance of some of the most popular ripple
size predictors [Nielsen, 1981; Mogridge et al., 1994;
Wiberg and Harris, 1994] is discussed in this section. These
predictors were presented in a companion paper, and the
complete details of each formulation can be found in the
original articles.
[35] Our new experimental data is shown in Figures 9 and
10 in terms of the dimensionless variables used by each of
the three predictors. In Figures 9a 9c and 10a 10c the
corresponding predictor equation and our experimental
results are shown. Also, the compiled literature data,
discussed in the companion paper, are shown in the background as a guide for the overall performance of each
predictor and to allow the comparison of our observations
with the ones of other researchers. Of the three predictors,
the Nielsen [1981] laboratory equation is clearly the one that
best captures the overall trend of our experiments. However,
a clear deviation from the laboratory equation is observed
for the round-crested ripples formed at high velocities. The
mobility number used by Nielsen [1981] predictor is defined
as y = U2max/(gRD50).
[36] Regarding Mogridge et al. [1994] predictor, the
constant dimensionless ripple wavelength l/l0 predicted
by Mogridge et al. [1994] for large values of the
dimensionless water excursion d/dl0 is only followed by
some of our experiments, while other experiments still
follow the orbital trend. Similarly, the overall trend of the
ripple height is to continuously increase as the dimensionless
water excursion increases, contrary to the decay predicted
by Mogridge et al. [1994]. Therefore the predictor tends to
underpredict the size of the observed ripples, specially for
long water excursions. The characteristic ripple dimensions
and near-bed water excursions (l0 and dl0, h0, and dh0)
used by Mogridge et al. [1994] are functions of the
dimensionless period parameter c = D50/(gRT2).
[37] For the case of Wiberg and Harris [1994] predictor,
the deviations are particularly dramatic, since the ripple
dimensionless wavelength and height (l/D50 and h/D50)
remain close to the orbital trend instead of following the
suborbital-anorbital trend as the predictor suggests. Similar
deviations have been reported by Traykovski et al. [1999]
and ODonoghue and Clubb [2001] for field and laboratory
experiments, respectively. Furthermore, for the same
dimensionless water excursion d/D50  8000 the ripple
wavelength was between l/D50  2000 and l/D50  5000,
depending on the water maximum orbital velocity.

9 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

Figure 9. Performance of existing ripple wavelength


predictors with the experimental data (Table 2). The data
displayed in gray corresponds to the data compiled from
the literature. (a) Nielsen [1981], (b) Mogridge et al. [1994],
(c) Wiberg and Harris [1994]. The symbols used are as
indicated in Figure 9a.

C12015

Figure 10. Performance of existing ripple height


predictors with the experimental data (Table 2). The data
displayed in gray corresponds to the data compiled from
the literature. (a) Nielsen [1981], (b) Mogridge et al. [1994],
(c) Wiberg and Harris [1994]. The symbols used are as
indicated in Figure 10a.

10 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

and l/d = 0.65 for the horizontal line in Figure 11a. For the
ripple height the equations are
h
i1
h
0:1 0:055 Umax =ws 3 1 ;
d

and h/d = 0.1 for the horizontal line in Figure 11b.


[39] The agreement between the new predictor and the
observations is clearly better than the agreement of
Mogridge et al. [1994] and Wiberg and Harris [1994]
predictors. It is also better that the agreement with Nielsen
[1981] laboratory equation. It should be emphasized that
despite being inspired by the trends found in our experimental data, the new predictor was calibrated using the
complete data set compiled from the literature, which is
displayed in light gray on the background of Figure 11.

Figure 11. Performance of the new ripple size predictor


[Pedocchi and Garca, 2009b] with the experimental data
(Table 2). The data displayed in gray corresponds to the data
compiled from the literature with Rep > 13. The symbols
used are as indicated in Figure 11a.
5.2. Performance of the New Predictor
[38] The new predictor proposed in the companion paper
is compared here with our experimental data. The new
predictors emphasize the existence of local and global
sediment transport mechanisms over a ripple bed. These
two mechanism are p
represented
by the dimensionless

particle size Rep = gRD50 D50/n and the ratio of the


maximum orbital velocity to the particle settling velocity
Umax/ws. For the sediment size and water temperatures in
the current experiments the value of Rep falls in the upper
sediment size range of the predictor (i.e., Rep > 13). The
predictive equations proposed for this sediment size range
are shown in Figure 11. For the ripple wavelength the
equations are
h
i1
l
0:65 0:050 Umax =ws 2 1 ;
d

5.3. Discussion
[40] The long water excursion experiments showed the
importance of including a second variable, other than the
water excursion, to describe the water oscillation. If attention is directed to experiments 07, 20, 12, 19, 13, and 11 in
Table 2, it can be noted that all of them had approximately
the same water excursion, d  1.9 m, but increasing
maximum orbital velocity, going from 0.25 m/s to 1 m/s.
In Table 3 a summary of these experiments sorted by
maximum orbital velocity is given. From Table 3 an
important observation can be made; initially, both ripple
wavelength and height decreased as the maximum orbital
velocity increased, but once orbital velocities exceed 0.5 m/s
the ripples began to increase their size again. The ripples
formed at velocities above 0.5 m/s had round crests and are
indicated in Figures 9, 10, and 11. As observed in the
companion paper the transition occurred at Umax/ws  25.
This corresponds to a value of the Shields parameter q  0.3.
Above this threshold the dimensionless ripple wavelength
l/d increased continuously as the dimensionless orbital
velocity Umax/ws increased. Instead, for the ripple height a
much more abrupt transition was found and the dimensionless
height jumped from h/d = 0.05 to h/d = 0.1.
[41] A first consequence of this observation is that
anorbital ripples were never observed in our experiments.
The lower limit of d/D50 for the occurrence of anorbital
ripples defined by Clifton and Dingler [1984] and Wiberg
and Harris [1994] are 5000 and 5600, respectively. The
dimensionless water excursion in the experiments discussed
here was d/D50  7600, well above these limits. Although a
reduction on the ripple size for increasing orbital velocities
was initially observed, this reduction stopped well before
ripples could be considered anorbital, and the ripple size
started to increase again. The absence of anorbital ripples is
in agreement with the observations from other researchers
shown in the background of Figure 11 and discussed in the
companion paper. According to the analysis in the companion
paper anorbital ripples would form in fine sands (Rep < 9),
which is not the case of the sand studied here.
[42] A second consequence of the above observation is
that the assumption that under high enough velocities the
ripple dimensions will decrease until the transition to flatbed
occurs is not universally valid. Nevertheless, a minimum
ripple size for a given water excursion and sediment still
exists. For the case of our experiments with d  1.9 m,

11 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

by our observations as shown in Figure 12. On the other


hand, the Lofquist [1978] criterion predicts two-dimensional
ripples for y < 21.3. This would be verified only if some
quasi two-dimensional ripples are considered as twodimensional. Similarly, the Sato [1987] criterion, which
can be approximated to the upper right quadrant of
Figure 12, is also verified if these quasi two-dimensional
ripples are grouped with the two-dimensional ones. Finally,
Vongvisessomjai [1984]ppredicts
ripples to be threedimensional if AUmax/( gRD50 D 50) > 5500, correctly
discriminating the planform configuration of our
experiments.

Figure 12. Performance of existing ripple planform


predictors with the experimental data (Table 2). The data
displayed in gray corresponds to the data compiled from
the literature. The cross indicates the results for which the
l/w ratio is larger than 1. Solid line Vongvisessomjai
[1984], dashed line Lofquist [1978], dash-dot line Carstens
et al. [1969].
they are l  0.50 m and h  0.10 m, which is observed for
Umax = 0.5 m/s. Similarly, an upper limit is given by the
orbital scale, l  0.65d  1.2 m and h  0.1d  0.19 m,
which was observed for both Umax = 0.25 m/s and Umax =
1.0 m/s. For a given water excursion the ripple size will be
between these minimum and maximum size limits depending of the maximum orbital velocity.
[43] Third, it should be noted that the increase in the size
of the bed forms, for velocities above 0.5 m/s, was associated with changes in the ripple crest shape, which became
rounder as the orbital velocity increased. This change in the
crest shape resulted in the migration of the detachment point
from the ripple crest to the lee side (see section 4.2.2). Also
as expected the increase in velocity produced an increase in
the sediment transport. All these characteristics, together
with the abrupt change in size observed in Figure 11 give a
strong indication that round-crested ripples are a different
class of bed forms. And probably they should be studied
separately from the more commonly observed sharp crested
ripples. At this point, our understanding of the bed response
in the transition from sharp crested ripples to round-crested
ripples is limited, and further experiments are needed to
fully understand how this transition occurs.

6. Planform Geometry
6.1. Performance of Existing Predictors
[44] In this section the planform geometry predictors
discussed in the companion paper are contrasted with our
experimental observations. The predictors considered were
Carstens et al. [1969], Lofquist [1978], Sato [1987], and
Vongvisessomjai [1984]. The results are displayed in
Figure 12.
[45] The Carstens et al. [1969] criterion predicts threedimensional ripples for d/D50 > 1550, which is not verified

6.2. Performance of the New Predictor and Discussion


[46] The planform geometry predictor proposed in the
companion paper is based on the observation that both
longer excursions and larger velocities tend to increase the
three-dimensionality of the ripples. Additionally, ripples
formed in fine sediments tend to present more threedimensionality than ripples formed in coarse ones. Our
new predictor takes these elements into account, giving
the following criterion for the occurrence of twodimensional ripples:
Rep > 0:06 Rew 0:5 :

Owing to variations in the water temperature between


experiments, the sediment used in our experiments did not
correspond to a unique value of Rep but rather fell into a
narrow range of Rep values, between 14 and 19. If a
characteristic Rep  16 is selected, the upper limit for
the occurrence of two-dimensional ripples is given
approximately by
Rew  7  104 :

The narrow range of Rep values involved allowed to


represent our experimental data in the (Tn/D250, UmaxD50/n)
plane. This is done in Figure 13, where the laminar-turbulent
transition Rew = 6.6*104 and the smooth-rough transition
Re* = 6 are indicated. Our predictor, indicated by the
Rew = 7*104 line, correctly discriminates between twodimensional and three-dimensional ripples. From the
constant A/D50 lines, also shown in Figure 13, it can be
observed that the Carstens et al. [1969] criterion is unable
to capture the two-dimensional to three-dimensional
transition. Lofquist [1978] criterion for Rep  16,
translates to UmaxD50/n = 74, which does not seem to
hold, either. Finally, Vongvisessomjai [1984] criterion
translates to Rew = 8.8*104, which is very close to the
value given by our predictor for this particular value of
Rep. The relations used, in these transformations among
dimensionless variables, are given in section 3.2 of the
companion paper.
[47] Further insight into the performance of the new
predictor can be gained if experimental data with similar
values of Rep obtained by other researchers are also plotted
in the (Tn/D250, UmaxD50/n) plane. In Figure 14 experiments
reporting planform geometry from Carstens et al. [1969],
Southard et al. [1990], ODonoghue and Clubb [2001],
Williams et al. [2004], and ODonoghue et al. [2006] are

12 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

Figure 13. Performance of the new ripple planform predictor with the experimental data (Table 2)
showed on the (Tn/D250, UmaxD50/n) plane. The limit for the transition from two-dimensional to threedimensional ripples is given by the Rew = 7*104 line. Several dimensionless oscillation amplitude A/D50,
constant wave Reynolds number Rew and constant shear Reynolds number Re* contours are shown. The
laminar-turbulent transition Rew = 6.6*104 and smooth-rough transition Re* = 6 are also highlighted. The
number next to the data points is the Rep value.
included. They are for Rep values between 13 and 24, close
to the ones in our experiments. Kennedy and Falcon [1965]
and Mogridge and Kamphuis [1972] also reported observations inside the range of Rep analyzed here. However,
their experiments involved lightweight, large-diameter
sediments, and the observations fall out of the range
displayed in Figure 14. Nevertheless, all of their ripples
were two-dimensional in agreement with the criterion given
by equation (5).
[48] The agreement between the literature data and our
own observations and the new criterion is excellent, and the
small deviations are due to differences in Rep, which cannot
be considered in the (Tn/D250, UmaxD50/n) plane. Experiments for which the ratio between the ripple wavelength and
the facility width were larger than 1 are highlighted in
Figure 14. Three-dimensional ripples observed in wide
facilities coexist with two-dimensional ripples found in
narrow facilities. A clear indication of the importance of the
constraint imposed by the facility width on the development of
bed form three-dimensionality.

7. Summary and Conclusions


[49] The long water excursion experiments showed that
for a given water excursion the ripple size initially reduces
as the maximum orbital velocity increases. However, after a
certain maximum orbital velocity threshold was crossed the

size of the bed forms started to increase again. These latter


ripples, produced by high velocities, have round crests. The
different geometry and behavior of round-crested ripples are
a strong indication that they belong to a different bed form
regime. The smallest ripples obtained in the long excursion
experiments were still too large to be considered to be
anorbital, and it should be concluded that anorbital ripples
are not observed for the selected sand D50 = 250 mm under
sinusoidal oscillations.
[50] For the selected sediment it was found that ripples
were two-dimensional when the wave Reynolds number
was smaller than 7*104 as predicted by our new predictor.
The LOWST 0.8 m width removed some of the limitations
found in narrower facilities for the development of bed form
three-dimensionality. The comparison with data from
narrower facilities showed that facility width restricts the
development of three-dimensional ripples, and that the
planform geometry results obtained in narrow facilities
should be used with care.
[51] The performance of the existing size and planform
geometry predictors and our new size and planform
geometry predictors was evaluated with our own experimental data. The proposed new wavelength and height
predictors were the ones performing best. Of the existing
ripple size predictors, the field equations of Nielsen [1981]
were the ones showing the best agreement with our
observations. Regarding the prediction of planform geometry,

13 of 17

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

C12015

Figure 14. Performance of the new ripple planform predictor with the new experimental data (Table 2)
and the data compiled from the literature with 13 < Rep < 24. The limit for the transition from two-dimensional
to three-dimensional ripples is given by the Rew = 7*104 line. The contours are the same of Figure 13, but
the labels have been omitted here. The number next to the data points is the Rep value.

the proposed predictor showed the best performance. The


predictor of Vongvisessomjai [1984], which for the sediment
used in our experiments gives a very similar condition to
ours, also performed very well. However, for other sediments Vongvisessomjai [1984] and our predictor would give
different results.

For A/k > 30 and Rew < 6.6*104, the flow is considered to
be laminar and the analytical solution is

Appendix A:

[54] For A/k < 30 the boundary layer concept brakes


down, since the roughness elements are of the order of
the water excursion. Furthermore, for this case the flow can
transition from rough turbulent to laminar without passing
trough a smooth regime [Jonsson, 1966]. No expression is
given for this A/k range.

Wave Friction Factor Summary

[52] In this article the friction factor for oscillatory flow fw


was computed according to Pedocchi and Garca [2009a].
A summary of its mathematical expression follows.
[53] For A/k > 30 and Rew > 6.6 * 104, the flow presents
rough to smooth turbulent transition and fw is obtained using
an iterative scheme over the following expressions:
1
1 A
p 1:9 ln
1:5 k
fw

r !
fw
Lw ;
2

A1

where
8 2
0 (
r)2 13
<1
1
Re
fw A5
w
41  exp@
Lw
...
:7:5
90 A=k 2
1
r!1 9
=
1 Rew fw
:
...
;
2:1 A=k 2

A2

1
Rew 1=4
p p :
fw
2

A3

Appendix B: Near-Bed Flow Generated


by Surface Waves
[55] In the LOWST arbitrary combinations of Umax and d
can be selected for the experiments. However, not every
combination would correspond to conditions that could be
generated by surface waves in the sea. Therefore it is of
interest to know if a realistic combination of wave heights
and water depths would generate the selected near-bed
oscillations. Clifton and Dingler [1984] presented a good
summary on this problem, and here we will just restrict the
discussion to present a dimensionless diagram computed
using Airy theory.
[56] The wavelength L is related to the water depth h and
period T by the dispersion equation [Dean and Dalrymple,

14 of 17

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

C12015

C12015

Figure B1. Wave height H as function of the near-bed maximum orbital velocity Umax and water
excursion d, made dimensionless with the water depth h. The H/h lines were computed using Airy theory
over the full plane. However, they are strictly valid below the Airy Theory line. The wave breaking limits
are also shown by the H/h = 0.78 line and the Miche [1951] criterion close to it. The conditions of our
experiments are plotted for h = 5 m, for which most of them fall under the range of validity of the Airy
Theory. Note that all the possible combinations
of H and h that would give a particular Umax and d pair
p
can be found along the constant Umax/ gd lines.
1991], which can be made dimensionless using h as a length
scale


L L0
2ph
tanh

;
h
h
L

with L0 = gT /(2p) the length of a wave with the samepperiod


2
in deep waters.
p Note that L0/h = p/2 [(d/h)/(Umax/ gh)] .
If Umax/ gh and d/h are given, L0/h can be computed
and L/h can be obtained from Equation (B1). Then the
dimensionless wave height H/h can be computed using
B2

The result of this procedure is shown in Figure B1 in


graphical form.
[57] The limits under which the Airy theory is valid are
given by [see Komar, 1976]
 
H L 2
< 32p2=3 ;
h h




H L
2ph 1
< 0:0625:
tanh
h h
L

B1



H d
2ph
sinh
:
h h
L

and
B4

And the H/h curves computed above are strictly valid under
these limits. To the right, for large values of d/h, cnoidal
wave theory should be used. To the left, second- and higherorder Stokes wave theories should be used; the results of
applying these other theories are not included here.
[58] When waves are too tall for a given water depth or
too steep, they break. This introduces a limit for the
conditions that may be found in the real world. In shallow
waters a very simple criterion, which takes neither the slope
of the beach nor the wave period into account is given by
Hb
0:78;
h

B5

where Hb is the wave height at which the wave would


break. In intermediate and deep waters the condition for
waver break is given by the Miche [1951] criterion

B3

15 of 17



Hb
L
2ph
:
0:142 tanh
h
h
L

B6

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

Both limits fall outside the area where the Airy theory is
strictly valid and they are shown in Figure B1 just to
indicate where this limitations would fall if the Airy theory
was used.
[59] The conditions of the present experiments are also
included in Figure B1, using h = 5 m as water depth for the
computations. For this water depth most of the performed
experiments fall into the range of validity of the Airy theory
and can be considered to correspond to conditions
found in
p
nature. Moving along the constant U max / gd lines
other possible combinations of wave heights and water
depths that would give the same near-bed conditions can
be found.
[60] Acknowledgments. The work presented in this article was
supported in part by the Ripples DRI of the Coastal Geosciences program,
with Thomas Drake as Program Director, of the U.S. Office of Naval
Research, grant N00014-05-1-0083, the support is gratefully acknowledged. The Large Oscillating Water-Sediment Tunnel was built with
support from the DURIP Program ONR grant N00014-01-1-0540. The
help of Sig Anderson Jr. (ELD, MN) and John Bushey (MTS, MN) with the
design and construction of the tunnel and the piston system, respectively, is
also gratefully acknowledged. The authors would like to thank J. Ezequiel
Martin for the enriching discussions during the course of this work. Thanks
to David Admiraal and Yovanni Catano-Lopera for their comments, which
helped to improve an earlier draft of this manuscript. Also thanks to Bill
Arnott and Peter Nielsen for their suggestions which helped to improve the
final form of the manuscript.

References
Allen, J. R. L., and M. R. Leeder (1980), Criteria for the instability of
upper-stage plane beds, Sedimentology, 27(2), 209 217.
Arnott, R., and J. Southard (1990), Exploratory flow-duct experiments on
combined-flow bed configurations, and some implications for interpreting
storm-event stratification, J. Sediment. Petrol., 60(2), 211 219.
Bagnold, R. (1946), Motion of waves in shallow water. Interaction between
waves and sand bottoms, Proc. R. Soc. London, Ser. A, 187, 1 15.
Carstens, M. R., F. M. Neilson, and H. D. Altinbilek (1969), Bed forms
generated in laboratory under an oscillatory flow. Analytical and experimental study, Tech. Memo. 28, Coastal Eng. Res. Cent., U.S. Army Corps
of Eng., Vicksburg, Miss.
Catano Lopera, Y. A., S. T. Demir, and M. H. Garca (2007), Self-burial
of short cylinders under oscillatory flows and combined waves plus
currents, IEEE J. Oceanic Eng., 32, 191 203.
Clifton, H. E. (1976), Wave-formed sedimentary structures: A conceptual
model, in Beach and Nearshore Sedimentation, edited by R. A. Davis Jr.
and R. L. Ethington, Spec. Publ. Soc. Econ. Paleontol. Mineral., 24,
126 148.
Clifton, H. E., and J. R. Dingler (1984), Wave-formed structures and
paleoenvironmental reconstruction, Mar. Geol., 60(1 4), 165 198.
Cummings, D. I., S. Dumas, and R. W. Dalrymple (2009), Fine-grained
versus coarse-grained wave ripples generated experimentally under largescale oscillatory flow, J. Sediment. Res., 79(2), 83 93.
Dean, R. G., and R. A. Dalrymple (1991), Water Wave Mechanics for
Engineers and Scientists, 353 pp., World Sci., Singapore.
Dingler, J. R. (1974), Wave formed ripples in nearshore sands, Ph.D. thesis,
Univ. of Calif., San Diego.
Dumas, S., R. Arnott, and J. B. Southard (2005), Experiments on
oscillatory-flow and combined-flow bed forms: Implications for interpreting
parts of the shallow-marine sedimentary record, J. Sediment. Res., 75(3),
501 513.
Faraci, C., and E. Foti (2001), Evolution of small scale regular patterns
generated by waves propagating over a sandy bottom, Phys. Fluids,
13(6), 1624 1634.
Garca, M. H. (2008), Sediment transport and morphodynamics, in
Sedimentation Engineering: Process, Measurements, Modeling and
Practice, ASCE Manual Rep. Eng. Pract., vol. 110, edited by M. H.
Garca, pp. 21 164, Am. Soc. of Civ. Eng., Reston, Va.
Hanes, D., V. Alymov, Y. Chang, and C. Jette (2001), Wave-formed
sand ripples at Duck, North Carolina, J. Geophys. Res., 106(C10),
22,575 22,592.
Hanson, H., and B. Camenen (2007), Closed form solution for threshold
velocity for initiation of sediment motion under waves, in Coastal
sediments 07: Proceedings of the Sixth International Symposium on
Coastal Engineering and Science of Coastal Sediment Processes, May

C12015

13 17, 2007, New Orleans, Louisiana, pp. 15 27, Am. Soc. of Civ.
Eng., Reston, Va.
Inman, D. L. (1957), Wave-generated ripples in nearshore sands, Tech.
Memo. 100, U.S. Army Corps of Eng., Washington, D. C.
Jensen, B. L. (1989), Experimental investigation of turbulent oscillatory
boundary layers, Ser. Pap. 45, Tech. Univ. of Denmark, Lyngby.
Jonsson, I. (1966), Wave boundary layers and friction factors, Proceedings
of Tenth Conference on Coastal Engineering, September, 1966, Tokyo,
Japan, vol. 1, pp. , 127 148, Am. Soc. of Civ. Eng., New York.
Kennedy, J. F., and M. Falcon (1965), Wave generated sediment ripples,
Rep. 86, Hydrodyn. Lab., Mass. Inst. of Technol., Cambridge.
Komar, P. D. (1976), Beach Processes and Sedimentation, Prentice Hall,
Upper Saddle River, N. J.
Lofquist, K. E. B. (1978), Sand ripple growth in an oscillatory-flow water
tunnel, Tech. Pap. 78-5, Coastal Eng. Res. Cent., U.S. Army Corps of
Eng., Vicksburg, Miss.
Miche, M. (1951), Le pouvoir reflechissant des ouvrages maritimes exposes
a laction de la houle, Ann. Ponts Chaussess, 121, 285 319.
Miller, M., and P. Komar (1980a), Oscillation sand ripples generated by
laboratory apparatus, J. Sediment. Petrol., 50(1), 173 182.
Miller, M., and P. Komar (1980b), A field investigation of the relationship
between oscillation ripple spacing and the near-bottom water orbital
motions, J. Sediment. Petrol., 50(1), 183 191.
Mogridge, G. R., and J. W. Kamphuis (1972), Experiments on bed form
generation by wave action, paper presented at 13th International
Conference on Coastal Engineering, Int. Assoc. of Hydro-Environ.
Eng., Vancouver, B. C., Canada.
Mogridge, G. R., M. Davies, and D. Willis (1994), Geometry prediction for
wave-generated bedforms, Coastal Eng., 22(3 4), 255 286.
Nielsen, P. (1981), Dynamics and geometry of wave-generated ripples,
J. Geophys. Res., 86(C7), 6467 6472.
ODonoghue, T., and G. Clubb (2001), Sand ripples generated by regular
oscillatory flow, Coastal Eng., 44(2), 101 115.
ODonoghue, T., J. Doucette, J. van der Werf, and J. Ribberink (2006), The
dimensions of sand ripples in full-scale oscillatory flows, Coastal Eng.,
53(12), 997 1012.
Pedocchi, F., and M. H. Garca (2009a), Friction coefficient for oscillatory
flow: The rough-smooth turbulent transition, J. Hydraul. Res., 47(4),
438 444.
Pedocchi, F., and M. H. Garca (2009b), Ripple morphology under
oscillatory flow: 1. Prediction, J. Geophys. Res., doi:10.1029/
2009JC005354, in press.
Ribberink, J., and A. Al-Salem (1994), Sediment transport in oscillatory
boundary layers in cases of rippled beds and sheet flow, J. Geophys. Res.,
99(C6), 12,707 12,727.
Sato, S. (1987), Oscillatory boundary layer flow and sand movement over
ripples, Ph.D. thesis, Univ. of Tokyo, Tokyo.
Shen, H. W., A. S. Harrison, and W. J. Mellema (1978), Temperature and
Missouri River stages near Omaha, J. Hydraul. Div., 104, 1 20.
Soulsby, R. L., and R. J. S. Whitehouse (1997), Threshold of sediment
motion in coastal environments, in Pacific Coasts and Ports 97,
Proceedings of 13th Australasian Coastal & Ocean Engineering
Conference, Christchurch, pp. 149 154, Cent. for Adv. Eng., Univ. of
Canterbury, Christchurch, New Zealand.
Southard, J. B. (1991), Experimental determination of bed-form stability,
Annu. Rev. Earth Planet. Sci., 19, 423 455.
Southard, J. B., J. Lambie, D. Federico, H. Pile, and C. Weidman (1990),
Experiments on bed configurations in fine sands under bidirectional
purely oscillatory flow, and the origin of hummocky cross-stratification,
J. Sediment. Petrol., 60(1), 1 17.
Sumer, B. M., A. Kozakiewicz, J. Fredsoe, and R. Deigaard (1996),
Velocity and concentration profiles in sheet-flow layer of movable bed,
J. Hydraul. Eng., 122(10), 549 558.
Traykovski, P., A. Hay, J. Irish, and J. Lynch (1999), Geometry, migration,
and evolution of wave orbital ripples at leo-15, J. Geophys. Res.,
104(C1), 1505 1524.
Vanoni, V. A. (1974), Factors determining bed forms of alluvial streams,
J. Hydraul. Div., 100, 363 377.
Vongvisessomjai, S. (1984), Oscillatory ripple geometry, J. Hydraul. Eng.,
110, 247 266.
Wiberg, P., and C. Harris (1994), Ripple geometry in wave-dominated
environments, J. Geophys. Res., 99(C1), 775 789.
Williams, J., P. Bell, P. Thorne, N. Metje, and L. Coates (2004), Measurement and prediction of wave-generated suborbital ripples, J. Geophys.
Res., 109, C02004, doi:10.1029/2003JC001882.
Xu, J. (2005), Observations of plan-view sand ripple behavior and spectral
wave climate on the inner shelf of San Pedro Bay, California, Cont. Shelf
Res., 25(3), 373 396.
Yalin, S., and R. C. H. Russell (1963), Similarity in sediment transport due
to waves, in Proceedings of eighth Conference on Coastal Engineering

16 of 17

C12015

PEDOCCHI AND GARCIA: RIPPLE MORPHOLOGY UNDER OSCILLATORY FLOW, 2

Mexico City, Mexico, November 1962, edited by J. W. Johnson,


pp. 1123 1142, Counc. on Wave Res., Eng. Found., Richmond, Calif.
Yang, B., R. W. Dalrymple, and S. Chun (2006), The significance of
hummocky cross-stratification (hcs) wavelengths: Evidence from an
open-coast tidal flat, South Korea, J. Sediment. Res., 76, 2 8.

C12015

Champaign, 205 N. Mathews Ave., Office 2535b, Urbana, IL 61801, USA.


(mhgarcia@uiuc.edu)
F. Pedocchi, Instituto de Mecanica de los Fluidos e Ingeniera Ambiental,
Faculatad de Ingeniera, Universidad de la Republica, Julio Herrera y
Reissig 565, CP 11300, Montevideo, Uruguay. (kiko@fing.edu.uy)



M. H. Garca, Ven Te Chow Hydrosystems Laboratory, Department of


Civil and Environmental Engineering, University of Illinois at Urbana-

17 of 17

Das könnte Ihnen auch gefallen