Sei sulla pagina 1di 19

Direct measurements of the bottom friction factor

beneath surface gravity waves


Hamid Mirfenderesk
a
, Ian R. Young
b,
*
a
School of Civil Engineering, University College of New South Wales, Canberra, ACT 2600, Australia
b
Swinburne University of Technology, P.O. Box 218, Hawthorn, Vic 3122, Australia
Abstract
This paper describes a detailed experimental study of the laminar and turbulent oscillatory boundary layer developed under the action of
waves for the purpose of evaluation of bottom energy dissipation. The measurements were conducted on hydraulically smooth and immobile
rough beds and under both monochromatic conditions and a spectrum of waves in a 50 m long wave ume. Bottom friction was measured in
two different ways: (i) by a ush mounted shear plate, developed as part of the project and set at the bottom of the ume, (ii) by using the
momentum integral equation and the measured velocity prole within the boundary layer. The velocity prole within the boundary layer was
measured using a Laser Doppler Anemometer. The roughness geometry was chosen to simulate both rough at and rippled beds. With
respect to the dissipation under a spectrum of waves, the results show that a spectrum of waves and a monochromatic wave whose maximum
bottom velocity is 1.88 times the root mean square of the bed velocity of the spectrum, dissipate the same amount of energy. As a
consequence, measurements of energy dissipation for monochromatic waves can be related to spectral conditions. The results for
monochromatic waves were found to be in good agreement with the existing formulae for the bottom friction factor, which have mainly been
based on experimental results in water tunnels or over oscillating trays. Therefore, it was concluded that the secondary effects due to the
nonuniformity of the bed velocity in the horizontal direction, which can be observed under real waves, does not have substantial inuence on
the bed friction factor. In addition, it was concluded that the nonlinearity of waves in nite depth water does not signicantly affect the
results.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Bottom friction; Finite depth; Water waves; Friction factor
1. Introduction
Waves propagating over a bed lose energy due to
interaction with the bed. One of the most important
dissipative mechanisms is associated with bed friction,
which causes a thin boundary layer to develop above the
ocean bed. The ow in the boundary layer is usually
turbulent in eld conditions. This turbulence causes
turbulent shear stresses which are associated with energy
dissipation. Putnam and Jonsson [1] considered monochro-
matic waves and expressed the rate of energy dissipation, E;
as the work done by the wave orbital velocity against shear
stress at the bottom
E t
0
u
b
1
where u
b
is the horizontal water particle velocity just outside
the boundary layer, which can be derived from potential
theory for waves and t
0
is the bottom shear stress, which
depends on wave and bottom characteristics.
In the case of a wave spectrum, Hasselmann and Collins
[2] showed that waves lose energy by bottom friction
according to:
S
bf
k 2kt
0
u
bk
l 2
where u
bk
is the orbital bottom velocity of the wave
component with wave number k; kl denotes the ensemble
mean and S
bf
k is the time rate of energy density loss at
wave number k: It should be noted that linear wave theory is
assumed in the development of the spectral formulation in
Eq. (2).
A review of previous research carried out on the bed
oscillatory boundary layer and resulting energy dissipation
shows that these studies can be categorized into two major
groups, (i) those dealing with monochromatic waves and (ii)
those dealing with spectral conditions.
The bed boundary layer under monochromatic waves
has been studied both analytically and experimentally.
0141-1187/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apor.2004.02.002
Applied Ocean Research 25 (2003) 269287
www.elsevier.com/locate/apor
* Corresponding author.
E-mail address: iyoung@swin.edu.au (I.R. Young).
Analytical studies in this eld are mainly based on solving
the linearized momentum equation for the ow
r

t
u 2u
b

t
z
3
where r is the density of water, z is the vertical coordinate
measured upwards from the bed, t is the shear stress (t t
0
at z 0) and u is the instantaneous velocity. Eq. (3)
represents one equation in two unknowns, i.e. u and t: Either
a drag law model or an eddy viscosity model can be used to
parameterize t in Eq. (3) and hence obtain closure.
In the drag law model, the maximum bed shear stress,
t
0max
is expressed as
t
0max

1
2
rf
w
U
2
b
4
where f
w
is a friction factor and U
b
is the maximum near bed
velocity.
In the eddy viscosity model the shear stress is
parameterized in terms of the velocity gradient in the
bottom boundary layer:
t n
t
u
z
5
where n
t
is the turbulent eddy-viscosity.
Nearly all the analytical solutions use an eddy viscosity
model to close the set of equations. The main difference
between these solutions is in the way in which they model
the eddy viscosity. Some authors have used time-variant
eddy viscosity, such as Trowbridge and Madsen [3,4].
Others have used time invariant eddy viscosity, such as
Kajiura [5], Grant and Madsen [6], Brevik [7], Myrhaug [8],
You et al. [9] and Hsu and Ou [10]. Time invariant models
differ in the way they prescribe the eddy viscosity
distribution within the boundary layer. Also, some authors
have used higher order turbulent closure schemes, (k 21
closure), for modeling the turbulent eld within the wave
boundary layer, such as Justesen [11] and Aydin and Shuto
[12].
Experimental studies of the bottom boundary layer and
bottom dissipation with monochromatic waves, fall into
three groups depending on the laboratory devices used.
They are: (i) oscillating trays, (ii) oscillating ow tunnels
and (iii) wave umes. Bagnold [13], Kalkanis [14,59] and
Sleath [15] used an oscillating tray, while Jonsson [16],
Carstens et al. [17], Kamphuis [18], Johnsson and Carlsen
[19], Lofquist [20], Van Dorn [21], Sato et al. [22,60],
Sumer et al. [23], Jenssen et al. [58], Hino et al. [24] and
Sawamoto and Sato [25] carried out their tests in oscillating
ow tunnels. Savage [26], Inman and Bowen [27],
Horikawa and Watanabe [28], Van Doorn [29], Brevik
[30,34] and Simons et al. [31] performed tests in wave
umes their tests were carried out in oscillating ow tunnels
[1625,58]. Tests were performed in wave umes [2631].
The experimental works can also be categorized based on
the way that the bed shear stress is obtained. A potentially
reliable method, used by Kamphuis [18], involved the direct
measurement of the shear force exerted on a segment of
the bed. Another method widely used by other authors, e.g.
[32,33], is the measurement of the velocity eld within the
boundary layer and the determination of the bed shear stress
by numerically solving the momentum Eq. (3). A brief
examination of the literature shows that the bulk of the
experimental studies have been performed in oscillating
ow tunnels with only a small number in wave umes.
These ume measurements are rather old and generally lack
velocity measurements close to the bed. Table A1 shows
various expressions for the friction factor suggested by
different authors for both mobile and immobile at and
rippled beds.
As can be seen in Table A1, (Appendix A), there are
numerous empirical relationships for the estimation of
bottom friction under waves. In Fig. 1 some of the best
known expressions for the bottom friction factor coefcient
are compared. It can be seen that all the curves show a
similar trend and cover a narrow region in the friction
factor-relative roughness plane, although the logarithmic
scaling of the graph masks signicant differences in
magnitude (factor of 2).
The relationships shown in Fig. 1 were obtained using a
variety of methods, both experimental and theoretical. The
available experimental data, are shown in Fig. 2.
An examination of the data and the summary curves in
Fig. 2 shows some interesting features. It can be seen that
the values of friction factor shown by the open circles are
much smaller in magnitude than other results. These data
were obtained by Sleath [33] through measurement of
the Reynolds stress within the boundary layer. The remainder
of the data were obtained either through the momentum
integral equation, direct measurement of the shear stress or
the decay of wave height. This signicant difference shows
that, unlike in unidirectional ow, the Reynolds stress
method does not give realistic values for bottom shear
stress. Sleath [33] cites the nite value of the correlation
between the stream-wise and vertical components of the bed
velocity as a reason for this discrepancy.
The second feature of note is that, generally, friction
factors obtained in a wave ume, e.g. [30,31,34], are larger
than those obtained from water tunnels or oscillating trays.
Brevik [7] attributed this difference to the inherent
inaccuracy in the wave attenuation method for measuring
the bottom friction factor, particularly when the slope of
the energy line is small and the side wall effects are
signicant. Simons et al. [31] cited the uncertainty in
velocity measurements as a reason for those data points
which show larger friction factors than the empirical
relationships. They also pointed out the low Reynolds
Numbers of these points and argued that they might be in the
transitional smooth-to-rough regime where the Nikuradse
roughness may be different from that determined in tests in a
current alone.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 270
The third feature in Fig. 2 is that for small values of
A
b
=k
s
; the friction factor continues to increase. Malarkey
and Davies [35] have found similar results with an
inviscid model. These results contradict the assumptions
made by Kajiura and Jonsson [5,36] that the friction
factor approaches a constant value as A
b
=k
s
becomes
small.
Studies of bottom dissipation under a wave spectrum can
also be categorized into two main groups, (i) analytical and
(ii) experimental. The analytical works can be further
divided into two groups, (i) drag law models, investigated
by Hasselmann and Collins [2] and Collins [37] and (ii)
eddy viscosity models, investigated by Madsen et al. [38]
and Weber [39].
Fig. 2. Summary of experimental friction factor data for oscillatory boundary layers.
Fig. 1. Friction factor expressions proposed by various authors for oscillatory boundary layers. In the gure, A
b
is the particle excursion at the outer edge of the
bottom boundary layer, k
s
is an equivalent sand grain roughness and f
w
is the friction factor.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 271
Experimental studies under spectral conditions have
mainly been conducted in the eld. Bretschneider and Reid
[40] made the rst such eld study, followed by Iwagaki and
Kakinuma [41], Ifuku and Kakinuma [42], Lambrakos [43],
Myrhaug et al. [44] and Young and Gorman [45]. Simons
et al. [46] conducted their experiments in a wave basin. The
eld measurements either determined the decay of the wave
spectrum or the velocity prole near the bed. The former
have a fundamental problem because there are invariably
other parameters active, in addition to bottom friction,
adding uncertainty to the results. The accuracy of the results
of the second group is also questionable because the
velocity is usually measured only at a limited number of
points at rather uncertain levels above the bed. In addition,
there is usually no accurate information about the bed forms.
The friction coefcients, mainly or partly based on eld
measurement, (Table 1), vary signicantly. In addition, their
accuracies have been questioned as a result of measure-
ments made by other authors, such as Bouws and Komen
[45], and Young and Gorman [47].
Table 1 summarizes relationships developed for spectral
conditions. The friction factors in this table can be used in
expression (6) proposed by Luo and Monbaliu [48] and
Weber [39,49] to obtain the bottom friction dissipation,
S
bf
2c
f
k
sinh 2kh
Ff ; w 6
where k is the wave number of the spectral component, h is
the water depth, w is the direction of propagation of the
spectral component with frequency f ; Ff ; w is the two
dimensional frequency spectrum, S
bf
is the time rate of
energy density loss at wave number k and c
f
is a dissipation
coefcient, the value of which is given in Table 1 for
different methods.
The terms in Table 1 require explanation:
c
J
is the friction factor proposed in the JONSWAP study.
c
DHC
is the friction factor obtained by Hasselmann and
Collins [2], using a drag law model. In this expression i
and j denote the orthogonal components of the bottom
velocity. The value of c
f
in this expression was obtained
from the decay of a wave spectrum in the Gulf of
Mexico.
c
DC
is the friction factor proposed by Collins [37] which
is an approximation of the Hasselmann and Collins [2]
expression. The main difference is that in Ref. [37] the
bracketed term in c
DHC
is approximated by the root-
mean-square bed velocity of the spectrum.
c
DM
is the friction factor proposed by Madsen et al. [38],
obtained from an eddy viscosity model. In this model a
spectrum of waves is approximated by an equivalent
monochromatic wave. Hence, the value of f
w
can be
obtained from existing friction factor diagrams for
monochromatic waves. The velocity term in this
expression is the root-mean-square bed velocity obtained
from the spectrum, u
br

2
__
S
ub
f ; wdf dw
_
and S
ub
is the near-bottom orbital velocity spectrum.
c
E
is the friction factor obtained by Weber [49], using
a one layer eddy viscosity model. The function T
k
; as
stated in her paper, expresses our ignorance of the
stress and the velocity prole in the boundary layer.
Therefore it is an unknown function. However, u
f
and
T
k
can be obtained by an iterative method, detailed in
her paper, provided the roughness height is given. In
this expression k
N
represents the roughness height
which was proposed to be equal to 4 cm.
A review of the literature reveals that analytical
methods are far from being applicable in engineering
problems. Also there are insufcient experimental data
under spectral conditions to provide certain and reliable
empirical expressions. In addition, the limited experimental
data under monochromatic waves, which form the basis for
the vast majority of the existing empirical relations, e.g.
[18,36], have been obtained from oscillating ow tunnels
rather than under more realistic wave conditions. Under
real waves the horizontal ow at the bed is nonuniform and
can not be exactly simulated in a water tunnel.
In addition, Asano and Iwagaki [50] state that the wave
nonlinearity produces signicant effects on near bottom
water particle velocity and the resultant bed shear stress.
They showed that the friction factors estimated by nonlinear
calculations can possibly be several times as large as those
proposed by existing expressions. Real waves, including
those generated in wave umes are usually non-linear to
some extent, while oscillating water tunnels and oscillating
trays typically generate a purely monochromatic oscillatory
boundary layer. Further, a review of the literature shows that
there are little experimental data on bed friction under a
spectrum of waves.
In view of the above, two questions are raised. (i) How
realistically can water tunnels and oscillating trays model
real waves in regard to bed friction? (ii) How can the
existing body of knowledge for monochromatic waves be
extended to a spectrum of waves? To generate more
experimental data to compare the friction factors obtained
in wave umes with those in oscillatory water tunnels, to
examine the effects that nonlinearity has on the friction
factor, and to nd a way to apply the considerable body
of knowledge for monochromatic wave conditions to
Table 1
Proposed dissipation coefcients for use in Eq. (6), after [48]
Formulation References Coefcients
c
J
2c
f
=g [57] c
f
0:038 m
2
s
23
c
DHC
2c
f
{d
ij
kul ku
ij
=ul} [2] c
f
0:015
c
DC
2c
f
ku
2
l
1=2
[37] c
f
0:015
c
DM
f
w
u
br
[38] f
w
or k
N
c
E
u
f
T
k
j
0
T
p
k
j [49] k
N
4 cm
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 272
a spectrum of waves, a series of tests in a wave ume have
been performed, and the results are presented here.
The arrangement of the paper is as follows. In Section 2
the experimental design and instrumentation is considered,
together with the accuracy of each of the systems. This is
followed in Section 3 by the determination of the bed
roughness for each of the bed conditions examined. Section
4 provides the results for hydraulically smooth and rough
beds under the action of monochromatic and spectral wave
conditions. Finally, conclusions are presented in Section 5.
2. Instrumentation and experimental design
To assess the wave energy dissipation due to bottom
friction, direct observations have been utilized in this study.
The basic parameters measured were:
Total frictional uid force on the bottom, using a shear
plate.
Velocity eld within the bottom boundary layer, using
Laser Doppler Anemometry (LDA) and Acoustic
Doppler Velocimetery (ADV) techniques.
Water surface elevation, using three capacitance wave
gauges.
Figs. 3 and 4 illustrate the experimental apparatus used.
As can be seen in these gures, the main parts of the
system are the: ume, wave generator, LDA, shear plate,
ADV and wave gauges. A description of each of these
elements is given below. In addition, an extensive
uncertainty analysis was performed for each of the
instrumentation systems, to determine the accuracy and
sampling variability [51]. The results of these analyses are
also presented below.
Flume. The tests were carried out in a wave ume with
dimensions: length 50 m, width 2:3 m, depth 2:0 m.
The sides and bottom of the ume were made of reinforced
concrete, with the exception of the working section, where
one of the side walls was of 12 mm plate glass for a length of
10 m. At one end of the ume there was a vertical, plane
wave generator, which was able to generate monochromatic
and spectral waves. At the other end of the ume provision
was made for dissipating the wave energy. The wave
dissipater consisted of 14 layers of steel mesh and four
layers of graded gravel. Tests indicated that the average
reection coefcient in the frequency range of interest, 0.2
0.7 Hz, was less than 4%, which was considered acceptable
for the present application. Extensive testing was conducted
to determine the transfer function for the wave generator/
ume system. As a result, pre-dened monochromatic and
spectral wave conditions could be produced.
Tests to determine the equivalent sand roughness, for
each of the bed materials used, were carried out in a second
smaller ume. This ume was 10 m long, 0.5 m wide and
0.6 m deep. The walls consisted of glass and the bottom was
covered by the bed roughness material being tested. The
ow to this ume was provided by a hydraulic pump which
was regulated using a gate valve. The ume also had
provision for regulation of the slope of the bed using a
hydraulic jack. To improve the ow uniformity, a ow
straightener made of wire mesh with 5 mm diameter cells
was inserted at the inlet of the ume.
Wave generator. To simulate realistic wave conditions in
the ume, a piston-type wave generator was used. This wave
generator was controlled numerically and generated either
monochromatic waves or a spectrum of waves by the
reciprocating motion of a vertical paddle with dimensions
2:3 2 m. The paddle was driven by a negative feed-back
type electro-hydraulic servo-system.
Shear plate. The shear stress was measured directly using
a bottom mounted shear plate. The principle of this
measurement technique is to detach a piece of the bed and
allow it to move freely under the shear stress exerted by
water owing over the bed. Figs. 5 and 6 show the form and
dimensions of the shear plate used.
The shear plate measured 45 cm in the across-ume
direction and 26 cm in the along-ume (wave propagation)
direction. The plate was surrounded by a section of
Fig. 3. Longitudinal section of the wave ume, showing the experimental
apparatus.
Fig. 4. Cross-section through the wave ume used for the measurements.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 273
the bottom which extended 3 m in both directions. This
section of the bed, and the plate itself were covered in
artical roughness elements. A number of initial tests were
performed to ensure the measured shear stress was
independent of the size of this region.
The shear plate itself, consisted of two major parts. The
rst part, the body, consisted of two parallel steel plates. The
plate at the top was ush with the bottom. Both the plate and
the surrounding bottom were covered with roughness
elements. The plate was supported at its corners and at the
middle by ve slender legs. The top plate and supports
functioned as a single degree of freedom structure. The legs
were rigidly joined to both the top and bottom plates. The
structural stiffness was small enough to allow the shear plate
to deect under the very small shear forces exerted on the
top plate, but was still large enough to justify a static
calibration of the plate.
The second component of the system was an eddy current
proximity sensor, which measured the distance between the
sensor head and a conductor, or target element, attached to
the shear plate. The output voltage from the sensor can be
shown to be linearly related to the distance between the
target and the sensor head. In this manner, the output
voltage time history yielded the displacement of the shear
plate. The relationship between the applied force and the
displacement of the plate was linear and determined through
a static calibration. Independent tests were performed to
access the dynamic response of the system. These tests
conrmed that within the frequency range of interest
(0.53 Hz), the transfer function was unity, and that the
static calibration was valid. This is not surprising as typical
amplitudes of displacement of the plate were only 0.1 mm.
The error analysis determined that the accuracy of the shear
plate system was approximately ^2:5%:
Laser Doppler Anemometer (LDA). The velocity
measurements in the wave ume were carried out with a
two-component Dantec LDA operating in backward-
scattering differential mode. Use of the LDA technique
enables the measurement of the desired velocity component
with great accuracy and without intrusion of a probe into the
ow. Light scattered from particles in the ow is collected
by a lens, focused on the receiving optical ber and
transmitted to a photo multiplier. The photo multiplier
converts the light to an analogue signal. The analogue signal
is processed using a DANTEC Burst Spectrum Analyzer
(BSA). In order to measure the velocity prole the optical
transducer was moved using a traversing system. This
system allows the optical transducer to traverse 460 mm
along the ume, 495 mm vertically and 449 mm across the
ume. The traversing system was under computer control,
and could position the measurement volume to an accuracy
of 0.05 mm in all three co-ordinate directions. The
traversing system also features a tilting facility, so that by
tilting the laser probe and hence the laser beams,
measurements very near the bottom could be made. The
error analysis determined the accuracy of velocity measure-
ments made with the system, operating with the level of ow
seeding available, to be approximately ^3%:
Acoustic Doppler Velocimeter (ADV). A Sontek ADV
was used to measure the velocity prole within the
boundary layer induced by the unidirectional ow in the
smaller ume. The purpose of these measurements was to
determine the roughness of the different bed surfaces used.
The ADV is a non-contact acoustic-sensing instrument for
the measurement of water particle velocities. In a similar
fashion to the LDA, the ADV measures water ow via the
Doppler shift of the acoustic signal scattered from particles
in the ow. The instrument has a single transmit transducer
and three receive transducers, focused on a measuring
volume 10 cm from the transmit transducer. In this manner,
the three orthogonal components of ow can be determined.
Wave gauges. Three WG-30 capacitance wave gauges,
supplied by Brackner Co., Canada, were used to measure the
water level. The sensing element for each of the probes was
a 1mm Teon coated copper wire stretched vertically
between the two ends of an aluminum frame. The frequency
response of the system was unity at the wave frequencies
being considered and the gauges proved insensitive to
variations in water quality (due to the introduction of LDA
seeding material). The error analysis indicated the resol-
ution of the system was less than ^1 mm or less than 1% of
the wave height being considered.
Data acquisition. Data acquisition was carried out using
a system of three IBM 486 computers. The wave maker,
capacitance gauges and shear plate were controlled via the
A/D and D/A channels of an Analog Devices RTI-815F
card. A base sampling rate of 25 Hz was chosen for this data
Fig. 6. Plan view of the shear plate.
Fig. 5. Cross-section through the shear plate in the along-ume direction.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 274
acquisition. In addition, a 25 Hz synchronization signal was
sent to both the LDA and ADV systems. The data rate for
the LDA system is variable, since data are acquired
whenever suitable light scattering particles enter the
measurement volume. This rate is dependent on the quantity
and size of scatterers within the ow. Seeding material
(titanium dioxide) was regularly introduced into the ow to
obtain suitable results. Sampling rates varied between
experiments, but were typically of the order of 1 KHz.
The 25 Hz synchronization signal from the RTI card was
stored along with the sampling time of each LDA velocity
measurement, for later correlation. The same synchroniza-
tion signal was also passed to the ADV system, which was
controlled through its own digital signal processor.
The external 25 Hz signal was, however, used as the trigger
for sampling.
3. Bed roughness conditions
Three different bed conditions were used to investigate
the bottom friction dissipation under waves in the wave
ume. They were: (i) sand grains, (ii) spherical aggregate
with diameters between 4 and 9.5 mm and (iii) rib roughness
elements with a triangular cross-section. The rib elements
were placed 6.5 cm apart in the ow direction (Fig. 7). As
described earlier, the area surrounding the shear plate was
also covered in these respective roughness elements.
Identical bed conditions were also produced in the smaller
ume under unidirectional ow for the determination of the
equivalent sand grain roughness for each surface.
The sand grain bed consisted of angular silicon carbide
aggregate with a diameter of 197 mm. The aggregate was
glued on a backing made from woven cloth. Considering the
fact that the equivalent Nikuradse sand grain roughness is
actually the diameter of the uniform sand grains, an
equivalent sand roughness of approximately 0.2 mm was
expected for the sand grains. To verify the accuracy of this
prediction, the velocity prole within the boundary layer of
a unidirectional ow over the sand grains was measured. For
this purpose the ADV was traversed in a vertical direction
with measurements being made at 0.1 mm increments.
The equivalent sand grain roughness can be obtained by
extrapolation of the logarithmic velocity prole. These
experimental results gave the equivalent sand grain rough-
ness as 0.16 mm, which compares well with the value of
0.2 mm obtained from the physical size of the sand grains.
The same tests performed on the sand grains, were
repeated on a bed covered with the spherical aggregate. The
velocity prole measurements gave an equivalent sand grain
roughness equal to 20 mm for the surface covered with
the aggregate, which compares well with the empirical
relations offered by other authors, e.g. [18]. To obtain friction
factor results in a fully turbulent rough ow, tests were
carried out on a surface covered with the transverse triangular
ribs, Fig. 7.
To relate the rib roughness geometry to an equivalent
sand grain size, the same tests conducted on the above
roughness elements were repeated on this surface. The
experimental results gave an equivalent sand grain rough-
ness equal to 50 mm, which compared well with existing
empirical relationships, such as Dirling, (sighted by
Hodge [52]).
4. Bottom friction measurements
The objective of the measurements presented below is
the determination of the bottom friction factor under a
progressing wave train. Two different measurement tech-
niques were used. For the sand grain material, spurious
forces exerted on the edges of the shear plate (see below)
made this an unreliable measurement technique. For this
case, however, a well formed boundary layer develops and
the friction factor can be obtained from measurements of the
velocity prole and solution of the momentum Eq. (3).
Conversely, for the spherical aggregate and rib roughness
elements, determination of the velocity prole is compli-
cated by roughness elements protruding into the thin
boundary layer. For these cases, however, the shear plate
provides a direct and reliable method of determination of the
friction factor (edge forces are smaller, compared to the
shear stress). As a result, solution of the momentum Eq. (3)
was used to determine f
w
for the sand grain roughness and
direct measurements with the shear plate were used for the
other roughness elements.
4.1. Monochromatic waves
Experiments were performed in the wave ume for both
regular (monochromatic) and irregular (spectral) wave
trains. Irregular waves are considered in Section 4.2.
Tests includedmeasurements of instantaneous longitudinal
velocities throughout the boundary layer using the LDA(sand
grain roughness), instantaneous bed shear stress using the
ush-mounted shear plate (aggregate and rib roughness) and
instantaneous water surface elevation. These values of
velocity, shear stress and water surface elevation were
decomposed into periodic and random components. The
periodic components were determined as the ensemble
average of the respective time series. An ensemble average
is an average of many realizations of the same phenomenon.
Due to the periodicity of the ow, for ensemble averaging,
everycycle has beenregardedas one realizationof the process.
The uctuating component was obtained as the difference
between the ensemble average and instantaneous values. Fig. 7. Geometry of the rib roughness elements.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 275
The shear plate has the obvious advantage that it provides
a direct measure of the shear stress exerted on the bed. Due
to the nite height of the wave train, however, a longitudinal
pressure gradient will be present. This pressure gradient will
exert a contaminating force on the edge of the shear plate,
and its magnitude must be carefully assessed to obtain
reliable results [18]. The pressure gradiant at the bed can be
estimated from the relationship r u
b
=t 2p=x; where
u
b
is the orbital velocity just outside the oscillatory
boundary layer and p is the pressure [31]. The bed velocity,
u
b
t, was measured directly using the LDA system. It was
assumed that the pressure varied linearly over the length of
the plate. As the shear plate was 26 cm long and the wave
length was typically 4 m, this seemed a reasonable
assumption. Calculations using linear wave theory con-
rmed the accuracy of the assumption (error is less than
0.5%). The pressure difference across the plate, Dp was
assumed to act on the cross sectional area of the edge of
the plate. This included both the thickness of the plate
itself, plus the roughness elements attached to the plate. For
instance, in the case of rib roughness, the base plate is 1 mm
thick whilst the shear plate is covered with three ribs each
with a height of about 13.6 mm. Thus the total height upon
which the pressure acts is 14.6 mm.
Calculations were performed for each of the experiments
conducted. The effect of the pressure gradient in the case of
a hydraulically smooth bed is quite signicant and can, in
fact, exceed the magnitude of the shear force exerted on the
plate. For this reason, the shear plate data were not used for
these cases. In the case of the rib roughness elements, the
effect of the pressure gradient was relatively small, less than
10% of the total shear stress, although in the case of the
spherical aggregates its effect was as large as 40% of the
total shear stress. For both of these cases, the measured
shear stresses were corrected for the additional edge force
effects, as described above. In the following section the
experimental procedures for hydraulically smooth and
rough beds are separately discussed.
4.1.1. Hydraulically Smooth bed
The rst series of tests were performed on an hydraulically
smooth bed for six different wave conditions. Since some
theoretical solutions exist for the monochromatic oscillatory
boundary layer on hydraulically smooth beds, the results of
this part of the experiment were compared with these
theoretical values to test the reliability of the whole
experimental set-up. The wave conditions and the resulting
values of f
w
are presented in Table 2. The water depth for
the various experiments performed ranged between 0.44
and 0.60 m (values reported in the attached tables),
although the depth was constant throughout each individual
experiment. Prior to each measurement the water was kept
oscillating for at least 5 min or the passage of approxi-
mately 150 waves, to ensure steady state conditions in the
ume.
The oscillatory stream-wise velocity was measured at
approximately 40 levels within the boundary layer for every
case. The duration of data acquisition at each level
was 120 s. The bed shear stress was then calculated
numerically, using the momentum integral Eq. (3).
Depending on their height, waves generated in a shallow
wave ume are not truly monochromatic, but are nonlinear
with more peaked crests and atter troughs. Consequently,
unlike monochromatic waves the absolute values of the
wave amplitude, bed velocity and shear stress beneath the
crest and trough are not equal. For small waves this
difference is insignicant, but as the wave becomes larger,
the difference also increases. For each case, three friction
factors were estimated, one using the data under the crest,
one using the data under the trough and the last using
Table 2
Experimental conditions for a hydraulically smooth bed. The quantities U
b
; A
b
and t
0
are the average values measured under the crest and trough. Using the
criteria of Kamphuis [18], run S1 represents transitional ow, all other runs represent laminar ow
For all runs, period 2 s
Run code Wave height (cm) U
b
(cm/s) A
b
(cm) t
0
(N/m
2
) R
e
f
w
Water depth d (m)
S1 12.0 18.7 6.0 0.37 11257 0.021 0.55
S2 12.3 17.0 5.5 0.41 9265 0.028 0.60
S3 6.6 14.3 3.3 0.28 6692 0.027 0.44
S4 3.9 9.5 3.1 0.15 2945 0.033 0.44
S5 3.0 6.6 2.1 0.094 1382 0.044 0.44
S6 2.1 4.5 1.5 0.072 645 0.072 0.44
Fig. 8. Measured values of the bottom friction factor over a at
hydraulically smooth bed (sand grains) under progressing waves. The
solid line represents the theoretical result given by Eq. (9).
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 276
the average values under the crest and trough. The
calculated friction factors are shown in Fig. 8, along with
the theoretical result for monochromatic waves, [53]. It can
be shown that for a laminar boundary layer the theoretical
value of the maximum bottom shear stress is equal to, [53]:
t
0max
r

nv
p
U
b
7
Substituting Eq. (7) into Eq. (4) yields:
f
w

2

nv
p
U
b
8
Dening the near bottom amplitude Reynolds Number as
R
e

A
b
U
b
n
and considering that U
b
A
b
v; yields the
theoretical value of the friction factor for a laminar
boundary layer as:
f
w

2

R
e
p 9
where A
b
and U
b
are the maximum particle excursion and
velocity at the outer edge of the bottom boundary layer, v
is the angular frequency of the waves and n is the
kinematic viscosity. It can be seen in Fig. 8 that the various
denitions of f
w
used, differ only marginally from each
other and are in excellent agreement with the theoretical
result, given by Eq. (9).
Figs. 9 and 10 illustrate typical results of the tests on an
hydraulically smooth bed.
Fig. 9 shows the average velocity proles as a function of
distance above the bed for case S1. Each prole is obtained
for a given phase in the period of the external ow. Fig. 10
shows the phase averaged water surface elevation, bed
velocity and shear stress for case S1. It can be observed that
the velocity and shear stress are out of phase. The phase
difference for all cases was approximately 458 as shown in
Fig. 10, with the shear stress leading the velocity (and water
surface elevation). Kamphuis [18] gives the critical
Reynolds Number for the upper limit of laminar ow to
be 10,000. Based on this criterion, case S1 is classied as
transitional ow whilst the remainder of the cases are
laminar. Hence, the phase difference observed in these cases
is consistent with the predictions of theory for such ow
conditions [53].
4.1.2. Rough beds
As stated above, two kinds of rough bed roughness
elements were used in this study, (i) spherical aggregate
with a diameter between 4 and 9.5 mm, and (ii) rib
roughness with a triangular cross section, 1.36 cm height
and 2.25 cm base (Fig. 7). The tests on rough beds included
measurements of wave height, direct measurements of shear
stress using the shear plate, and measurements of the
stream-wise velocity just outside the boundary layer using
the LDA. The tests were performed for 71 different cases
presented in Tables 35.
For each case, 600 s of data were acquired (between 300
and 375 wave periods). The acquired data were phase
averaged in the same manner as for the hydraulically
smooth bed. Fig. 11 shows a typical result for run code 25.
The phase averaged water surface elevation, bed velocity
and bed shear stress are shown.
As explained for the hydraulically smooth bed, the
velocity and shear stress under the crest and trough have
different values for the waves generated in a shallow wave
ume. To obtain the value of friction factor from Eq. (4),
representative values for the maximum bed velocity
Fig. 9. Velocity prole within the boundary layer. The phase angle (in degrees) is shown by the number at the top of each prole. The case shown is S1, as
described in Table 2.
Fig. 10. Phase averaged bed velocity, water surface elevation and bed shear
stress for case S1 of Table 2 (hydraulically smooth bed).
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 277
and shear stress are necessary. In addition, to calculate the
Reynolds Number a representative value of water particle
excursion at the bed is also required. One possible option is to
use the average of the values measured beneath the crest and
trough. Another alternative is to calculate the maximum bed
velocity and water particle excursion via the root mean
square value of the bed velocity. For a monochromatic wave
the maximum value of the bed velocity can be obtained by
multiplying the root mean square velocity by

2
p
: The water
particle excursion at the bed can then be obtained by dividing
the maximum bed velocity by the angular frequency of the
waves. The measured results show that the mean of the crest
and trough values of the phase averaged velocity, shown by
Table 3
Experimental conditions for a rough bed (aggregate). For all cases, the
water depth d 0:44 m. Using the criteria of Kamphuis [18], the ow
conditions for all cases were rough-turbulent
Aggregate roughness, period 1:6 s
Run code Wave height (cm) U
b
(m/s) f
w
A
b
=k
s
R
e
63 2.60 0.07 0.63 0.86 1171
64 3.30 0.09 0.43 1.17 2139
65 4.20 0.11 0.39 1.45 3310
66 5.00 0.14 0.33 1.74 4736
67 6.40 0.16 0.27 2.01 6315
68 6.90 0.18 0.26 2.28 8200
69 7.80 0.20 0.24 2.55 10186
70 8.50 0.22 0.21 2.83 12580
71 9.50 0.24 0.19 3.08 14922
Table 4
Experimental conditions for a rough bed (aggregate). For all cases, the
water depth d 0:44 m. Using the criteria of Kamphuis [18], the ow
conditions for all cases were rough-turbulent
Aggregate roughness, period 2 s
Run code Wave height (cm) U
b
(m/s) f
w
A
b
=k
s
R
e
25 2.30 0.05 0.82 0.81 830
26 3.10 0.07 0.58 1.13 1592
27 3.30 0.08 0.55 1.28 2054
28 3.80 0.09 0.44 1.48 2740
29 4.00 0.10 0.39 1.64 3392
30 4.60 0.12 0.34 1.85 4281
31 4.80 0.13 0.31 2.01 5077
32 5.20 0.14 0.29 2.23 6239
33 5.50 0.15 0.28 2.39 7207
34 6.00 0.16 0.25 2.61 8566
35 6.60 0.18 0.25 2.79 9757
36 6.90 0.19 0.23 2.99 11261
37 7.10 0.20 0.22 3.17 12657
38 7.70 0.22 0.20 3.57 15993
39 7.70 0.22 0.19 3.57 16034
40 8.00 0.24 0.18 3.78 17968
41 8.90 0.26 0.15 4.16 21788
42 9.70 0.29 0.15 4.55 25977
43 10.60 0.31 0.13 4.91 30282
44 11.20 0.33 0.14 5.25 34680
45 12.60 0.35 0.13 5.52 38338
46 12.50 0.36 0.13 5.68 40492
47 13.10 0.37 0.12 5.85 43035
48 13.00 0.38 0.12 5.99 45146
49 13.70 0.39 0.12 6.13 47203
50 13.50 0.39 0.11 6.28 49626
51 14.20 0.40 0.11 6.42 51746
52 14.00 0.41 0.11 6.56 54132
53 14.30 0.42 0.10 6.72 56763
54 15.10 0.44 0.10 6.96 60864
55 15.90 0.45 0.10 7.20 65108
56 16.30 0.46 0.09 7.38 68407
57 17.00 0.49 0.09 7.76 75686
58 18.20 0.50 0.09 7.97 79914
59 18.60 0.51 0.08 8.18 84119
60 19.40 0.53 0.08 8.44 89620
61 19.30 0.54 0.08 8.58 92509
62 19.80 0.55 0.08 8.72 95444
Table 5
Experimental conditions for a rough bed (aggregate). For all cases, the
water depth d 0:44 m. Using the criteria of Kamphuis [18], the ow
conditions for all cases were rough-turbulent
Rib roughness, period 2 s
Run code Wave height (cm) U
b
(m/s) f
w
A
b
=k
s
R
e
1 2.20 0.05 1.00 0.31 762
2 2.90 0.07 0.78 0.44 1529
3 3.80 0.09 0.67 0.59 2690
4 4.20 0.11 0.57 0.73 4177
5 4.80 0.14 0.52 0.87 5990
6 5.50 0.16 0.45 1.04 8419
7 6.60 0.19 0.42 1.19 11092
8 6.90 0.21 0.39 1.35 14324
9 7.80 0.24 0.36 1.52 18182
10 8.70 0.26 0.33 1.67 22025
11 9.70 0.29 0.29 1.84 26493
12 10.70 0.31 0.28 1.97 30533
13 11.70 0.33 0.26 2.12 35457
14 12.40 0.36 0.25 2.27 40428
15 13.10 0.37 0.24 2.39 44707
16 13.80 0.39 0.23 2.49 48847
17 14.70 0.41 0.23 2.60 53171
18 15.40 0.42 0.21 2.70 57296
19 16.20 0.44 0.21 2.81 61971
20 16.90 0.46 0.20 2.90 66007
21 17.70 0.48 0.20 3.03 72300
22 18.20 0.48 0.20 3.07 74027
23 18.90 0.50 0.20 3.17 78880
24 19.50 0.51 0.20 3.27 83888
Fig. 11. Phase averaged bed velocity, wave height and bed shear stress for
run code 25 of Table 4 (rough bed).
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 278
the circles in Fig. 12, is very close to

2
p
U
rms
; denoted by the
symbols p. The difference appears to be less than 3%. Hence,
the maximum water particle velocity outside the boundary
layer was considered equal to

2
p
U
rms
; and the maximum
water particle excursion outside of the boundary equal to

2
p
U
rms
=v; where v is the angular frequency. The maximum
shear stress was taken as the mean of the crest and trough
values of the phase averaged shear stress.
4.1.3. Flow conditions
Many different criteria have been proposed for the lower
limit of rough turbulent ow. In Fig. 13, these criteria are
shown on a single graph to make a comparison between them
simpler. Also, the experimental points presented in Tables
35 are presented on the same graph, denoted by circles.
It can be seen that the data points follow a general
pattern, in the form of two parallel lines, related to the two
types of roughness used in the study. The data cluster with
the smaller values of A
b
=k
s
are related to the rib roughness
elements and the data cluster with the larger values of A
b
=k
s
are related to the spherical aggregate roughness. It can be
observed that, depending on the criteria adopted, some of
the points are in the rough turbulent region and some are in
the transitional region. To show the position of each test in
the gure, the portion of the graph where the experimental
points are located, has been expanded and shown in Fig. 14.
The numbers on this graph are the run codes mentioned in
Tables 35.
4.1.4. Comparison with existing relationships
and experimental data
In this section, the measured friction factors are presented
and a comparison with existing formulae and experimental
data is made. As mentioned in Section 1, numerous empirical
relationships for the estimation of bottom friction factors
under waves have previously been proposed. Amongst these,
the relationships suggested by Kamphuis [18], Jonsson [36]
and Kajiura [5] are best known and have been referenced by
many authors. The Kamphuis and Jonsson relationships are
based ontests performed in an oscillating water tunnel, whilst
the Kajiura formula is theoretical, based on a time invariant
eddy viscosity model. In addition to these relationships, there
are also a large body of experimental data available. These
experimental data include that obtained by Sleath [33], using
an oscillating tray in a water tank, [31] in a wave ume, [54]
also in a wave ume, [55] in an oscillating water tunnel, [30,
34] in a wave ume and [23] in an oscillating water tunnel. To
make the comparison simpler, all the experimental data and
the formulae, along with the experimental results of this
study are drawn on the same graph. This comparison is made
in the friction factor-relative roughness domain in Fig. 15,
Fig. 12. A comparison of different methods for the determination of the
maximum bed velocity for nonlinear waves generated in the wave ume.
The means of the phase averaged values beneath the crest and the trough are
shown by o, whilst values given by

2
p
U
rms
are shown by p .
Fig. 13. Flow condition criteria proposed by different authors. Data from the present experiment are shown by the open circles.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 279
whereby it is assumed that conditions are rough turbulent and
not R
e
dependent. It can be observed that generally the
experimental results of this study match well with the
empirical relationship of Kamphuis [18]. At relative rough-
ness less than 1, the data points deviate from all the
relationships, but the agreement with the experimental data
of Simons et al. [56] and Kemp and Simons [54] is very good.
This agreement is better for the experimental results on the
rib surface than on the aggregate. It is worth pointing out that
the tests performed by Simons et al. [56] and Kemp and
Simons [54] were also performed on rib roughness elements
in a wave ume. The relative roughness for spherical
aggregates and rib roughness were chosen as 20 and 50 mm,
respectively. These values were based on the experiments in
unidirectional currents as described in Section 4.1.3 and the
existing empirical data.
To make a comparison between the friction factors
obtained in this study and the data of other authors in
Fig. 14. Flow condition criteria as shown in Fig. 13. The numbers refer to the run identiers of Tables 35.
Fig. 15. The wave friction factor in the rough turbulent regime. A comparison is shown between data from the present experiment and previous studies.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 280
a friction factorReynolds Number domain, the experimen-
tal results are presented on the friction factor diagram
suggested by Kamphuis [18]. This comparison is presented
in Fig. 16. The circles in the gure denote the friction
factors obtained in this study as a function of Reynolds
Number. It can be seen that the general pattern of the data
is in the form of two parallel lines related to the two
roughness elements used in the study. The upper line is
related to the rib roughness elements and the lower one
to the spherical aggregate. In addition, the hydraulically
smooth bed data (average vales), previously shown in
Fig. 8 are also plotted on this gure (triangles).
To show the relative roughnesses related to the friction
factors, the portion of Fig. 16 where the data points are
located has been expanded and depicted in Fig. 17. The
numbers in this graph denote the A
b
=k
s
values related to each
data point. It can be seen that the agreement between the
results of this study and those of Kamphuis [18] in the range
of the experimental data is good.
The same comparison has been made between the
friction factors obtained in this study and Jonsson [36]
suggested diagram in a friction factorReynolds Number
domain. Fig. 18 shows this comparison. The circles in the
gure show the friction factors as a function of Reynolds
Number. As in Fig. 16, the hydraulically smooth bed data is
also included as shown by the triangles. Again, the portion
of Fig. 18 where the data are located has been expanded
in Fig. 19. It can be seen that for a relative roughness
of approximately 1, the agreement between this study and
the Jonsson diagram is good but for larger relative
roughnesses, for the same Reynolds Number, the Jonsson
diagram yields smaller friction factors. The difference is
estimated to be approximately 5%.
4.2. Monochromatic representation of the wave spectrum
As mentioned previously, experimental and theoretical
studies of wave bottom friction have mainly been performed
for monochromatic waves. To apply this considerable body
of knowledge to a spectrum of waves, it is necessary to
approximate a wave spectrum by a monochromatic wave.
As far as the energy dissipation of waves due to bottom
friction is concerned, a monochromatic wave and a
spectrum of waves can be considered equivalent provided
they dissipate the same amount of energy under similar
circumstances. To this end, a series of tests have been
performed to examine such a simplication.
In these tests, several JONSWAP spectra with different
peak frequencies and total energy were simulated and the
bottom shear stress and associated velocity were measured
directly. As mentioned in Section 1, the energy dissipated
due to bottom friction, for a monochromatic wave, averaged
over one wave period, can be expressed by Eq. (1) and in the
case of a wave spectrum by Eq. (2).
Based on this denition, the dissipated energy due to
bottom friction and the root-mean-square water velocity,
u
rms
; were calculated for each spectrum. The results are
presented in Table 6. In the following, the root-mean-square
velocity for a wave spectrum is denoted as u
rms
and that for a
monochromatic wave as u
rms;mono
:
In the next phase of the tests, four series of monochromatic
waves were generated. The periods selected were 1.6, 2 and
Fig. 16. The friction factor diagram proposed by Kamphuis [18] with the data from the present study shown. The circles represent the rough bed data (aggregate
and triangular ribs), the triangles represent the hydraulically smooth bed data also shown in Fig. 8.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 281
2.4 s, to match the peak spectral periods of the tested spectra.
The waves in each series had the same period but different
amplitudes. The bottom velocity and shear stress were
measured and the dissipation of energy due to bottomfriction
was calculated for each case. The graphs in Figs. 2022 show
the variation of dissipated energy versus root-mean-square
bottom velocity for each series of monochromatic waves on
hydraulically smooth and rough beds.
Fig. 17. The friction factor diagramproposed by Kamphuis [18] as in Fig. 16 with data from the present study. The numbers shown on the R
e
f
w
plain represent
measured values of A
b
=k
s
; which can be compared with Kamphuis [18] values shown by the solid lines (whose values of A
b
=k
s
are shown at the extreme right of
the diagram).
Fig. 18. The friction factor diagramproposed by Jonsson [36], with the data fromthe present study shown, as in Fig. 16. The circles represent the rough bed data
(aggregate and triangular ribs), the triangles represent the hydraulically smooth bed data also shown in Fig. 8.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 282
It can be seen that in all cases there is an approximately
logarithmic relationship, irrespective of the wave period.
Using these graphs, it is possible to determine monochro-
matic waves which dissipate the same amount of energy as
the spectra presented in Table 7 and determine their root-
mean-square bottom velocity, u
rms;mono
: The second column
in Table 7 shows the spectral values of u
rms
; based on direct
measurement. The third column gives the equivalent
monochromatic root-mean-square velocity (dissipates the
same energy), u
rms;mono
; obtained form Figs. 2022 and the
fourth column gives the ratio between these velocities.
Although there is some variation, the ratio is approximately
constant at 1.33. Therefore, since the orbital velocity
amplitude is equal to

2
p
u
rms;mono
; it is suggested that a
wave spectrum can be represented by a monochromatic
wave, the bottom velocity amplitude of which is equal to
1:88u
rms
; and with a period equal to the spectral peak period.
It is interesting to speculate as to the signicance of this
number. The results are clear, for the same rms bed velocity,
a monochromatic wave will dissipate less energy than a
spectrum of waves. The obvious reason is that the friction
factor, f
w
is different for the two cases. In the rough
turbulent zone, the friction factor is largely insensitive to
changes in bed velocity or R
e
(i.e. in the plateau region of
Figs. 16 and 18). The effective value of A
b
could, however,
result in a change in f
w
: In the hydraulically smooth bed
region, however, f
w
varies signicantly as the velocity R
e

varies. The fact that the value (1.33) is almost the same in
these two quite different regimes indicates that it is not
Table 6
Characteristics of tested wave spectra on hydraulically smooth and rough
beds. The parameters a; s and g are the standard JONSWAP parameters.
Cases with run codes commencing with the letter R refer to strip rippled
bed cases, those commencing with S to aggregate beds and those
commencing with J to hydraulically smooth beds
For all spectra, a 0:01; s 0:1; g 3:3
Run code T
p
(s)
H
s
(mm)
u
rms
(m/s)
Dissipated energy
(kg/s
3
)
Bed condition
J161 1.6 67 0.0767 2.04 10
24
Smooth
J162 1.6 56 0.0610 8.26 10
24
Smooth
J163 1.6 41 0.0447 0.28 10
24
Smooth
J201 2 153 0.1407 15.10 10
24
Smooth
J202 2 124 0.1170 8.86 10
24
Smooth
J203 2 89 0.1030 4.29 10
24
Smooth
J204 2 54 0.0825 1.85 10
24
Smooth
J205 2 25 0.0574 0.43 10
24
Smooth
J241 2.4 92 0.109 13.00 10
24
Smooth
J242 2.4 78 0.094 8.03 10
24
Smooth
S201 2 52.5 0.0616 0.099 Rough
S202 2 73 0.0883 0.224 Rough
S203 2 93 0.1118 0.359 Rough
S161 1.6 54 0.0640 0.139 Rough
S162 1.6 70.5 0.0806 0.225 Rough
S241 2.4 80.5 0.100 0.251 Rough
S242 2.4 92 0.1170 0.357 Rough
R201 2 53 0.0625 0.150 Rough
R202 2 71 0.0894 0.345 Rough
R203 2 95 0.113 0.580 Rough
R241 2.4 78.5 0.101 0.436 Rough
R242 2.4 90 0.1187 0.618 Rough
Fig. 19. The friction factor diagram proposed by Jonsson [36] as in Fig. 18 with data from the present study. The numbers shown on the R
e
f
w
plain represent
measured values of A
b
=k
s
; which can be compared with Jonsson [36] values shown by the solid lines (whose values of A
b
=k
s
are shown at the extreme right of the
diagram), as in Fig. 17.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 283
a simple case of a change in the effective bed velocity
causing the difference in the energy dissipation.
Although the rms velocities may be the same in the two
cases, the ows may be quite different. In the spectral case,
oscillatory ows will occur across a range of frequencies. It
is possible that this range of scales gives rise to turbulence at
a broader range of scales. Nonlinear interaction between
these different scales possibly giving rise to enhanced
dissipation. The exact reason is beyond the scope of this
paper and awaits further work.
The present results clearly show that the factor is
remarkably constant (1.33) under very different bed
conditions. The present data have not, however, investigated
whether this is the case for different spectral shapes.
Although a number of different peak frequencies were
investigated, a spectral shape approximating the mean
JONSWAP form was retained throughout. In the limit, as
the spectrum becomes narrower, it will approach a
monochromatic wave. Therefore, one would expect
the energy dissipation to decrease as the spectrum narrows.
The exact dependence of the energy dissipation will require
further study.
5. Conclusions
The literature on bottom friction dissipation under
oscillatory waves is very large. The vast majority of this
literature has, however, been for monochromatic waves and
has been obtained in laboratory studies using water tunnels
or oscillatory trays. When applying these studies to eld
applications, two signicant questions are raised:
1. How can friction factor values obtained for monochro-
matic conditions be applied to more typical spectral
conditions?
2. Are measurements made in oscillatory water tunnels or
over oscillating trays really representative of the case
of nonlinear water waves interacting with the bottom?
The experimental program described in this paper was
designed to address these problems. Measurements were
Fig. 20. Variation of the root mean square of the dissipated energy versus
the root mean square of the bottomvelocity on an hydraulically smooth bed.
Fig. 21. Variation of the root mean square of the dissipated energy versus
the root mean square of the bottom velocity on a rough bed (aggregate).
Fig. 22. Variation of the root mean square of the dissipated energy versus
the root mean square of the bottom velocity on a rib rippled bed.
Table 7
Comparison between a wave spectrum and its equivalent monochromatic
wave
Run code u
rms
(m/s) u
urm;mono
Ratio A
rms
=k
s
J161 0.0767 0.1015 1.32 Smooth
J162 0.0610 0.0804 1.31 Smooth
J163 0.0447 0.0610 1.36 Smooth
J201 0.1407 0.1794 1.26 Smooth
J202 0.1170 0.1604 1.31 Smooth
J203 0.1030 0.1283 1.24 Smooth
J204 0.0825 0.1055 1.24 Smooth
J205 0.0574 0.0711 1.21 Smooth
J241 0.1090 0.1480 1.36 Smooth
J242 0.0944 0.1320 1.39 Smooth
S201 0.0616 0.0803 1.36 1.5
S202 0.0883 0.1262 1.43 2.2
S203 0.1118 0.1600 1.43 2.8
S161 0.0640 0.0888 1.39 0.8
S162 0.0806 0.1138 1.41 1
S241 0.100 0.1347 1.35 1.9
S242 0.117 0.1555 1.33 2.3
R201 0.0625 0.0877 1.40 1.6
R202 0.0894 0.1262 1.41 2.2
R203 0.113 0.1583 1.40 2.8
R241 0.101 0.1364 1.35 0.8
R242 0.1187 0.1558 1.32 0.9
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 284
made using a combination of velocity measurements within
the bottom boundary layer and direct measurements of the
bottom shear stress using a shear plate. Such measurements
are demanding, and the shear plate results need to be
carefully corrected to remove spurious edge forces exerted
on the plate.
The results obtained for monochromatic waves are in
excellent agreement with earlier measurements made using
water tunnels and oscillatory trays, thus providing conr-
mation that these measurements are indeed representative of
the situation under water waves. Thus, the very signicant
body of data obtained in such measurements can be applied
with condence.
Measurements of the energy dissipated at the bed have
shown that, for the purposes of calculating bottom
friction decay, a spectrum can be represented by a
monochromatic wave, the maximum bed velocity of
which is equal to 1.88 times the rms bed velocity of the
spectrum. Hence, typical friction factor diagrams
expressed in terms of R
e
and A
b
=k
s
can be directly
applied to spectral conditions.
The present study is limited to spectra approximating
the mean JONSWAP form. Although not conrmed in
this study, it is speculated that the factor 1.88 will be
smaller for narrower spectral forms, approaching

2
p
for
the limiting condition of a monochromatic wave (e.g.
swell).
Appendix A. Expressions for the friction factor
suggested by different authors for mobile and immobile
at and rippled beds
Table A1
Bed friction factor in an oscillatory ow
Authors name Relation
Bagnold (1947)
For
A
b
l
. 1 !f
w
0:216
A
b
l
_ _
23=4
for
A
b
l
, 1 !f
w
0:24
Jonsson [32] Smooth turbulent ow
1
4

f
w
p 2 log
1
4

f
w
p 21:55 log R
e
Rough turbulent ow
1
4

f
w
p log
1
4

f
w
p 20:08 log
A
b
k
s
Kajiura [5] Smooth turbulent ow
1
8:1

f
w
p log
1

f
w
p 20:135 log

R
e
_
Rough turbulent ow
1
4:05

f
w
p log
1
4

f
w
p 20:254
A
b
k
s
Table A1 (continued)
Authors name Relation
Kamphuis [18] For rough turbulent ow, k
s
. 2D
90
1
4

f
w
p log
1
4

f
w
p 20:35
4
3
log
A
b
k
s
for laminar ow
2

R
p
Swart [61,62]
For
k
s
A
b
, 0:63 !f
w
exp 5:213
k
s
A
b
_ _
0:194
25:977
_ _
For
k
s
A
b
$ 0:63 !f
w
0:3
Immobile bed u , 0:05 !k
s
2:5D
Flat mobile bed u . 1 !k
s
190

u 20:05
p
D
k
s
25
h
2
l
, [61,62]; or for mobile bed 0:05 , u , 1
k
s

8h
2
l
190

u 20:05
p
D, [63]
Vitale [64] f
w
22:3r
0:9
p
. 37:5r
p
r
p
0:01
h
A
b
2
D
90
A
b
0:01
h
A
b

200D
90
A
b
_ _
Swart [65]
for
A
b
k
s
, 7 !f
w

0:30
1 0:28
A
b
k
s
_ _
For 7 ,
A
b
k
s
# 160 !f
w
0:0066 0:483
A
b
k
s
_ _
20:91
For 160 ,
A
b
k
s
!f
w
0:0146
A
b
k
s
_ _
20:157
0:483
A
b
k
s
_ _
20:91
Grant and
Madsen [66]
For
A
b
k
s
. 1 !f
w

0:08
ker
2
2z
0

1=2
kei
2
2z
0

1=2

For
A
b
k
s
# 1 !f
w
0:23
z
0

k
s
30l
; l 0:4u
p
=v; u
p

f
w
2
_
U
b
k
s
A
b
28
h
A
b
h
l
160s c
m

d
A
b
c
c
c
0
c
c
_ _
1=2
20:7
_ _
2
Vongvisessomjai
[67]
Fixed bed, k
s
2D
f
w
0:287
A
b
k
s
_ _
22=3
Mobile bed
f
w
0:049F
20:44
p
a
0:21
p
D
0:2
p
D; mean diameter of sediment, a
p
A
b
=D;
D
p
s 21gD
3
=n
2
and F
p
1:84 10
23
a
4=3
p
Wilson [68]
0:4

2
f
w
_
ln
3
20p

2
f
w
_
gs 21T
2
A
b
_ _
for c . 0:8
(continued on next page)
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 285
Symbols used in Table A1 here
U
b
amplitude of monochromatic near bottom uid
velocity,
c
m
coefcient which accounts for the effect of the
added mass of the moving sediment particle in
water,
D grain size,
ker and kei Kelvin functions of zeroth order,
A
b
amplitude of uid particles outside boundary layer
at the bed,
h ripple height,
l wave length of ripples,
k
s
equivalent Nikuradse sand roughness,
u mobility number,
c Shields parameter,
c
0
Shields parameter based on grain roughness,
c
c
critical Shields parameter, associated with inci-
pient motion of sediment,
s relative density of sediments,
R
e
Reynolds number,
T wave period.
References
[1] Putnam JA, Jonsson JW. The dissipation of wave energy by bottom
friction. Trans Am Geophys Union 1949;30.
[2] Hasselmann K, Collins JI. Spectral dissipation of nite-depth gravity
waves due to turbulent bottom friction. J Marine Res 1968;26:112.
[3] Trowbridge J, Madsen OS. Turbulent wave boundary layers. Model
formulation and rst order solution. J Geophys Res 1984;89:798997.
[4] Trowbridge J, Madsen OS. Turbulent wave boundary layers 2. Second
order theory and mass transport. J Geophys Res 1984;89:79998007.
[5] Kajiura K. A model of the bottom boundary layer in water waves. Bull
Earthq Res Inst 1968;46:75123.
[6] Grant WD, Madsen OS. Combined wave and current interaction with
a rough bottom. J Geophys Res 1979;84:1797808.
[7] Brevik I. Oscillatory rough turbulent boundary layers. Journal of the
Waterway, Port, Coastal and Ocean Division, Proceedings of
the American Society of Civil Engineers, ASCE, 107 No. ww3
1981;17588.
[8] Myrhaug D. On a theoretical model of rough turbulent wave boundary
layers. Ocean Engng 1982;9:54765.
[9] You ZJ, Wilkinson DL, Nielsen P. Velocity distribution in turbulent
oscillatory boundary layer. Coastal Engng 1992;18:2138.
[10] Hsu T, Ou S. Wave boundary layers in rough turbulent ow. J Coastal
Engng 1997;24:2543.
[11] Justesen P. Prediction of turbulent oscillatory ow over rough beds.
Coastal Engng 1988;12:25784.
[12] Aydin I, Shuto N. An application of the k-epsilon model to oscillatory
boundary layers. Coastal Engng Jpn 1998;30(2):1124.
[13] Bagnold RA. Motion of waves in shallow water interaction between
waves and sand bottom. Proc R Soc Lond A 1946;187:148.
[14] Kalkanis G. Turbulent ow near an oscillating wall. Technical
Memorandum, Beach Erosion Board Corps of Engineers; 1957. p. 97.
[15] Sleath JFA. Sea bed mechanics. New York: Wiley; 1984.
[16] Jonsson IG. Measurements in the turbulent wave boundary layer.
Proceedings 10th Congress, vol. 1. International Association Hydro-
logical Research; 1963. pp. 8592.
[17] Carstens MR, Nielsen FM, Altinblek HD. Bed forms generated in the
laboratory under an oscillatory ow: analytical and experimental
study. Technical memorandum No. 28, US Army Corps of Engineers,
Coastal Engineering Research Center; 1969.
[18] Kamphuis JW. Friction factor under oscillatory waves. Journal of the
Waterways Harbors and Coastal Engineering Division. Proceedings
of the American Society of Civil Engineering 101; 1975. p. 13544.
[19] Jonsson IG, Carlsen NA. Experimental and theoretical investigations
in an oscillatory turbulent boundary layer. J Hydraulic Res 1976;14:
4560.
[20] Lofquist KEB. Measurements of oscillatory drag and sand ripples.
Coastal Engng Conf 1980;3:3087100.
[21] Van Doorn Th. Computations and comparison with measurements of
the turbulent bottom boundary layer in an oscillatory ow. Report on
Computations. M 1562, Part, 2. Delft Hydraulic Lab; 1983.
[22] Sato S, Mimura N, Watanabe A. Oscillatory boundary layer ow over
rippled beds. Coastal Engng Conf 1984;3:2293309.
[23] Sumer BM, Jensen BL, Fredsoe J. Turbulence in oscillatory boundary
layers. In: Bellot GC, Mthieu J, editors. Advances in Turbulence. New
York: Springer; 1987. p. 55667.
[24] Hino M, Kashiwayanage M, Nakayama A, Hara T. Experiments on
the turbulence statistics and the structure of a reciprocating oscillatory
ow. J Fluid Mech 1983;131:363400.
[25] Sawamoto M, Sato E. The structure of oscillatory turbulent boundary
layer over rough bed. Coastal Engng Jpn 1991;34:113.
[26] Savage RD. Laboratory study of wave energy losses by bottom
friction and percolation. Technical Memorandum. No 31, United
States Beach Erosion Board; 1953.
[27] Inman DL, Bowen AJ. Flume experiments on sand transport by waves
and currents. Proceedings of the 8th Conference Coastal Engineering
1963;13750.
[28] Horikawa K, Watanabe A. Laboratory study on oscillatory boundary
layer ow. Coastal Engng Jpn 1968;11:1328.
[29] Van Doorn Th. Experimental investigation of near bottom velocities
in water waves without and with a current. Report investigation M
1423, Delft Hydraulic Lab; 1981. Part 1.
Table A1 (continued)
Authors name Relation
Aydin and
Shuto (12)
Rough bed
5 ,
U
b
vk
s
, 10
4
1
4

f
w
p log
1
4

f
w
p 0:082 0:959 log
U
b
vk
s
_ _
Smooth bed
R
e
, 7 10
4
!f
w

2:0

R
e
p
R
e
. 4 10
5
!f
w

0:043
R
1=6
e
Myrhaug [69]
Rough
1
4:07

f
w
p log
1
4

f
w
p 0:256 log
A
b
k
s
Smooth
1
8:14

f
w
p log
1
4

f
w
p 20:51 log

R
e
_
Transition
1
4:07

f
w
p log
1
4

f
w
p
0:2562
log 12exp 20:0262
R
e
A
b
=k
s

f
w
_
_ _
4:71
A
b
=k
s
R
e

f
w
p
_ _
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 286
[30] Brevik I. Flume experiment on waves and currents I. Smooth bed.
Coastal Engng 1980;3.
[31] Simons RR, Grass AJ, Mansour-Tehrani M. Bottom shear stresses in
the boundary layers under waves and currents crossing at right angles.
Proceedings of the 23rd International Conference on Coastal
Engineering, vol. 1.; 1992. p. 60417.
[32] Jonsson IG. Wave boundary layer and friction factors. Coastal
Engineering Conference; 1966. pp. 125148.
[33] Sleath JFA. Transition in oscillatory ow over rough beds.
J Waterway, Port, Coastal Ocean Engng 1988;114:1833.
[34] Brevik I. Flume experiment on waves and currents II. Rippled bed.
Coastal Engng 1980;4:89110.
[35] Malarkey J, Davies AG. Discrete vortex modelling of oscillatory ow
over ripples. Appl Ocean Res 2002;24:12745.
[36] Jonsson IG.Friction factor diagrams for oscillatory boundary layers.
Technical University of Denmark, Coastal Engineering Laboratory,
Progress Report No. 10; 1965.
[37] Collins JI. Prediction of shallow water spectra. J Geophys Res 1972;
77:2693707.
[38] Madsen OS, Poon Y, Graber HC. Spectral wave attenuation by bottom
friction: theory. Coastal Engng Conf 1988;1:492504.
[39] Weber SL. Eddy-viscosity and drag-law models for random ocean
wave dissipation. J Fluid Mech 1991;232:7398.
[40] Bretschneider CL, Reid RO. Changes in wave height due to bottom
friction, percolation and refraction. Technical Memooir US Army
Erosion Board No. 45 1954;.
[41] Iwagaki Y, Kakinuma T. On the bottom friction factors off ve
Japanese Coasts. Coastal Engng Jpn 1967;10:1322.
[42] Ifuku M, Kakinuma T. Bottomfriction factors off ve Japanese coasts.
Coastal Engng Jpn 1982;25:3749.
[43] Lambrakos KF. Seabed wave boundary layer measurements and
analysis. J Geophys Res 1982;87:417189.
[44] Myrhaug D, Lambrakos KF, Slattelid OH. Wave boundary layer in
ow measurements near the seabed. Coastal Engng 1992;18:
15381.
[45] Young IR, Gorman RM. Measurements of the evolution of ocean
wave spectra due to bottom friction. J Geophys Res 1994;100:
109871004.
[46] Simons RR, Grass AJ, Saleh WM, Mansour-Tehrani M. Bottom shear
stresses under random waves with a current superimposed. Proceed-
ings of the 24th International Conference on Coastal Engineering
1994;56578.
[47] Bouws E, Komen GJ. On the balance growth and dissipation in an
extreme depth-limited wind-sea in the southern North sea. J Phys
Oceanogr 1983;13:16538.
[48] Luo W, Monbaliu J. Effects of the bottom friction formulation on the
energy balance for gravity waves in shallow water. J Geophys Res
1994;99:1850111.
[49] Weber SL. Bottom friction for wind sea and swell in extreme depth-
limited situations. J Phys Oceanogr 1991;21:14972.
[50] Asano T, Iwagaki Y. Nonlinear effects on velocity elds in turbulent
wave boundary layer. Coastal Engng Jpn 1986;29:5263.
[51] Mirfenderesk H. The dissipation of ocean wave spectra due to
bottom friction. PhD Thesis. University of NSW, Australia; 1999.
p. 230.
[52] Hodge BK. An assessment of rough surface boundary-layer calculation
methods. Turbulent boundary layer, forced, incompressible, non-
reacting, The American Society of Mechanical Engineers; 1979.
[53] Nielsen P. Coastal bottom boundary layers and sediment transport.
Advance Series on Ocean Engineering, Vol. 4. Singapore: World
Scientic; 1992.
[54] Kemp PH, Simons RR. The interaction between waves and a turbulent
current: waves propagating with the current. J Fluid Mech 1982;116:
22750.
[55] Jensen BL. Experimental investigation of turbulent oscillatory
boundary layers. Institute of hydrodynamic and hydraulic engineer-
ing. Technical university of Denmark series paper No 45; 1988.
[56] Simons RR, Grass AJ, Kyriacou A. The inuence of currents on wave
attenuation. Coastal Engng Conf 1988;1:36376.
[57] Hasselmann K. Measurements of wind-wave growth and swell decay
during the Joint North Sea Wave Project (JONSWAP). Dtsch
Hydrogh Z 1973;8, 12(Suppl. A):95 pp.
[58] Jensen BL, Sumer BM, Fredsoe J. Turbulent oscillatory boundary
layer at high Reynolds Numbers. J Fluid Mech 1989;206:26597.
[59] Kalkanis G. Transportation of bed material due to wave action. US
Army, Coastal Engineering Research Center. Technical Memoran-
dum No. 2; 1964.
[60] Sato S, Khimosako K, Watanabe A. Measurements of oscillatory
turbulent boundary layer ow above ripples with a Laser Doppler
Velocimeter. Coastal Engng Jpn 1987;30:8998.
[61] Swart DH. Predictive equations regarding coastal transports. Proc
15th Conf Coastal Engng 1976;2:111332.
[62] Swart DH. Computation of long shore transport. Delft Hydraulics
Laboratory Report R968; 1976.
[63] Nielsen P. Dynamics and geometry of wave-generated ripples.
J Geophys Res 1981;86:646772.
[64] Vitale P. Sand bed friction factors for oscillatory ows. J Waterways,
Port, Coastal Ocean Div 1979;105:22945.
[65] Swart DH, Fleming CA. Long shore water and sediment movement.
Proceedings of the 17th Conference on Coastal Engineering
Conference, vol. 2.; 1980.
[66] Grant WD, Madsen OS. Movable bed roughness in unsteady
oscillatory ow. J Geophys Res 1982;87:46981.
[67] Vongvisessomjai S. Wave friction factor on sand ripples. Coastal
Sediments87 1987;1:393408.
[68] Wilson KC. Friction of wave induced sheet ow. Coastal Engng 1989;
12:3719.
[69] Myrhaug D. A rational approach to wave friction coefcients for
rough, smooth and transitional turbulent ow. Coastal Engng 1989;
13:1121.
H. Mirfenderesk, I.R. Young / Applied Ocean Research 25 (2003) 269287 287

Potrebbero piacerti anche