Sie sind auf Seite 1von 11

Coastal Engineering 55 (2008) 1102–1112

Contents lists available at ScienceDirect

Coastal Engineering
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c o a s t a l e n g

Characteristics of turbulent boundary layers over a rough bed under saw-tooth waves
and its application to sediment transport
Suntoyo a,b,⁎, Hitoshi Tanaka b, Ahmad Sana c
a
Department of Ocean Engineering, Faculty of Marine Technology, Institut Teknologi Sepuluh Nopember (ITS), Surabaya 60111, Indonesia
b
Department of Civil Engineering, Tohoku University, 6-6-06 Aoba, Sendai 980-8579, Japan
c
Department of Civil and Architectural Engineering, Sultan Qaboos University, P.O. Box 33, AL-KHOD 123, Oman

A R T I C L E I N F O A B S T R A C T

Article history: A large number of studies have been done dealing with sinusoidal wave boundary layers in the past.
Received 14 August 2007 However, ocean waves often have a strong asymmetric shape especially in shallow water, and net of
Received in revised form 30 March 2008 sediment movement occurs. It is envisaged that bottom shear stress and sediment transport behaviors
Accepted 4 April 2008
influenced by the effect of asymmetry are different from those in sinusoidal waves. Characteristics of the
Available online 21 May 2008
turbulent boundary layer under breaking waves (saw-tooth) are investigated and described through both
Keywords:
laboratory and numerical experiments. A new calculation method for bottom shear stress based on velocity
Turbulent boundary layers and acceleration terms, theoretical phase difference, φ and the acceleration coefficient, ac expressing the
Sheet flow wave skew-ness effect for saw-tooth waves is proposed. The acceleration coefficient was determined
Sediment transport empirically from both experimental and baseline k–ω model results. The new calculation has shown better
Skew waves agreement with the experimental data along a wave cycle for all saw-tooth wave cases compared by other
Saw-tooth waves existing methods. It was further applied into sediment transport rate calculation induced by skew waves.
Sediment transport rate was formulated by using the existing sheet flow sediment transport rate data under
skew waves by Watanabe and Sato [Watanabe, A. and Sato, S., 2004. A sheet-flow transport rate formula for
asymmetric, forward-leaning waves and currents. Proc. of 29th ICCE, ASCE, pp. 1703–1714.]. Moreover, the
characteristics of the net sediment transport were also examined and a good agreement between the
proposed method and experimental data has been found.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction and Sana and Shuy (2002) have compared the direct numerical
simulation (DNS) data for sinusoidal oscillatory boundary layer on
Many researchers have studied turbulent boundary layers and smooth bed with various two-equation turbulence models and, a
bottom friction through laboratory experiments and numerical quantitative comparison has been made to choose the best model for
models. The experimental studies have contributed significantly specific purpose. However, these models were not applied to predict
towards understanding of turbulent behavior of sinusoidal oscillatory the turbulent properties for asymmetric waves over rough beds.
boundary layers over smooth and rough bed (e.g., Jonsson and Carlsen, Many studies on wave boundary layer and bottom friction asso-
1976; Tanaka et al., 1983; Sleath, 1987, Jensen et al., 1989). These ciated with sediment movement induced by sinusoidal wave motion
studies explained how the turbulence is generated in the near-bed have been done (e.g., Fredsøe and Deigaard, 1992). These studies have
region either through the shear layer instability or turbulence bursting shown that the net sediment transport over a complete wave cycle is
phenomenon. Such studies included measurement of the velocity zero. In reality, however ocean waves often have a strongly non-linear
profiles, bottom shear stress and some included turbulence intensity. shape with respect to horizontal axes. Therefore it is envisaged that
An extensive series of measurements and analysis for the smooth bed turbulent structure, bottom shear stress and sediment transport be-
boundary layer under sinusoidal waves has been presented by Hino haviors are different from those in sinusoidal waves due to the effect
et al. (1983). Jensen et al. (1989) carried out a detailed experimental of acceleration caused by the skew-ness of the wave.
study on turbulent oscillatory boundary layers over smooth as well as Tanaka (1988) estimated the bottom shear stress under non-linear
rough bed under sinusoidal waves. Moreover, Sana and Tanaka (2000) wave by modified stream function theory and proposed formula to
predict bed load transport except near the surf zone in which the
acceleration effect plays an important role. Schäffer and Svendsen
⁎ Corresponding author. Department of Civil Engineering, Tohoku University, 6-6-06
Aoba, Sendai 980-8579, Japan.
(1986) presented the saw-tooth wave as a wave profile expressing
E-mail addresses: suntoyo@oe.its.ac.id, suntoyo@kasen1.civil.tohoku.ac.jp (Suntoyo), wave-breaking situation. Moreover, Nielsen (1992) proposed a bottom
tanaka@tsunami2.civil.tohoku.ac.jp (H. Tanaka), sana@squ.edu.om (A. Sana). shear stress formula incorporating both velocity and acceleration

0378-3839/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.coastaleng.2008.04.007
Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112 1103

terms for calculating sediment transport rate based on the King's 2. Experimental study
(1991) saw-tooth wave experiments with the phase difference of 45°.
Recently, Nielsen (2002), Nielsen and Callaghan (2003) and Nielsen 2.1. Turbulent boundary layer experiments
(2006) applied a modified version of the formula proposed by Nielsen
(1992) and applied it to predict sediment transport rate with various Turbulent boundary layer flow experiments under saw-tooth
experimental data. They have shown that the phase difference waves were carried out in an oscillating tunnel using air as the
between free stream velocity and bottom shear stress used to evaluate working fluid. The experimental system consists of the oscillatory
the sediment transport is from 40° up to 51°. Whereas, many flow generation unit and a flow-measuring unit. The saw-tooth wave
researchers e.g. Fredsøe and Deigaard (1992), Jonsson and Carlsen profile used is as presented by Schäffer and Svendsen (1986) by
(1976), Tanaka and Thu (1994) have shown that the phase difference smoothing the sharp crest and trough parts. The definition sketch for
for laminar flow is 45° and drops from 45° to about 10° in the saw-tooth wave after smoothing is shown in Fig. 1. Here, Umax is the
turbulent flow condition. However, Sleath (1987) and Dick and Sleath velocity at wave crest, T is wave period, tp is time interval measured
(1991) observed that the phase difference and shear stress were from the zero-up cross point to wave crest in the time variation of free
depended on the cross-stream distance from the bed, z for the mobile stream velocity, t is time and α is the wave skew-ness parameter. The
roughness bed. It is envisaged that the phase difference calculated at smaller α indicate more wave skew-ness, while the sinusoidal wave
base of sheet flow layer may be very close to 90°, while the phase (without skew-ness) would have α = 0.50.
difference just above undisturbed level may only 10–20° and the The oscillatory flow generation unit comprises of signal control
phase difference about 51° as the best fit value obtained by Nielsen and processing components and piston mechanism. The piston
(2006) may be occurred at some depth below the undisturbed level. displacement signal is fed into the instrument through a PC. Input
More recently, Gonzalez-Rodriguez and Madsen (2007) presented digital signal is then converted to corresponding analog data through
a simple conceptual model to compute bottom shear stress under a digital–analog (DA) converter. A servomotor, connected through a
asymmetric and skewed waves. The model used a time-varying servomotor driver, is driven by the analog signal. The piston mecha-
friction factor and a time-varying phase difference assumed to be the nism has been mounted on a screw bar, which is connected to the
linear interpolation in time between the values calculated at the crest servomotor. The feed-back on piston displacement, from one instant
and trough. However, this model does not parameterize the fluid to the next, has been obtained through a potentiometer that com-
acceleration effect or the horizontal pressure gradients acting on the pared the position of the piston at every instant to the input signal,
sediment particle. Moreover, this model under predicted most of and subsequently adjusted the servomotor driver for position at the
Watanabe and Sato's (2004) experimental data induced by skew next instant. The measured flow velocity record was collected by
waves or acceleration-asymmetric waves. means of an A/D converter at 10 millisecond intervals, and the mean
Hsu and Hanes (2004) examined in detail the effects of wave velocity profile variation was obtained by averaging over 50 wave
profile on sediment transport using a two-phase model. They have cycles. According to Sleath (1987) at least 50 wave cycles are needed to
shown that the sheet flow response to flow forcing typical of successfully compute statistical quantities for turbulent condition. A
asymmetric and skewed waves indicates a net sediment transport in schematic diagram of the experimental set-up is shown in Fig. 2.
the direction of wave propagation. However, for a predictive near- The flow-measuring unit comprises of a wind tunnel and one
shore morphological model, a more efficient approach to calculate the component Laser Doppler Velocimeter (LDV) for flow measurement.
bottom shear stress is needed for practical applications. Moreover, Velocity measurements were carried out at 20 points in the vertical
investigation of a more reliable calculation method to estimate the direction at the central part of the wind tunnel. The wind tunnel has a
time-variation of bottom shear stress and that of turbulent boundary length of 5 m and the height and width of the cross-section are 20 cm
layer under saw-tooth wave over rough bed have not been done as yet. and 10 cm, respectively (Fig. 2). These dimensions of the cross-section
Bottom shear stress estimation is the most important step, which is of wind tunnel were selected in order to minimize the effect of
required as an input to the practical sediment transport models. sidewalls on flow velocity. The triangular roughness having a height of
Therefore, the estimation of bottom shear stress from a sinusoidal 5 mm (a roughness height, Hr = 5 mm) and 10 mm width was pasted
wave is of limited value in connection with the sediment transport over the bottom surface of the wind tunnel at a spacing of 12 mm
estimation unless the acceleration effect is incorporated therein. along the wind tunnel, as shown in Fig. 3. Moreover, it was confirmed
In the present study, the characteristics of turbulent boundary layers that the velocity measurement at the center of the roughness and at
under saw-tooth waves are investigated experimentally and numeri- the flaking off region around the roughness has shown a similar flow
cally. Laboratory experiments were conducted in an oscillating tunnel distribution as shown in Jonsson and Carlsen (1976).
over rough bed with air as the working fluid and smoke particles as These roughness elements protrude out of the viscous sub-layer at
tracers. The velocity distributions were measured by means of Laser high Reynolds numbers. This causes a wake behind each roughness
Doppler Velocimeter (LDV). The baseline (BSL) k–ω model proposed by element, and the shear stress is transmitted to the bottom by the
Menter (1994) was also employed to and the experimental data was pressure drag on the roughness elements. Viscosity becomes irrelevant
used for model verification. Moreover, a quantitative comparison
between turbulence model and experimental data was made. A new
calculation method for bottom shear stress is proposed incorporating
both velocity and acceleration terms. In this method a new acceleration
coefficient, ac and a phase difference empirical formula were proposed
to express the effect of wave skew-ness on the bottom shear stress under
saw-tooth waves. The proposed ac constant was determined empirically
from both experimental and the BSL k–ω model results. The new
calculation method of bottom shear stress under saw-tooth wave was
further applied to calculate sediment transport rate induced by skew or
saw-tooth waves. Sediment transport rate was formulated by using the
existing sheet flow sediment transport rate data under skew waves by
Watanabe and Sato (2004). Moreover, the acceleration effect on both the
bottom shear stress and sediment transport under skew waves were
examined. Fig. 1. Definition sketch for saw-tooth wave.
1104 Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112

Table 1
Experimental conditions for saw-tooth waves

Case T (s) Umax (cm/s) v (cm2/s) α am/ks Re S ks/zh


SK1 4.0 398 0.145 0.314 168.9 6.96 × 105 25.3 0.15
SK2 4.0 399 0.147 0.363 169.3 6.89 × 105 25.4 0.15
SK3 4.0 400 0.147 0.406 169.8 6.93 × 105 25.5 0.15
SK4 4.0 400 0.151 0.500 169.8 6.75 × 105 25.5 0.15

2.2. Sediment transport experiment

The experimental data from Watanabe and Sato (2004) for


oscillatory sheet flow sediment transport under skew waves motion
were used in the present study. The flow velocity wave profile was the
acceleration asymmetric or skew wave profile obtained from the time
variations of acceleration of first-order cnoidal wave theory by
Fig. 2. Schematic diagram of experimental set-up. integration with respect to time. These experiments consist of 33
cases. Three values of the wave skew-ness (α) were used; 0.453, 0.400
for determining either the velocity distribution or the overall drag on and 0.320. Moreover, the maximum flow velocity at free stream, Umax
the surface. And the velocity distribution near a rough bed for steady ranges from 0.72 to 1.45 m/s. The sediment median diameters are
flow is logarithmic. Therefore the usual log-law can be used to estimate d50 = 0.20 mm and d50 = 0.74 mm and the wave periods are T = 3.0 s and
the time variation of bottom shear stress το(t) over rough bed as shown T = 5.0 s.
by previous studies e.g., Jonsson and Carlsen (1976), Hino et al. (1983),
Jensen et al. (1989), Fredsøe and Deigaard (1992) and Fredsøe et al. 3. Turbulence model
(1999). Moreover, some previous studies (e.g., Jonsson and Carlsen,
1976; Hino et al., 1983; Sana et al., 2006) also have shown that the For the 1-D incompressible unsteady flow, the equation of motion
values of bottom shear stress computed from the usual log-law and the within the boundary layer can be expressed as
momentum integral methods gave a quite similar, especially by virtue
Au 1 Ap 1 As
of the phase difference in crest and trough values of the shear stress. ¼ þ ð1Þ
At q Ax q Az
Nevertheless, this usual log-law may be under estimated by as much as
20% up to 60% in accelerating flow and overestimated by as much as At the axis of symmetry or outside boundary layer u = U, therefore
20% up to 80% in decelerating flow, respectively, for unsteady flow as
shown by Soulsby and Dyer (1981). The usual log-law should be Au AU 1 As
¼ þ ð2Þ
modified by incorporating velocity and acceleration terms to estimate At At q Az
the bed shear stress for unsteady flow, as given by Soulsby and Dyer
For turbulent flow,
(1981).
Experiments have been carried out for four cases under saw-tooth s Au P
waves. The experimental conditions of present study are given in ¼v  u VvV ð3Þ
q Az
Table 1. The maximum velocity was kept almost 400 cm/s for all the P P
cases. The Reynolds number magnitude defined for each case has The Reynolds stress q uVv V may be expressed as q uVv V¼
sufficed to locate these cases in the rough turbulent regime. Here, v is qvt ðAu=AzÞ, where νt is the eddy viscosity.
the kinematics viscosity, am/ks is the roughness parameter, ks, And Eq. (3) became,
Nikuradse's equivalent roughness defined as ks = 30zo in which zo is
the roughness height, am = Umax/σ, the orbital amplitude of fluid just s Au
ðv þ vt Þ ð4Þ
above the boundary layer, where, Umax, the velocity at wave crest, σ, q Az
the angular frequency, T, wave period, S (=Uo/(σzh)), the reciprocal of
For practical computations, turbulent flows are commonly computed
the Strouhal number, zh, the distance from the wall to the axis of
by the Navier–Stokes equation in averaged form. However, the
symmetry of the measurement section.
averaging process gives rise to the new unknown term representing
the transport of mean momentum and heat flux by fluctuating
quantities. In order to determine these quantities, turbulence models
are required. Two-equation turbulence models are complete turbu-
lence models that fall in the class of eddy viscosity models (models
which are based on a turbulent eddy viscosity are called as eddy
viscosity models). Two transport equations are derived describing
transport of two scalars, for example the turbulent kinetic energy k
and its dissipation ε. The Reynolds stress tensor is then computed
using an assumption, which relates the Reynolds stress tensor to the
velocity gradients and an eddy viscosity. While in one-equation
turbulence models (incomplete turbulence model), the transport
equation is solved for a turbulent quantity (i.e. the turbulent kinetic
energy, k) and a second turbulent quantity is obtained from algebraic
expression. In the present paper the base line (BSL) k–ω model was
used to evaluate the turbulent properties to compare with the ex-
Fig. 3. Definition sketch for roughness. perimental data.
Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112 1105

pffiffiffiffiffiffiffiffiffiffiffiffi
The baseline (BSL) model is one of the two-equation turbulence UT ¼ F js0 j=q is friction velocity and the parameter SR is related to
models proposed by Menter (1994). The basic idea of the BSL k–ω model the grain-roughness Reynolds number, k+s = ks(U ⁎/v),
is to retain the robust and accurate formulation of the Wilcox k–ω model  2
in the near wall region, and to take advantage of the free stream 50 100
SR ¼ for kþ
s b25 and SR ¼ for kþ
s z25 ð12Þ
independence of the k–ε model in the outer part of boundary layer. It kþ
S kþ
s
means that this model is designed to give results similar to those of the
original k–ω model of Wilcox, but without its strong dependency on The instantaneous bottom shear stress can be determined using
arbitrary free stream of ω values. Therefore, the BSL k–ω model gives Eq. (4), in which the eddy viscosity was obtained by solving the
results similar to the k–ω model of Wilcox (1988) in the inner part of transport equation for turbulent kinetic energy k and the dissipation
boundary layer but changes gradually to the k–ε model of Jones and of the turbulent kinetic energy ω in Eq. (7). While, the instantaneous
Launder (1972) towards to the outer boundary layer and the free stream value of u(z,t) and vt can be obtained numerically from Eqs. (1)–(7)
velocity. In order to be able to perform the computations within one set with the proper boundary conditions.
of equations, the Jones–Launder model was first transformed into the k–
ω formulation. The blending between the two regions is done by a 3.2. Numerical method
blending function F1 changing gradually from one to zero in the desired
region. The governing equations of the transport equation for turbulent A Crank–Nicolson type implicit finite-difference scheme was used
kinetic energy k and the dissipation of the turbulent kinetic energy ω to solve the dimensionless non-linear governing equations. In order to
from the BSL model as mentioned before are, achieve better accuracy near the wall, the grid spacing was allowed to
increase exponentially in the cross-stream direction to get fine
   2 resolution near the wall. The first grid point was placed at a distance
Ak A Ak Au
¼ ðv þ vt rkx Þ þ vt bTxk ð5Þ of Δz1 = (r − 1) zh/(rn − 1), where r is the ratio between two consecutive
At Az Az Az
grid spaces and n is total number of grid points. The value of r was
   2 selected such that Δz1 should be sufficiently small in order to maintain
Ax A Ax Au 1 Ak Ax fine resolution near the wall. In this study, the value of Δz1 is given
¼ ðv þ vt rx Þ þg bx2 þ 2ð1  F1 Þrx2 ð6Þ
At Az Az Az x Az Az equal to 0.0042 cm from the wall which correspond to z+ = zU⁎/v = 0.01.
It may be noted that in k–ε model where wall function method is used
From k and ω, the eddy viscosity can be calculated as to describe roughness the first grid point should be lie in the
logarithmic region and corresponding boundary conditions should be
k applied for k and ε. In the k–ω model, as explained before the effect of
vt ¼ ð7Þ
x roughness can be simply incorporated using Eq. (11). In space 100 and
where, the values of the model constants are given as σkω = 0.5, in time 7200 steps per wave cycle were used. The convergence was
β⁎ = 0.09, σω = 0.5, γ = 0.553 and β = 0.075 respectively, and F1 is a achieved through two stages; the first stage of convergence was based
blending function, given as: on the dimensionless values of u, k and ω at every time instant during
a wave cycle. Second stage of convergence was based on the maximum
 
F1 ¼ arg 41 ð8Þ wall shear stress in a wave cycle. The convergence limit was set to
1 × 10− 6 for both the stages.
where,
" rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ! # 4. Mean velocity distributions
k 500v 4rx2 k
arg1 ¼ min max ; 2 ; ð9Þ
0:09xz z x CDkx z2 Mean velocity profiles in a rough turbulent boundary layer under
saw-tooth waves at selected phases were compared with the BSL k–ω
here, z is the distance to the next surface and CDkω is the positive model for the cases SK2 and SK4 presented in Figs. 4 and 5, respectively.
portion of the cross-diffusion term of Eq. (6) defined as
 
1 Ak Ax
CDkx ¼ max 2rx2 ; 1020 ð10Þ
x Az Az

Thus, Eqs. (2), (5) and (6) were solved simultaneously after nor-
malizing by using the free stream velocity, U, angular frequency, σ
kinematics viscosity, ν and zh.

3.1. Boundary conditions

Non slip boundary conditions were used for velocity and turbulent
kinetic energy on the wall (u =k = 0) and at the axis of symmetry of
the oscillating tunnel, the gradients of velocity, turbulent kinetic energy
and specific dissipation rate were equated to zero, (at z =zh, ∂u/∂z = ∂k/
∂z = ∂ω/∂z = 0). The k–ω model provides a natural way to incorporate the
effects of surface roughness through the surface boundary condition.
The effect of roughness was introduced through the wall boundary
condition of Wilcox (1988), in which this equation was originally
recognized by Saffman (1970), given as follow,

xw ¼ UTSR =v ð11Þ

where ωw is the surface boundary condition of the specific dissipation


ω at the wall in which the turbulent kinetic energy k reduces to 0, Fig. 4. Mean velocity distribution for Case SK2 with α = 0.363.
1106 Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112

the model prediction is excellent. A similar result was obtained by Sana


and Shuy (2002) using DNS data for model verification.

5. Prediction of turbulence intensity

The fluctuating velocity in x-direction u' can be approximated


using Eq. (13) that is a relationship derived from experimental data for
steady flow by Nezu (1977),
pffiffiffi
uV¼ 1:052 k ð13Þ

where k is the turbulent kinetic energy obtained in the turbulence


model.
Comparison made on the basis of approximation to calculate the
fluctuating velocity by Nezu (1977) may not be applicable in the whole
range of cross-stream dimension since it is based on the assumption of
isotropic turbulence. This assumption may be valid far from the wall,
where the flow is practically isotropic, whereas the flow in the region
near the wall is essentially non-isotropic. The BSL k–ω model can
predict very well the turbulent intensity across the depth almost all at
phases, but, near the wall underestimates at phases A, C, D and E (Case
Fig. 5. Mean velocity distribution for Case SK4 with α = 0.500. SK2) and at phases A, C, D, E and H (Case SK3) as shown in Figs. 6 and 7,
respectively. However, the model qualitatively reproduces the
turbulence generation and mixing-processes very well.
The solid line showed the turbulence model prediction while open and
6. Bottom shear stress
closed circles showed the experimental data for mean velocity profile
distribution. The experimental data and the turbulence model show that
6.1. Experimental Results
the velocity overshoot is much influenced by the effect of acceleration
and the velocity magnitude. The difference of the acceleration between
Bottom shear stress is estimated by using the logarithmic velocity
the crest and trough phases is significant. The velocity overshooting is
distribution given in Eq. (14), as follows,
higher in the crest phase than the trough as shown at phase B and F for
Case SK2 (α = 0.363). As expected this difference is not visible for  
U⁎ z
symmetric case (Case SK4) (α = 0.500). Moreover, the asymmetry of the u¼ ln ð14Þ
j z0
flow velocity can be observed in phase A and E. Due to the higher
acceleration at phase A the velocity overshooting is more distinguished where, u is the flow velocity in the boundary layer, κ is the von
in the wall vicinity. Karman's constant (= 0.4), z is the cross-stream distance from
The BSL k–ω model could predict the mean velocity very well in the theoretical bed level (z = y + Δz) (Fig. 3). For a smooth bottom zo = 0,
whole wave cycle of asymmetric case. Moreover, it predicted the velocity but for rough bottom, the elevation of theoretical bed level is not a
overshooting satisfactorily (Fig. 4). For symmetrc case (Case SK4) as well single value above the actual bed surface. The value of zo for the fully

Fig. 6. Turbulent intensity comparison between BSL k–ω model prediction and experimental data for Case SK2.
Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112 1107

Fig. 7. Turbulent intensity comparison between BSL k–ω model prediction and experimental data for Case SK3.

rough turbulent flow is obtained by extrapolation of the logarithmic ing with acceleration effect. The increase in wave skew-ness causes an
velocity distribution above the bed to the value of z = zo where u increase the asymmetry of bottom shear stress. The wave without
vanishes. The temporal variations of Δz and zo are obtained from the skew-ness shows a symmetric shape, as seen in Case SK4 for α = 0.500
extrapolation results of the logarithmic velocity distribution on the (Fig. 8).
fitting a straight line of the logarithmic distribution through a set of
velocity profile data at the selected phases angle for each case. These 6.2. Calculation methods of bottom shear stress
obtained values of Δz and zo are then averaged to get zo = 0.05 cm for
all cases and Δz = 0.015 cm, Δz = 0.012 cm, Δz = 0.023 cm and 6.2.1. Existing methods
Δz = 0.011 cm, for Case SK1, Case SK2, Case SK3 and Case SK4, There are two existing calculation methods of bottom shear stress
respectively. The bottom roughness, ks can be obtained by applying for non-linear wave boundary layers. The maximum bottom shear
the Nikuradse's equivalent roughness in which zo = ks/30. By plotting u stress within a basic harmonic wave-cycle modified by the phase
against ln(z/z0), a straight line is drawn through the experimental difference is proposed by Tanaka and Samad (2006), as follows:
data, the value of friction velocity, U⁎ can be obtained from the slope
of this line and bottom shear stress, τo can then be obtained. The u
1
so t  ¼ qfw U ðt ÞjU ðt Þj ð15Þ
obtained value of Δz and zo as the above mentioned has a sufficient r 2
accuracy for application of logarithmic law in a wide range of velocity Here τo(t), the instantaneous bottom shear stress, t, time, σ, the
profiles near the bottom. Suzuki et al. (2002) have given the details of angular frequency, U(t) is the time history of free stream velocity, φ is
this method and found good accuracy. phase difference between bottom shear stress and free stream velocity
Fig. 8 shows the time-variation of bottom shear stress under saw- and fw is the wave friction factor. This method is referred as Method 1
tooth waves with the variation in the wave skew-ness parameter α. It in the present study.
can be seen that the bottom shear stress under saw-tooth waves has
an asymmetric shape during crest and trough phases. The asymmetry
of bottom shear stress is caused by wave skew-ness effect correspond-

Fig. 8. The time-variation of bottom shear stress under saw-tooth waves. Fig. 9. Calculation example of acceleration coefficient, ac for sawtooth wave.
1108 Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112

the BSL k–ω model results of bottom shear stress using following
relationship:
pffiffiffiffiffiffiffiffiffiffi  
U ⁎ðt Þ  fw =2U t þ ur
ac ðt Þ ¼ pffiffiffiffiffiffiffi ð19Þ
fw =2 AU ðt Þ
r At

Fig. 9 shows an example of the temporal variation of the accel-


eration coefficient ac(t) for α = 0.300 based on the numerical com-
putations. The results of averaged value of acceleration coefficient ac
from both experimental and numerical model results as function of
the wave skew-ness parameter, α are plotted in Fig. 10. Hereafter, an
equation based on regression line to estimate the acceleration
coefficient ac as a function of α is proposed as:

Fig. 10. Acceleration coefficient ac as function of α.


ac ¼ 036 ln ðaÞ  0:249 ð20Þ
Nielsen (2002) proposed a method for the instantaneous wave
friction velocity, U⁎(t) incorporating the acceleration effect, as follows: The increase in the wave skew-ness (or decreasing the value of α)
rffiffiffiffiffi  brings about an increase in the value of acceleration coefficient, ac. For
fw sin u AU the symmetric wave where α = 0.500, the value of ac is equal to zero. In
U⁎ðt Þ ¼ cos uU ðt Þ þ ð16Þ
2 r At others words the acceleration term is not significant for calculating
the bottom shear stress under symmetric wave. Therefore, for
so ðt Þ ¼ qU⁎ðt ÞjU⁎ðt Þj ð17Þ sinusoidal wave Method 3 yields the same result as Method 1.

This method is based on the assumption that the steady flow


component is weak (e.g. in a strong undertow, in a surf zone, etc.). This
method is termed as Method 2 here. It seems reasonable to derive the
το(t) from u(t) by means of a simple transfer function based on the
knowledge from simple harmonic boundary layer flows as has been
done by Nielsen (1992).

6.2.2. Proposed method


The new calculation method of bottom shear stress under saw-
tooth waves (Method 3) is based on incorporating velocity and
acceleration terms provided through the instantaneous wave friction
velocity, U⁎(t) as given in Eq. (18). Both velocity and acceleration terms
are adopted from the calculation method proposed by Nielsen (1992,
2002) (Eq. (16)). The phase difference was determined from an
empirical formula for practical purposes. In the new calculation
method a new acceleration coefficient, ac is used expressing the wave
skew-ness effect on the bottom shear stress under saw-tooth waves,
that is determined empirically from both experimental and BSL k–ω
model results. The instantaneous friction velocity, can be expressed as:
qffiffiffiffiffiffiffiffiffiffi 
u
ac AU ðt Þ
U⁎ðt Þ ¼ fw =2 U t þ þ ð18Þ
r r At

Here, the value of acceleration coefficient ac is obtained from the


average value of ac(t) calculated from experimental result as well as

Fig. 12. Comparison among the BSL k–ω model, calculation methods and experimental
Fig. 11. Phase difference between the bottom shear stress and the free stream velocity. results of bottom shear stress, for Case SK1.
Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112 1109

α = 0.500 in Eq. (24) yields the same result as Eq. (22). As seen in Fig. 11
the phase difference at crest, trough and average between crest and
trough for Case SK4 with α = 0.500 is about 19.1°, this value agrees well
with the result obtained from Eq. (22) as well as Eq. (24) for α = 0.500.
The increase in the wave skew-ness or decreasing α causes the
average value of phase difference in experimental results to gradually
decrease as shown in Fig. 11.

6.3. Comparison for bottom shear stress

In the previous section it has been shown that the bottom shear
stress under saw-tooth waves has an asymmetric shape in both wave
crest and trough phases. The increase in wave skew-ness causes an
increase in the asymmetry of bottom shear stress under saw-tooth
waves. Figs. 12, 13, 14 and 15 show a comparison among the BSL k–ω
model, three calculation methods and experimental results of bottom
shear stress under saw-tooth waves, for Case SK1, Case SK2, Case SK3
and Case 4, respectively.
Method 3 has shown the best agreement with the experimental
results along a wave cycle for all saw-tooth wave cases. Method 2
slightly underestimated the bottom shear stress during acceleration
phase for the higher wave skew-ness (Case SK1) as shown in Fig. 12.
While, it overestimated the same in the crest phase for Case SK2 and
SK3 as shown in Figs. 13 and 14, and in the trough phase for Case SK4
as shown in Fig. 15.

Fig. 13. Comparison among the BSL k–ω model, calculation methods and experimental
results of bottom shear stress, for Case SK2.

6.2.3. Wave friction factor and phase difference


The wave friction coefficient proposed by Tanaka and Thu (1994)
was used in all the calculation methods in the present study as
follows:
(  0:100 )
am
fw ¼ exp 7:53 þ 8:07 ð21Þ
zo

1 þ 0:00279C 0:357
us ¼ 42:4C 0:153 ðdegreeÞ ð22Þ
1 þ 0:127C 0:563

0:111 1
for smooth : C ¼ ; for rough : C ¼ qffiffiffiffi ð23Þ
j f2w Re j f2w azm0

u ¼ 2aus ðdegreeÞ ð24Þ

Where, φs is phase difference between free stream velocity and


bottom shear stress proposed by Tanaka and Thu (1994) based on
sinusoidal wave study and C defined by Eq. (23).
Fig. 11 shows the phase difference obtained from measured data
under saw-tooth waves, as well as from theory proposed by Tanaka
and Thu (1994) in Eq. (22) for sinusoidal wave. The wave skew-ness Fig. 14. Comparison among the BSL k–ω model, calculation methods and experimental
effect under saw-tooth waves was included using Eq. (24). A value of results of bottom shear stress, for Case SK3.
1110 Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112

Here, Φ(t) is the instantaneous dimensionless sediment transport


rate, ρs is density of the sediment, g is gravitational acceleration, d50 is
median diameter of sediment, A is a coefficient, τ⁎(t) is the Shields
parameter defined by (το(t) / (((ρs/ρ) − 1)gd50)) in which το(t) is the
instantaneous bottom shear stress calculated from both Method 1 and
Method 3. While τ⁎cr is the critical Shields number for the initiation of
sediment movement (Tanaka and To, 1995).
 
s⁎cr ¼ 0:055 1  exp 0:09S ⁎0:58 þ 0:09S0:72
⁎ ð26Þ

Where, S⁎ is dimensionless particle size defined as:


qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðqs =q  1Þgd350
S⁎ ¼ ð27Þ
4v
The net sediment transport rate, qnet, which is averaged over one-
period is expressed in the following expression according to Eq. (25).

qnet
U ¼ AF ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð28Þ
ðqs =q  1Þgd350

Z T
1
F¼ signfs⁎ðt Þgjs⁎ðt Þj0:5 fjs⁎ðt Þj  s⁎cr gdt ð29Þ
T 0

Here, Φ is the dimensionless net sediment transport rate, F is a


function of Shields parameter and qnet is the net sediment transport
rate in volume per unit time and width. Moreover, the integration of
Eq. (29) is assumed to be done only in the phase |τ⁎(t)| N τcr⁎ and
during the phase |τ⁎(t)| b τcr⁎ the function of integration is assumed to
be 0.
Sheet-flow condition occurs when the tractive force exceeds a
certain limit, sand ripples disappear, replaced by a thin moving layer
of sand in high concentration. Many researchers have shown that the
characteristic of Nikuradse's roughness equivalent (ks) may be defined
to be proportional to a characteristic grain size for evaluating the
friction factor. For sheet-flow sediment transport ks = 2.5 d50 as shown
by Swart (1974), Nielsen (2002) and Nielsen and Callaghan (2003).
Therefore, in the present study the same relationship is used to
Fig. 15. Comparison among the BSL k–ω model, calculation methods and experimental
results of bottom shear stress, for Case SK4. formulate the sheet-flow sediment transport rate under skew wave.
First of all, the wave velocity profile, U(t) which was obtained from
the time variation of acceleration of first order cnoidal wave theory by
integrating with respect to time as in the experiment by Watanabe and
As expected, Method 1 yielded a symmetric value of the bottom Sato (2004). The bottom shear stress calculated from Method 1 was
shear stress at the crest and trough part for all the cases of saw-tooth substituted into Eq. (29) and the result is shown in Fig. 16 by open
waves. Moreover, the BSL k–ω model results showed close agreement symbols. As expected that Method 1 yields a net sediment transport rate
with the experimental data and Method 3 results. Therefore, Method 3
can be considered as a reliable calculation method of bottom shear
stress under saw-tooth waves for all cases.
It can be concluded that the proposed method (Method 3) for
calculating the instantaneous bottom shear stress under saw-tooth
waves has a sufficient accuracy.

7. Application to the net sediment transport induced by skew waves

7.1. Sediment transport rate formulation

The proposed calculation method of bottom shear stress is further


applied to formulate the sheet-flow sediment transport rate under
skew wave using the experimental data by Watanabe and Sato (2004).
At first, the instantaneous sheet flow sediment transport rate q(t) is
expressed as a function of the Shields number τ⁎(t) as given below:

qðt Þ
Uðt Þ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ A signfs⁎ðt Þgjs⁎ðt Þj0:5 fjs⁎ðt Þj  s⁎cr g ð25Þ
ðqs =q  1Þgd350
Fig. 16. Formulation of sediment transport rate under skew waves.
Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112 1111

Fig. 17. The relation between the net sediment transport rates and Umax in variation of α Fig. 19. Comparison of experimental and calculation result of the net sediment transport
for T = 3 s and d50 = 0.20 mm. rates in variation of maximum velocity Umax and the wave skew-ness α for
d50 = 0.20 mm and T = 5 s.
to be zero, because the integral value of F for a complete wave cycle is
zero. In other word, it can be concluded that (Method 1) is not suitable bottom shear stress and consequently yields a higher net sediment
for calculating the net sediment transport rate under skew waves. transport rate (Fig. 17).
Furthermore, the relation between F and the dimensionless net Onshore and offshore sediment transport rate is shown in Fig. 18
sediment transport rate (Φ) obtained by the proposed method along with the net sediment transport. In this figure the values of
(Method 3) is shown in Fig. 16 by closed symbols. Since of the Umax, T and d50 are fixed and only α has been changed. As obvious for a
acceleration effect has been included in this calculation method (Eq. wave profile without skew-ness (α = 0.500) the amount of onshore
(18)), which causes the bottom shear stress at crest differ from that at sediment transport is equal to that in offshore direction, therefore the
trough, and therefore yields a net positive or negative value of F from net sediment transport rate is zero. The difference between the
Eq. (29). A linear regression curve is also shown in with the value of onshore and the offshore sediment transport becomes more promi-
A = 11 (Eq. (28)). nent due to an increase in the wave skew-ness and thus causing in a
significant increase the net sediment transport.
7.2. Net sediment transport by skew waves A similar comparison is made for another of experimental
condition for T = 5 s and d50 = 0.20 mm in Fig. 19.
The characteristics of the net sediment transport induced by skew Recently, Nielsen (2006) applied an extension of the domain filter
waves are studied using the present calculation method for bottom method developed by Nielsen (1992) to evaluate the effect of
shear stress (Method 3) and the experimental data for the sheet flow acceleration skew-ness on the net sediment transport based on the
sediment transport rate from Watanabe and Sato (2004). Fig. 17 shows data of Watanabe and Sato (2004). A good agreement between
a comparison between the experimental data and calculations based calculated and experimental data of the net sediment transport was
on Method 3 for the net sediment transport rates, qnet and maximum found using φ = 51°, a value much different from the usual notion that
velocity, Umax for the wave period T = 3 s and the median diameter of the phase difference is of the order of 10o for rough turbulent wave
sediment particle d50 = 0.20 mm along with the wave skew-ness boundary layers.
parameter (α). It is clear that an increase in the wave skew-ness and Figs. 20 and 21 show the correlation of the net sediment transport
the maximum velocity produces an increase in the net sediment experimental data from Watanabe and Sato (2004) and the net
transport rate depicted in both experimental data and calculation
results. The proposed method shows very good agreement with the
data with minor differences. However, the present model has a
limitation that does not simulate the sediment suspension. As
mentioned previously higher wave skew-ness produces a higher

Fig. 20. Correlation of the net sediment transport experimental data from Watanabe
Fig. 18. Change in amount of sediment transport rate according to an increasing α. and Sato (2004) and the net sediment transport calculated by the present model.
1112 Suntoyo et al. / Coastal Engineering 55 (2008) 1102–1112

Acknowledgments

The first author is grateful for the support provided by Japan


Society for the Promotion of Science (JSPS), Tohoku University, Japan
and Institut Teknologi Sepuluh Nopember (ITS), Surabaya, Indonesia
for completing this study. This research was partially supported by
Grant-in-Aid for Scientific Research from JSPS (No. 18006393).

References

Dick, J.E., Sleath, J.F.A., 1991. Velocities and concentrations in oscillatory flow over beds
of sediment. Journal of Fluids Mechanics 233, 165–196.
Fredsøe, J., Deigaard, R., 1992. Mechanics of coastal sediment transport. Advanced Series
on Ocean Engineering, vol. 3. World Scientific Publication.
Fredsøe, J., Andersen, K.H., Sumer, B.M., 1999. Wave plus current over a ripple-covered
bed. Coastal Engineering 38, 177–221.
Gonzalez-Rodriguez, D., Madsen, O.S., 2007. Seabed shear stress and bedload transport
due to asymmetric and skewed waves. Coastal Engineering 54 (12), 914–929.
Hino, M., Kashiwayanag, M., Nakayama, A., Nara, T., 1983. Experiments on the
turbulence statistics and the structure of a reciprocating oscillatory flow. Journal
of Fluid Mechanics 131, 363–400.
Hsu, T.J., Hanes, D.M., 2004. Effects of wave shape on sheet flow sediment transport.
Fig. 21. Correlation of the net sediment transport experimental data from Watanabe and Journal of Geophysical Research 109 (C05025). doi:10.1029/2003JC002075.
Sato (2004) and the net sediment transport calculated by Nielsen's model (2006). Jensen, B.L., Sumer, B.M., Fredsøe, J., 1989. Turbulent oscillatory boundary layers at high
Reynolds numbers. Journal of Fluid Mechanics 206, 265–297.
Jones, W.P., Launder, B.E., 1972. The prediction of laminarization with a two-equation
model of turbulence. International Journal of Heat and Mass Transfer 15, 301–314.
Jonsson, I.G., Carlsen, N.A., 1976. Experimental and theoretical investigations in an
sediment transport calculated by Nielsen's model (2006) and by the oscillatory turbulent boundary layer. Journal of Hydraulic Research 14 (1), 45–60.
present model, respectively. The present method shows a slightly King, D.B., 1991. Studies in oscillatory flow bed load sediment transport. PhD Thesis,
better correlation than Nielsen's model (2006) with a reasonable University of California, San Diego, USA.
Menter, F.R., 1994. Two-equation eddy-viscosity turbulence models for engineering
value of the phase difference (φ ranges from 9.6° to 16.5°). The model
applications. AIAA Journal 32 (8), 1598–1605.
performance is indicated by the coefficient of determination. The Nezu, I., 1977. Turbulent structure in open channel flow. Ph.D Dissertation, Kyoto
present model shows the coefficient of determination (R2 = 0.655), University, Japan.
Nielsen, P., 1992. Coastal bottom boundary layers and sediment transport. Advanced
which higher than that for Nielsen's model as (R2 = 0.557). Although
Series on Ocean Engineering, vol. 4. World Scientific Publication.
the present model is marginally better than the Nielsen's model Nielsen, P., 2002. Shear stress and sediment transport calculations for swash zone
(2006), the present model used a more realistic value of the phase modeling. Coastal Engineering 45, 53–60.
difference obtained from well-established formula. Nielsen, P., 2006. Sheet flow sediment transport under waves with acceleration
skewness and boundary layer streaming. Coastal Engineering 53, 749–758.
Nielsen, P., Callaghan, D.P., 2003. Shear stress and sediment transport calculations for
8. Conclusions sheet flow under waves. Coastal Engineering 47, 347–354.
Saffman, P.G., 1970. Dependence on Reynolds number of high-order moments of
velocity derivatives in isotropic turbulence. Physics Fluids 13, 2192–2193.
The characteristics of the turbulent boundary layer under saw- Sana, A., Tanaka, H., 2000. Review of k − e model to analyze oscillatory boundary layers.
tooth waves were studied using experiments and the BSL k–ω Journal of Hydraulic Engineering 126 (9), 701–710.
turbulence model. The mean velocity distributions under saw-tooth Sana, A., Shuy, E.B., 2002. Two-equation turbulence models for smooth oscillatory
boundary layers. Journal of Waterway, Port, Coastal and Ocean Engineering 128 (1),
waves show different characteristics from those under sinusoidal 38–45.
waves. The velocity overshooting is much influenced by the effect of Sana, A., Tanaka, H., Yamaji, H., Kawamura, I., 2006. Hydrodynamic behavior of asymmetric
acceleration and the velocity magnitude. The velocity overshooting oscillatory boundary layers at low Reynolds numbers. Journal of Hydraulic Engineering
132 (10), 1086–1096.
has different appearance in the crest and trough phases caused by the
Schäffer, A.H., Svendsen, I.A., 1986. Boundary layer flow under skew waves. Inst. Hydro-
difference of acceleration. The BSL k–ω model shows a good dynamics and Hydraulic Engineering, Tech. Univ. Denmark, Prog. Report, vol. 64, pp.
agreement with all the experimental data for saw-tooth wave 13–33.
Sleath, J.F.A., 1987. Turbulent oscillatory flow over rough beds. Journal of Fluid Mechanics
boundary layer by virtue of velocity and turbulence kinetics energy
182, 369–409.
(T.K.E). The model prediction far from the bed is generally good, while Soulsby, R.L., Dyer, K.R., 1981. The form of the near-bed velocity profile in a tidally
near the bed some discrepancies were found for all the cases. accelerating flow. Journal of Geophysical Research 86 (C9), 8067–8074.
A new calculation method for calculating bottom shear stress Suzuki, T., Tanaka, H., Yamaji, H., 2002. Investigation of rough bottom boundary layer
under irregular waves. Annual Journal of Hydraulic Engineering 46, 869–874 (in
under saw-tooth waves has been proposed based on velocity and Japanese).
acceleration terms where the effect of wave skew-ness is incorporated Swart, D.H., 1974. Offshore Sediment Transport and Equilibrium Beach Profile. Delft
using a factor ac, which is determined empirically from experimental Hydraulics Laboratory Publication, No. 131.
Tanaka, H., 1988. Bed load transport due to non-linear wave motion. Proceedings of 21st
data and the BSL k–ω model results. The new method has shown the International Conference on Coastal Engineering, ASCE, Malaga, Spain, pp. 1803–1817.
best agreement with the experimental data along a wave cycle for all Tanaka, H., Thu, A., 1994. Full-range equation of friction coefficient and phase difference
saw-tooth wave cases in comparison with the existing calculation in a wave-current boundary layer. Coastal Engineering 22, 237–254.
Tanaka, H., To, D.V., 1995. Initial motion of sediment under waves and wave-current
methods. combined motions. Coastal Engineering 25, 153–163.
The new calculation method of bottom shear stress (Method 3) Tanaka, H., Samad, M.A., 2006. Prediction of instantaneous bottom shear stress for
was applied to the net sediment transport experimental data under turbulent plane bed condition under irregular wave. Journal of Hydraulic Research
44 (1), 94–106.
sheet flow condition by Watanabe and Sato (2004) and a good
Tanaka, H., Chian, C.S., Shuto, N., 1983. Experiments on an oscillatory flow accompanied
agreement was found. with a unidirectional motion. Coastal Engineering in Japan 26, 19–37.
The inclusion of the acceleration effect in the calculation of bottom Watanabe, A., Sato, S., 2004. A sheet-flow transport rate formula for asymmetric,
forward-leaning waves and currents. Proc. of 29th ICCE, ASCE, pp. 1703–1714.
shear stress has significantly improved the net sediment transport
Wilcox, D.C., 1988. Reassessment of the scale-determining equation for advanced
calculation under skew waves. It is envisaged that the new calculation turbulent models. AIAA Journal 26 (11), 1299–1310.
method may be used to calculate the net sediment transport rate
under rapid acceleration in surf zone in practical applications, thus
improving the accuracy of morphological models in real situations.

Das könnte Ihnen auch gefallen