This article was downloaded by: [UOV University of Oviedo]

On: 31 October 2014, At: 11:23

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Hydraulic Research

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tjhr20

Review of seepage effects on turbulent open-

channel flow and sediment entrainment
a b b
Yan Lu , Yee-Meng Chiew & Nian-Sheng Cheng
a
Nanjing Hydraulic Research Institute, State Key Laboratory of Hydrology,Water
Resources and Hydraulic Engineering , Nanjing , 210029 , China
b
School of Civil and Environmental Engineering , Nanyang Technological University ,
Nanyang Avenue, Singapore , 639798
Published online: 26 Apr 2010.

To cite this article: Yan Lu , Yee-Meng Chiew & Nian-Sheng Cheng (2008) Review of seepage effects on turbulent open-
channel flow and sediment entrainment, Journal of Hydraulic Research, 46:4, 476-488, DOI: 10.3826/jhr.2008.2942

To link to this article: http://dx.doi.org/10.3826/jhr.2008.2942

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of
the Content. Any opinions and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of information. Taylor and Francis shall
not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other
liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Journal of Hydraulic Research Vol. 46, No. 4 (2008), pp. 476–488
doi:10.3826/jhr.2008.2942
© 2008 International Association of Hydraulic Engineering and Research

Review of seepage effects on turbulent open-channel flow

and sediment entrainment
Examen des effets de l’infiltration sur les écoulements turbulents en canal
à surface libre et sur l’entraînement des sédiments
YAN LU, Postdoctoral Fellow, Nanjing Hydraulic Research Institute, State Key Laboratory of Hydrology,
Water Resources and Hydraulic Engineering, Nanjing 210029, China

YEE-MENG CHIEW, Professor, School of Civil and Environmental Engineering, Nanyang Technological University,
Nanyang Avenue, Singapore 639798
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014

NIAN-SHENG CHENG, (IAHR Member), Associate Professor, School of Civil and Environmental Engineering, Nanyang
Technological University, Nanyang Avenue, Singapore 639798

ABSTRACT
This paper presents a review on the state-of-the-art knowledge of how seepage affects the turbulence characteristics in open-channel flow and its
implication on sediment entrainment. Published literature shows that some effects have been intensively examined and the results are well known,
such as seepage effects on mean flow velocity distributions. Understanding of the other effects remains rudimentary, such as variations of turbulence
intensity and bed shear stresses. In fact, many of these issues remain ambiguous with contradicting inferences and conclusions. For example, the
published literature is still not unanimous as to how turbulence intensities, bed shear stresses and bed particle stability change in the presence of
seepage. By reviewing literature in this area published over the past 35 years, this paper highlights the main conflicting results and attempts to explain
these deviations with certain recommendations.

RÉSUMÉ
Cet article passe en revue l’état de l’art des connaissances sur la manière dont l’infiltration affecte les caractéristiques de la turbulence des écoulements
en canal ouvert et son implication sur l’entraînement de sédiment. La littérature éditée prouve que certains effets ont été intensivement examinés et
les résultats sont bien connus, comme les effets de l’infiltration sur des distributions moyennes de vitesse. La compréhension des autres effets reste
rudimentaire, comme les variations d’intensité de la turbulence et des efforts de cisaillement de lit. En fait, beaucoup de questions restent ambigües
avec des inférences et des conclusions contradictoires. Par exemple, la littérature éditée n’est toujours pas unanime quant à la façon dont les intensités
de turbulence, les efforts de cisaillement et la stabilité des particules de lit changent en présence d’infiltration. En passant en revue la littérature éditée
sur le sujet au cours des trente-cinq dernières années, le présent article met en lumière les principaux résultats contradictoires et tente d’expliquer ces
écarts en donnant certaines recommandations.

Keywords: Open channel flow, sediment, seepage, shear stress, turbulence, transport

1 Introduction as compared to that with an impermeable boundary, as is often

Natural streams often comprise permeable beds in the form of
sand particles and gravels. An essential and well-known feature of
permeable boundaries is that mass and momentum transfer takes
place across the interface between the fluid and porous media.
Interactions between flow in the main stream and the ambient
groundwater, or hyporheic exchanges, are important because of
their role in controlling the transport of contaminants and main-
taining a healthy stream ecosystem (Brunke and Gonser, 1997;
Jones and Mulholland, 2000). The hydrodynamic process in such
a physical system may result in changes to the structural features
of the flow, implying modification to flow-resistance, sediment
entrainment characteristics and morphology of the streambed,
encountered in a laboratory study.
With the level of the free surface in a river being different
from the adjoining groundwater table, two typical seepage flows
(suction and injection) may occur through the river boundary. In
the case of suction or downward seepage, water seeps out of the
river; while with injection or upward seepage, the river receives
additional water.
In most cases, seepage velocity within the porous bound-
ary is small in comparison to the free-stream velocity above
the bed. However, this small seepage flow can effect an addi-
tional hydrodynamic force on the bed sediment. This, in turn,
will significantly affect the process of sediment transport, such
as sediment entrainment and formation of bed features. For

Revision received January 8, 2007/Open for discussion until February 28, 2009.

476
 Journal of Hydraulic Research Vol. 46, No. 4 (2008) Seepage effects on turbulent open-channel flow and sediment entrainment 477

Table 1 Summary of results of previous studies

Source Flume/seepage zone dimensions Sediment transport rate Bed shear stress Turbulence intensity

Length (m) Width (m) Suction Injection Suction Injection Suction Injection

Watters and Rao (1971) 4.6/1.5 0.7/0.7 Increase Decrease — — Decrease Increase
Oldenziel and Brink (1974) 15/4 0.5/0.5 Decrease Increase — — — —
Willetts and Drossos (1975) 3.6/0.125 0.076/0.076 Increase — — — — —
Nezu (1977) — — Little effect Little effect — — Decrease Increase
Richardson et al. (1985) 9.45/3 0.6/0.6 Decrease Increase — — — —
Maclean and Willetts (1986) 5/5 0.076/0.076 — — Increase — — —
Maclean (1991a) 12/0.28 0.3/ 0.215 — — Increase — — —
Maclean (1991b) 5/0.13 0.075/0.075 Increase — Increase — — —
Ramakrishna Rao et al. (1994) 14.16/12.75 0.615/0.615 — — Increase/ Increase/ — —
Decrease Decrease
Antonia and Zhu (1995) 2.6/0.04 0.6/0.6 — — — — Decrease —
Cheng (1997) 7.6/0.5 30/2 0.21/0.21 0.7/0.7 — Increase — Decrease — Increase
Cheng and Chiew (1998a) 30/2 0.7/0.7 — — — Decrease — Increase
Cheng and Chiew (1998b) 30/2 0.7/0.7 — — — Decrease — —
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014

Cheng and Chiew (1999) 7.6/0.5 0.21/ 0.21 — Increase — Decrease — —

Ramakrishna Rao and 3.6/2.4 0.158/0.158 Increase Decrease Increase/ Increase/ Increase Decrease
Nagaraj (1999) Decrease Decrease
Krogstad and Kourakine (2000) 3.5/0.12 0.46/0.46 — — — Decrease — Increase
Chen and Chiew (2004a) 30/2 0.7/0.7 — — Increase — — —
Chen and Chiew (2004b) 30/2 0.7/0.7 — — — — Decrease Increase

instance, a water intake, when drawing faster-moving flow
towards the sediment bed, can increase the local bed shear
stresses, resulting in bed erosion around the intake structure
(Willetts and Drossos, 1975). Other practical examples relate
to hydraulic structures used for channel stabilization, bank ero-
sion prevention and selection of water intake location. More
complex situations occur in the coastal environments, for exam-
ple in swash zone, where seepage varies both temporally and
spatially (Turner, 1995; Karambas, 2003). Additionally, suction
and injection have been encountered in other engineering prac-
tices, such as boundary layer control, drying processes in the
chemical industry and turbine-blade cooling. Knowledge of flows
over permeable boundaries clearly is not only useful for funda-
mental research, but also has practical importance in engineering
applications.
Despite the important influence of seepage on many aspects of
practical engineering, relevant information in this area remains
scarce. Moreover, results from recent research efforts in this
area are often inconclusive and sometimes conflicting. Table 1,
which summarizes the important recent findings over the past
35 years, shows these inconsistencies. The summary in the table
only contains findings that are relevant to matters discussed in
this paper.
The objective of this paper is to review the state-of-the-
art knowledge concerning open-channel flow and cohesionless
sediment entrainment with seepage. The paper first presents pub-
lished results on the variation of flow velocity and turbulence in
the presence of bed seepage. Changes in bed shear stresses due
to seepage and their implication on sediment entrainment is then
examined. Conflicting results are highlighted, and the reasons for
differences are discussed.
2 Velocity distributions subject to seepage
Analyses of the complex processes of contaminants transport in a
permeable streambed require detailed prediction of flows through
the porous bed. Many experimental and numerical studies have
been done on flow through such a medium (Ruff and Gelhar,
1972; Shimizu et al., 1990; Zhou and Mendoza, 1993). To be
consistent with the objective of this study, emphasis is placed on
previous studies of seepage effects on open-channel flows.
The well-established law of the wall was originally derived
for the inner layer of turbulent shear flows, and is widely used in
boundary layer, pipe and open-channel flows. In the presence of
seepage, the flow structure near the boundary and thus the dis-
tribution of the streamwise velocity are significantly modified.
For example, turbulent momentum transfer across the stream–
groundwater interface induces a non-zero velocity at the bed
surface (Zhou and Mendoza, 1993). The induced velocity at
the bed surface is termed “slip velocity”, which invalidates the
no-slip condition that is commonly used in fluid mechanics for
impermeable beds.
In the presence of boundary mass transfer, velocity profiles
have been reported to be different from the logarithmic law in the
field of aerodynamics. Various analytical models have been pro-
posed, for example Clarke et al. (1955), Willetts and Drossos
(1975) and Schlichting (1979). Direct numerical simulations
 478 Y. Lu et al. Journal of Hydraulic Research Vol. 46, No. 4 (2008)

and experimental studies have been performed for testing bed 40

Measurements
injection or suction in a turbulent boundary layer (Antonia and
Zhu, 1995; Krogstad and Kourakine, 2000;Yoda and Westerweel, 30
Log-law of the wall Eq(1)

2001; Kim and Sung, 2003). Velocity measurements over the Modified log-law with injection Eq(3)
seepage zone in open-channel flow have been performed by

u/u∗
Maclean (1991a), Cheng and Chiew (1998a, b) and Chen and 20

Chiew (2004a).
For a two-dimensional flow over a horizontal bed in the case of
10
zero seepage, that is flows over an impermeable bed, the velocity
distribution can be described by the well-known law of the wall
(Yalin, 1977) as: 0
3 4 5 6 7
u 1 y
= ln (1) ln(y/y0)
u∗ κ y0
√ Figure 2 Comparison of experimental data with modified log-law with
where u is the streamwise velocity; u∗ = τb /ρ, the friction injection (Cheng and Chiew, 1998b)
velocity; τb , the boundary shear stress; ρ, the mass density of
fluid; κ, the von Karman constant and y0 , the zero-velocity
displacement measured from the theoretical bed surface. If an streamwise extent (0.12 m) on a turbulent boundary layer. They
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014

equivalent roughness height ks is used, y0 can be expressed as: reported that injection causes an upward shift in the mean velocity
log-law, which is in good agreement with the finding of Cheng
y0 = ks exp(−κB) (2) and Chiew (1998b).
where ks can be taken as 2D50 ; D50 is the median diameter of Substituting Eq. (2) into Eq. (3) leads to:
sediment particles and B, the roughness function, being taken as
2
u 1 y vs 1 y
8.5 for a completely rough flow (Raudkivi, 1990). = ln + B + ln + B (4)
u∗ κ ks 4u∗ κ ks
It is of interest to note that with the assumption of uniform
2.1 Flow with injection suction, Eq. (4) was also derived by Schlichting (1979) using
When injection is applied to an open-channel flow, the resulting Prandtl’s mixing-length theory. In fact, this equation can be
velocity distribution over a flat bed is schematically shown in regarded as a generalization of Eq. (1), the latter being applica-
Fig. 1 (Nezu, 1977), implying a reduction in the streamwise flow ble only for impermeable beds. The experimental data by Rotta
velocity in the region close to the bed. Under this condition, a (1966) agrees well with Eq. (4) in the case of both injection and
modified logarithmic law was proposed by Cheng and Chiew suction, if a suitable value of B is found. In the case of injection
(1998b) as follows: only, Cheng and Chiew (1998b) proposed that B is related to

 vs /u∗ for the completely rough regime in the form:
u 1 y vs 1 y 2
= ln + ln (3)
u∗ κ y0 4u∗ κ y0 8.5
B= (5)
Cheng and Chiew (1998b) used a two-dimensional acoustic 1 + vs /u∗
Doppler velocimeter (ADV) for measuring the flow velocity The value of the roughness function with injection is appar-
and found that Eq. (3) compares well with their experimental ently smaller than that with zero-seepage flow, and reverts to 8.5,
data. Their measurement for vs /u∗ = 0.137 is reproduced in a commonly accepted value in the literature when the seepage
Fig. 2, showing that the commonly used log-law Eq. (1) deviates velocity vanishes.
considerably from the measured velocity profile. Numerical studies of seepage effect in open-channel flow are
Krogstad and Kourakine (2000) experimentally examined the comparatively scarce. Bose and Dey (2007) studied the effect
effect of localized injection through a porous strip of limited of bed injection on free surface flow field by numerically solv-
ing the Reynolds-averaged Navier–Stokes equations. In this way,
y the expressions for the streamwise and normal components of
velocity have been obtained with some constants estimated using
experimental data for flow in the outer and inner layers. The
u
velocity distributions computed are in good agreement with the
Velocity profile over
impermeable bed
experimental data for both cases with no-seepage and injection.
0
x
2.2 Flow with suction
vs
In the case of suction, the velocity distribution over a flat
Figure 1 Schematic diagram of velocity profile over permeable bed bed is schematically shown in Fig. 3 with the support of flow
with injection (after Nezu, 1977) visualization by Willetts and Drossos (1975). They proposed a
 Journal of Hydraulic Research Vol. 46, No. 4 (2008) Seepage effects on turbulent open-channel flow and sediment entrainment
y With y0 , u∗ and us determined experimentally, the authors
A B
yx
y1
y0x yxx
dx x age zone. This result is consistent with that obtained by Maclean
0 y0
vs Suction Zone us
Figure 3 Schematic diagram of velocity profile over permeable bed
with suction
modified seventh power law to describe the streamwise velocity
profile as:


Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014
u ky + y0x 1/7
=k (6)
umax y1
where k = (y1 − y0x )/yx ; y0x is the depth below which water is
removed from the stream between the two sections; y and yx , the
stream depths at A and B; and umax , the maximum velocity which
occurs at distance y1 from the bed. The experimental data mea-
sured using a hot-wire anemometer in a wind tunnel by Maclean
(1983) fit well with this profile. One main drawback of the method
is that many assumptions were made for computing y0x .
Eight years after his original work, Maclean (1991a) presented
another study on open-channel flow velocity over a zone of rapid
infiltration. In the latter paper, he argued that Eq. (4) is not valid
for the description of velocity distributions when suction veloc-
ity is high (vs /u∗ ∼ 2) because the x-derivatives (∂p/∂x, ∂u/∂x),
which are ignored in deriving Eq. (4), are no longer small. More-
over, he also defined an inner boundary layer close to the bed
to describe the velocity profile subjected to seepage. Based on
experimental data, he concluded that the flow structure in the
inner boundary layer is significantly different from that described
by Eq. (4); and the velocity close to the bed is increased, which
becomes apparent from the upstream edge of the suction region.
Above the inner layer, however, the velocity profile is still loga-
rithmic, if a suitable value is chosen for the roughness function,
B (see Eqs. (1) and (2)). Furthermore, the velocity gradient in
the logarithmic layer is found to reduce with an increase in the
suction velocity, arising from an acceleration of the flow close to
the bed and a deceleration of the flow near the water surface due
to continuity.
More recently, an approach similar to that proposed by Cheng
and Chiew (1998b) was used by Chen and Chiew (2004a) to
derive the velocity profile for flows subjected to suction. They
assumed that the boundary occurs at y = −y0 instead of y = 0,
as is the case for an impermeable bed (see Fig. 3). Taking the bed
surface as the reference level at which the flow has a slip velocity
of us , Chen and Chiew (2004a) proposed the following formula
analytically:


u − us 1 y + y0 vs 1 y + y0 2
= ln + ln (7)
u∗ κ y0 4u∗ κ y0
479
reported that Eq. (7) may be used to describe the measured pro-
files. Their measurements also confirmed the existence of the
slip velocity and a significant increase in the near bed velocity.
An example of the velocity profile for |vs /U0 | = 1.53% at the
middle of the seepage zone is reproduced in Fig. 4, where U0
is the depth-averaged velocity at the leading section of the seep-
(1991a), who showed an increased flow velocity close to the bed
and a reduced flow velocity near the free surface.
In addition, it should be mentioned here that in principle,
Eq. (3) or (4) should also apply for the case of suction, as the
direction of seepage velocity (suction or injection) is not speci-
fied in the relevant derivation. However, a general evaluation of
the parameters involved, y0 or B, needs further efforts for both
suction and injection situations. On the other hand, the modi-
fied logarithmic law may work only for the seepage velocity that
is relatively small when compared to the main-channel flow. A
high seepage velocity actually invalidates the theoretical analyses
related to Eqs (3) and (7).
Particle image velocimetry (PIV) was used by Yoda and
Westerweel (2001) to measure the velocity components in the
boundary layer with and without localized suction. It is worth
noting that the suction was produced through two 3 mm-diameter
holes spaced five-hole diameters apart, which is different from
the experimental setup employed, for example, by Cheng and
Chiew (1998b). Their results showed that suction decreased the
near-bed flow by an average of 6% in the region between the
suction holes, which are contrary to the findings of the other
researchers mentioned earlier. A plausible explanation is that the
use of the small suction hole could impose additional complica-
tions through the surface modification and pressure disturbances.
The experimental data measured using a single hot wire in a wind
tunnel by Antonia and Zhu (1995) support this explanation. In
their tests, suction was applied through a short porous bed strip
instead of an open hole or slot. The mean velocity distributions in
the presence of a sufficiently large rate of suction showed that an
increase in the flow velocities occurs immediately downstream
of the strip, relative to that in the case of no-seepage.
25
20
15
u(cm/s)
Measurements
10
Modified log-law with suction Eq (7)
5 Log-law of the wall Eq (1)
0
0 2 4 6 8
y(cm)
Figure 4 Comparison of experimental data with modified log-law for
suction (Chen and Chiew, 2004a)
 480 Y. Lu et al.
The effect of bed suction on the velocity distribution was stud-
ied numerically by Prinos (1995) using the Reynolds-averaged
Navier–Stokes equations and adopting low-R turbulence models
of the k-type for solving the system of equations. The compu-
tational results indicated an increase in near-bed velocities and
a decrease in flow velocities near to the free surface, which is
in agreement with the findings abovementioned (Willetts and
Drossos, 1975; Maclean, 1991a; Chen and Chiew, 2004a).
3 Turbulence
Although turbulence intensity in impervious open-channel flow
has been studied extensively, how it responds to seepage is not
clearly understood. Based on a study of the fluid dynamic forces
acting on a typical bed particle, Watters and Rao (1971) con-
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014
cluded that suction decreases turbulent fluctuations in the flow
field resulting in less momentum exchange between fluid par-
ticles, while the opposite is true for the case of injection. An
extreme case reported in their study showed that with strong
suction, the particle boundary layer changes from a turbulent
to laminar condition.
All three components of turbulence intensities in open-channel
flow for both injection and suction were measured by Nezu
(1977). His experimental data showed that seepage has a signif-
icant effect on the turbulence level, that is turbulence intensities
increase with injection and decrease with suction.
Other studies on the response of a turbulent boundary layer
to seepage, either through porous strips or open slots, generally
supported the observation that injection enhances (Krogstad and
Kourakine, 2000; Kim and Sung, 2003), while suction diminishes
turbulence intensities (Antonia and Zhu, 1995).
From experimental investigations of injection effects on open-
channel flow, Cheng (1997) observed an increase in the RMS
values of velocity fluctuations and the Reynolds shear stresses
near the bed, although little change was detected near the water
surface. This behavior, which was also observed by Nezu (1977),
is in agreement with the general expectation. Since seepage inter-
acts with the main-channel flow immediately near the bed, the
turbulence intensities are expected to change more readily near
the bed than those near the water surface. An example of the
Reynolds
   shear stress distribution along the seepage zone for
vs U0  = 1.07% at three specified depths from Cheng (1997)
is reproduced in Fig. 5.
A series of experiments were performed by Ramakrishna Rao
and Nagaraj (1999) to investigate the behavior of turbulence char-
acteristics. Velocity fluctuations were measured using a hot film
anemometer on a plane and non-transporting bed for both suction
and injection. They concluded qualitatively that turbulence inten-
sities were higher for suction than for injection or no-seepage. It
is interesting to note that the results of Ramakrishna Rao and
Nagaraj are contrary to those by Cheng (1997) and Nezu (1977).
Although the effect of seepage on the variation of turbu-
lence intensities has been investigated extensively, the reason
of how seepage modifies the turbulence structure still remains
Journal of Hydraulic Research Vol. 46, No. 4 (2008)
14
12
10
-u'v' (cm 2/s2)
8
6
4
0
-40 -20 0 20 40 60 80 100 120 140 160 180
x (cm)
Figure 5 Variations of Reynolds shear stress at specified depths
(|vs /U0 | = 1.07%) (Cheng, 1997)
uncertain. The main physical mechanism for the increased tur-
bulence intensities by injection may be twofold: (1) injection
causes a significant increase in the turbulent kinetic energy pro-
duction rate (Sumitani and Kasagi, 1995) and (2) injection also
diffuses upwards the streamwise vortices (Park and Choi, 1999),
thereby leading to a higher diffusion of turbulent energy into the
flow. One may envision that the corollary will apply to suction
effects on turbulence intensities, that is, a reduction of the turbu-
lent kinetic energy production rate and damping of the turbulent
energy diffusion, leading to reduced turbulence intensities.
Seepage may also result in significant near-bed turbulence
anisotropy, which plays an important role in turbulence produc-
tion. Such a modification has been investigated largely by direct
numerical simulation (e.g. Antonia et al., 1994; Park and Choi,
1999; Chung and Sung, 2001; Chung et al., 2002). The results
show that the near-bed anisotropy was enhanced in the presence of
suction, but may not be affected significantly by injection. Some
experimental evidences in this aspect can be found in Antonia
et al. (1988), Krogstad and Kourakin (2000) and Oyewola et al.
(2004).
4 Bed shear stress variations
A detailed investigation of flow and sediment movement in allu-
vial stream requires an accurate evaluation of the bed shear stress
or shear velocity. The latter, which is the most fundamental veloc-
ity scale for near-bed flows, can be determined using different
methods. One typical approach is to measure the water surface
slope under uniform flow conditions, which can then be used to
compute the bed shear stress or shear velocity. The shear veloc-
ity may also be determined from measured velocity distributions
in conjunction with the logarithmic law, or inferred from the
measured Reynolds shear stress distribution. Another approach
is to measure the bed shear stress directly with a Preston tube.
However, all these methods require an impervious flow boundary
and thus fail to work in the presence of seepage, which invali-
dates both flow uniformity and the no-slip boundary condition.
Consequently, other approaches have been developed to deter-
mine bed shear stresses on a permeable boundary (Oldenziel
 Journal of Hydraulic Research Vol. 46, No. 4 (2008) Seepage effects on turbulent open-channel flow and sediment entrainment 481

and Brink, 1974; Maclean and Willetts, 1986; Maclean, 1991b;
Ramakrishna Rao et al., 1994; Cheng and Chiew, 1998b; Chen
and Chiew, 2004a). Due to the complexity of the problem consid-
ered, obvious uncertainties exist among these studies, as detailed
below.
Oldenziel and Brink (1974) calculated the change in bed shear
stress due to suction and injection by considering the excess bed
shear stress arising from three factors: (1) the variation of energy
slope, (2) the difference in momentum flux between the two end
sections of a seepage zone and (3) the additional momentum
losses due to the mixing of channel flow and seepage. In comput-
ing the additional momentum losses, they introduced a reference
height, ys , at which this interaction takes place:

1/2
2 µLvs
ys =
3 τb0
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014
bed. By considering the infinitesimal element in section B shown
in Fig. 3, the excess bed shear stress is given by:
τe dx = ρU0x
2
yxx (11)
where yxx is the depth below which water is sucked into the bed
across the infinitesimal element and U0x , the average velocity
within the depth of yxx . Then, by applying the mass conservation,
the following expression was obtained:
τbs = τb0 + ρU0x vs (12)
Maclean (1991b) reported that Eq. (12) overestimated the bed
shear stress. By observing the initiation of indicator grains whose
critical shear stress for particle entrainment had been predeter-
mined in uniform flow, Maclean and Willetts (1986) measured
the bed shear stress with and without suction. The so-measured
(8)
shear stresses were found to increase with the suction velocity
ratio, as shown in Fig. 6. However, it must be noted that the

where L is the length of the seepage zone; vs , the seepage veloc-
ity; µ, the dynamic viscosity of fluid and τb0 , the bed shear stress
without seepage. Comparing the situations with and without
seepage, the following equation was finally obtained:

1/2
2 Lτb0 vs
τbs − τb0 = ρgh (Is − I0 ) − 2ρUvs − ηρvs
3 µ
where Is , I0 is the energy slope with and without seepage, respec-
tively; τbs , the bed shear stress with seepage; U, the mean flow
velocity without seepage; g, the gravitational acceleration; h, the
water depth and η, a correction factor. The correction for addi-
tional loss due to the mixing of discharges has to be determined
experimentally. Moreover, this method is practical only when the
energy slope is measured accurately.
The idea of Oldenziel and Brink was also adopted by Wil-
letts and Drossos (1975) for flow subjected to suction. In their
study, the streamwise velocity profile in the suction zone was first
described using the modified seventh power law Eq. (6) discussed
earlier (also see Fig. 3). Then, the value of y0x was determined by
assuming that the water removed between two sections per unit
time was equal to the flow rate within the depth y0x . Finally, the
excess shear stress due to suction was calculated as:
 x
1 1 1 2.4
ρgy1 + x(yx − y1 )ρgJb − ρgyx −
2 2
τe dx
2 2 2 0
 yx  y1
mechanism of the incipient motion of sediment particles subject
to bed seepage is different from that without seepage because dif-
ferent hydrodynamic forces are involved. Therefore, the results
of Maclean and Willetts (1986) could be seriously flawed.
To obtain bed shear stresses from the flow velocity, Maclean
(1991b) also proposed an alternative approach by relating the bed
(9)
shear stress to the square of the water velocity at a given reference
height above the bed:
τbs u2
= 2rs (13)
τb0 ur0
where ur0 , urs is the velocities measured at the reference height
without and with suction, respectively. This approach was found
to give a reasonable fit to his experimental data. However, the
choice of the reference height was arbitrary, which had signifi-
cant effects on the results (see Fig. 6). The method was verified
numerically by Prinos (1995) for estimating the bed shear stress
in open-channel flow with suction with an appropriate reference
height.
By conducting a series of laboratory experiments, Ramakr-
ishna Rao et al. (1994) observed that the variation of bed shear
stress resulting from seepage depends on the critical shear stress
2.2

=ρ u2x dy − ρ u21 dy (10)

0 0 2

1.8 Results of experiments

where Jb is the bed slope; ux , u1 , the velocity at y from bed in (Maclean and Willetts, 1986)
τbs /τb0

cross-section at B and A, respectively; and τe , the excess shear Theoretical results, Eq (13)
1.6
stress due to suction. A numerical method is necessary to solve Reference Height =1mm

Eq. (10). 1.4 Theoretical results, Eq (13)

Reference Height = 0.7mm
The results computed using Eq. (10) showed a sudden increase
1.2 Theoretical results, Eq (13)
in the shear stress at the leading edge of the suction zone, whereas Reference Height = 2mm
the experimental observation indicated a relatively gentle incre- 1
ment (Maclean, 1983). A further modification of this method was 0 0.04 0.08 0.12 0.16
proposed by Maclean and Willetts (1986) and Maclean (1991b) vs U 0
by assuming that the excess bed shear stress due to suction was Figure 6 Comparison of bed shear stress computed using the method
equal to the momentum given up by the fluid as it entered the of reference height (Maclean, 1991a, b)
 482 Y. Lu et al. Journal of Hydraulic Research Vol. 46, No. 4 (2008)

for particle entrainment without seepage, the sediment concentra- 1.8

v s /U0 = 0
v s /U0 = 0.47%
tion and seepage velocity. Two empirical relationships were thus v s /U0 = 1%
presented to estimate the variation for the flow with and without 1.6 v s /U0 = -0.86%
v s /U0 = -1.53%
sediment transport, respectively. In the case of a non-transporting
1.4
bed, the ratio of bed shear stress with and without seepage was

u*/u*0
suction
expressed as: 1.2

−CN
τbs τb0
= −N (14) 1
τb0 τc0
where N = (2ρUs vs /τb0 )(R/ h) is the seepage intensity parame- 0.8
injection
ter (vs takes negative for suction); τc0 , the critical shear stress for
0.6
particle entrainment without seepage; Us , the mean velocity with -0.2 0 0.2 0.4 0.6 0.8 1
seepage; h, the water depth; R, the hydraulic radius of the bed x/L
after smooth wall correction and C = 2.7. However, the C-value Figure 7 Bed shear stress distribution over injection and suction zone
was later revised by Ramakrishna Rao and Nagaraj (1999) to 2.2 (Cheng and Chiew, 1998b; Chen and Chiew, 2004a)
based on a wider range of experimental data. In the case of a
transporting sand bed, the following relationship, which is valid
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014

only for suction, was proposed: Figure 7 also shows that the shear stress reduces markedly at


τbs 600 1.75N the beginning of the injection zone and the reduction becomes
= +N (15) more apparent for higher seepage intensity. However, a gradual
τb0 c
increase of bed shear stress has been found towards the down-
where c is the sediment concentration in ppm by weight. It is of
stream end of the injection zone; the reason for this gradual
interest to note that one can deduce from Eqs (14) and (15) that
increase can be attributed to the influx of mass to the main flow.
both injection and suction can cause an increase or decrease in
Based on this result, one may deduce that injection effects on
the bed shear stress, an anomaly that will be discussed later.
bed shear stressed are related to two important features. First, it
By integrating the continuity and momentum equations,
alters the flow configuration near the bed resulting in a sudden
Cheng and Chiew (1998a) derived the following equation to
reduction of the bed shear stress directly. Second, it introduces
evaluate the shear velocity:

 additional mass to flow, thus causing an increase in the bed shear
dh βU 2 stress indirectly.
τbs = −ρgh 1− − 2βρUvs (16)
dx gh A recent theoretical study was conducted by Dey and Cheng
in which U is the depth-averaged streamwise velocity and h, the (2005), who employed the modified logarithmic law for injection
h
flow depth and β = ( 0 u2 dy)/(hU 2 ), the momentum correction Eq. (3) together with the Reynolds and continuity equations of
factor. The bed shear stresses derived based on the momentum two-dimensional open-channel flow. The derived equation for
integral equation (16) compare well with the results determined calculating the bed shear stress is quite complicated, which is
from the measured Reynolds shear stress distribution (Cheng, not reproduced herein, and the computed results are generally in
1997). Clearly, this approach depends on accurate measurements agreement with the experimental data of Cheng (1997).
of the water surface slope in the case of seepage. Another method The effect of mass transfer (introduced by injection or suction)
was proposed by Cheng and Chiew (1998b) for evaluating the bed may be substantiated by the experimental results of Ramakrishna
shear stress by directly fitting the modified logarithmic law for Rao et al. (1994), who conducted their tests in a flume with a
injection, that is Eq. (3) or (4), to the velocity measurements. seepage zone with length = 12.75 m. This length is much longer
To avoid the effect of uncertainties involved in near-bed velocity than that used by Cheng and Chiew (1998a, b), which is only
measurements, the depth-averaged velocity is used instead by 2 m. The large seepage zone definitely induces a relatively large
integrating Eq. (3) from y0 to h, which leads to (noting y0 / h  1): seepage influx, which when being added to the main flow will

 increase the bed shear stress to a significant extent. This may be
U 1 h
= ln −1 one of the reasons why their results are contrary to the general
u∗ κ y0

 expectation that injection tends to reduce bed shear stress. An
vs 2 h h opposite case is that if point-source seepage is considered, the
+ 2 ln − 2 ln −1 (17)
4κ u∗ y0 y0 effect of additional discharge influx on the bed shear stress would
An iterative procedure is necessary to evaluate the shear veloc- vanish.
ity with Eq. (17) because y0 is dependent on B, and thus on the An injection section consisted of a 0.12-m long porous strip
shear velocity. This approach is not subject to the uncertainty spanning the entire width of the wind tunnel was used by Krogstad
when fitting the modified logarithmic law to measured velocity and Kourakine (2000) to investigate localized injection effects on
profile. Cheng and Chiew (1998b) used it to evaluate the bed the turbulence structure in a boundary layer. Their experimen-
shear stress distribution over an injection zone; some results are tal results showed that as the incoming flow enters the injection
reproduced in Fig. 7, in which the injection and suction rates are region, the bed shear stress is significantly reduced. However, the
represented by positive and negative numbers, respectively. bed shear stress recovered sharply immediately downstream of
 Journal of Hydraulic Research Vol. 46, No. 4 (2008)
the seepage strip, indicating that the reduction of the bed shear
stress is directly due to the injection through altering the flow
configuration near the bed rather than the introducing additional
mass. Their observation appears to support the experimental data
of Cheng and Chiew (1998b) shown in Fig. 7.
Suction effects on bed shear stresses in open-channel flow
were experimentally investigated by Chen and Chiew (2004a). As
was discussed in the earlier section on velocity distributions, the
modified logarithmic law (7) for suction includes three unknown
parameters y0 , us and u∗ . To determine the shear velocity, the
measured velocity profiles were compared to Eq. (7). A software
package was used to fit the experimental data to a second-order
polynomial function to determine the best values for the three
parameters, including the shear velocity. For comparison, their
results on the bed shear stress distribution over the suction zone
are superimposed in Fig. 7. For the suction case, the data also
show significant reduction in the shear stress. This phenomenon
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014
is similar to that observed with injection, that is a decrease in
the bed shear stresses may largely be due to the reduction of
mass caused by suction. An interesting phenomenon of their data
is another reversal of the bed shear stress at around x/L = 0.3.
This second reversal is not seen for data associated with injection,
and the reason for its occurrence is as yet unknown.
By fitting their experimental data of the dimensionless shear
velocity (u∗ /u∗0 ) with relative suctions, Chen and Chiew (2004a)
presented an empirical equation as:
 3  2  
u∗  vs   vs   vs 
= 0.073   − 0.44   + 0.9   + 1 (18)
u∗0 U0 U0 U0
where u∗0 is the local shear velocity without seepage. This
approach shows that the rate of increase of shear velocity is much
bigger when |vs /U0 | < 0.6% than that when |vs U0 | > 0.6% (see
Fig. 8). Unfortunately, all the suction velocities in their studies
are more than 0.6% of the depth-averaged stream velocity due
to experimental setup limitation and no verification can be done
for small values of seepage rate. Further accurate measurements
at small values of suction velocities are necessary to verify the
empirical equation.
The shear velocity data provided by Cheng and Chiew
(1998a, b) for injection and by Chen and Chiew (2004a) for
1.8
1.6
u*/u*0
1.4
Eq (18)
1.2 Results of
Experiment
1 FS
0 0.005 0.01 0.015 0.02
vs U 0
Figure 8 Relationship between shear velocity and suction velocity
ratios (Chen and Chiew, 2004a)
Seepage effects on turbulent open-channel flow and sediment entrainment 483
1.8
experiment data with suction (Chen and
1.6 Chiew 2004a)
1.4
experiment data with injection (Cheng
and Chiew 1998a)
1.2
u*/u*0
1
0.8
0.6
0.4
0.2
-0.03 -0.02 -0.01 0 0.01 0.02 0.03
vs U 0
Figure 9 Effect of injection and suction on shear velocity
suction are re-plotted in Fig. 9. In this figure, the value of the
shear velocity for the respective seepage rate refers to that mea-
sured at the center of the seepage zone, that is, 1 m from its
leading edge. At this location, the data show a clear increase and
decrease of shear velocity for suction and injection, respectively.
It, however, does not reveal how mass addition (due to injection)
and extraction (due to suction) would influence shear velocities.
In other words, the data in Fig. 9 should only apply to a relatively
short seepage zone.
5 Stability of bed particles
Despite that existing data are generally incomplete and incon-
sistent, published studies have shown that the open-channel
flow characteristic can be significantly modified by seepage.
One may infer from this result that sediment transport theo-
ries developed without consideration of seepage effects are not
applicable for cases with suction or injection. Current research
results in this respect are, however, still very rudimentary. They
are largely limited to the simple case of incipient sediment
motion.
Since the boundary flow condition varies with seepage, bed
particles are subject to additional hydrodynamic forces besides
drag and lift forces. Cheng and Chiew (1999) performed a force
analysis for the threshold condition of sediment movement by
including an additional force due to injection. With reference to
a spherical particle at the incipient condition on a slope as shown
in Fig. 10, an equilibrium consideration of forces acting on the
FL
FD
0.025
W
Figure 10 Forces acting on a sphere resting on the bed
 484 Y. Lu et al. Journal of Hydraulic Research Vol. 46, No. 4 (2008)

3
particle results in:
Eq (22)
FD + W sin φ
tan θ = (19) 2.5
W cos φ − FL − FS Results ofExperiments Cheng and
Chiew (1999)D50=1.02mm
2
where θ is the angle of repose of the sediment particles; Results ofExperiment Ramakrishna

u*c (cm/s)
Rao and Nagaraj (1999) D50=1mm
W = (ρs − ρ)g(πd 3 /6), the submerged particle weight; 1.5
FD = CD (πd 2 /8)ρu2b , the drag force; FL = CL (πd 2 /8)ρu2b , the
lift force; FS = (iρgπd 3 /6(1 − ε)), the seepage force; CD , CL , 1

the drag and lift coefficients, respectively; d, the diameter of

0.5
particle; ρs , the density of particle; i, the seepage hydraulic gra-
dient; ε, the porosity of the granular materials and ub , the effective
0
velocity acting on particles at the bed. Furthermore, the effective 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
velocity was related to the shear velocity in the following form v s (cm / s)
(Chiew and Parker, 1994): Figure 11 Comparison of experimental results on threshold condition
u∗c for sediment entrainment
ub = √ (20)
f∗
where f∗ is a measure of friction; and u∗c , the critical shear without seepage, which may be computed using the Shields dia-
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014

velocity with seepage. Comparing the situation with and without
seepage, one gets:


u∗c 2 i
=1− (21)
u∗c0 ic
where ic = (1−ε)(ρs −ρ)/ρ for a horizontal bed, which describes
the critical condition of fluidization, under which the seepage
force acting on a particle just balances its submerged weight force
at the inception of motion. If the hydraulic gradient is related to
seepage velocity, Eq. (21) can be rewritten as:


m
u∗c 2 vs
=1− (22)
u∗c0 vsc
where vsc is the critical seepage velocity in a quick state and
m = 1 − 2, which depends on the characteristics of sedi-
ments. Equation (21) or (22) shows the relationship between
the critical shear velocities with and without injection under
the threshold condition for sediment transport. Increasing the
hydraulic gradient or injection velocity results in a reduction of
the critical shear velocity and vice versa. Equation (22) compares
well with the experimental data of Cheng and Chiew (1999),
which are re-plotted in Fig. 11. Subsequently, Cheng (2003)
derived an alternative expression for computing the critical shear
stress reduced by upward seepage, where Ergun equation was
used to describe the relationship between the seepage velocity
and seepage-related hydraulic gradient. More recently, Dey and
Zanke (2004) presented an analytical model and their results were
verified using the data of Cheng and Chiew (1999).
By fitting the experimental data collected in a 3.6-m long chan-
nel with a 2.4-m seepage zone, Ramakrishna Rao and Nagaraj
(1999) suggested that the critical shear stress for particle entrain-
ment due to suction τcs is related to both the bed shear stress
without seepage τb0 and the critical shear stress without seepage
τc0 , as:


−2.917
τb0 τcs
ln = −0.2525 for τb0 τc0 < 1 (23)
τc0 τc0
where τcs is the critical shear stress for particle entrainment with
seepage and τc0 , the critical shear stress for particle entrainment
gram. In their experiments, the values of τcs /τc0 varied from about
0.6 to 1.4, showing that suction can significantly affect the crit-
ical condition for sediment entrainment. By relating the critical
shear stress for particle entrainment with suction to the seepage
intensity parameter N defined in Eq. (14), Eq. (23) is modified to:

−2.2N
τcs τb0
= −N for τb0 /τc0 < 1 (24)
τb0 τc0
In the above analysis, Ramakrishna Rao and Nagaraj (1999)
further assumed that τbs = τcs and C = 2.2. Equations (23) and
(24) may be used to calculate the critical shear stress for sediment
entrainment with suction for any given particles, if the initial bed
shear stress is known. In the case of injection, the following
equation was fitted to their experimental data:


1.68
τb0 τcs
ln = 0.2525 for τb0 /τc0 > 1 (25)
τc0 τc0
A significant variation in τcs /τc0 , from about 0.5 to more
than 2, was shown in their investigation, which indicates that
injection can significantly affect the critical condition for sedi-
ment entrainment. However, no method was proposed in their
studies to relate the critical shear stress with injection to the
seepage intensity parameter, N.
It is of interest to note that the value of τcs /τc0 can be more
or less than unity for both cases of suction and injection in the
study by Ramakrishna Rao and Nagaraj (1999), whereas those
by Cheng and Chiew (1999) (i.e., Eq. (22)), Cheng (2003) and
Dey and Zanke (2004) showed that τcs /τc0 is always less than
unity in the case of injection. Ramakrishna Rao and Nagaraj
argued that in the presence of injection, the resistance decreases
due to the reduction of the effective weight of bed particles, and
the hydrodynamic forces acting on the bed particles also reduce
due to the decrease in the near-bed velocity as well as turbu-
lence intensity; the opposite trend holds for suction. Therefore,
it appears that the threshold of sediment movement is depen-
dent on variations in both changes in the effective particle weight
and hydrodynamic forces. No satisfactory explanations are given
at this stage, although it does not appeal to common sense that
 Journal of Hydraulic Research Vol. 46, No. 4 (2008)
injection should cause an increase in the threshold condition for
sediment entrainment.
The experimental data of Ramakrishna Rao and Nagaraj
(1999), which were collected in tests conducted with a similar
grain size as those conducted by Cheng and Chiew (1999), are
superimposed in Fig. 11. The figure shows that the two sets of
experimental results are very different, even for their trend.
It must be stated that these results are affected to some extent
by the length of the seepage zone. Since the bed shear stress is
not uniform along the seepage zone (see Fig. 7) due to variation
of the main flow discharge resulting from mass transfer associ-
ated with either injection or suction, it is questionable to use any
average value, such as average transport rate, or that measured
in the center of the seepage zone to define the threshold of sedi-
ment movement. To further illustrate this, it may be worthwhile
to conduct a thought experiment, whose setup is schematically
shown in Fig. 12. The 2 m-long-seepage zone is divided into five
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014
sections evenly. Each section is covered by the same uniform
sediment particles that are color-coded.
In order to examine the incipient motion criterion, the sed-
iment particles that are entrained are collected in a box placed
downstream of the test section. For a given water depth and main
flow discharge, either injection or suction is introduced, together
with the adjustment of the tail gate to achieve the incipient con-
dition. For the case of injection, for example, the variation of bed
shear stress (Fig. 7) presented by Cheng and Chiew (1998b) along
the seepage zone is superimposed in Fig. 12 to aid discussion.
The imaginary threshold shear stress for sediment entrainment is
also shown in Fig. 12 as a dash line. With these considerations,
the amount of sediment particles collected at the downstream end
of the seepage zone may reasonably be shown in Fig. 13.
white red blue
Bed Shear Stress
Figure 12 Thought experimental setup for study of incipient motion of
sediment with seepage
Sands Collected
White
Blue
Red
Figure 13 Schematic diagram of quantity of sediment particles
collected
Seepage effects on turbulent open-channel flow and sediment entrainment 485
Figure 13 shows that the quantity of the different color parti-
cles collected are different. This is because the bed particles in
the seepage zone are identical except for their color, thus their
respective critical shear stress must remain the same. However,
the bed shear stress experienced by each of these color-coded
sediment particles is different and dependent on their location
in the seepage zone as the distribution of bed shear stress along
the zone changes. Therefore, it is not reasonable to define the
incipient condition by considering the average transport rate at a
particular location in the seepage zone. Additionally, it is ques-
tionable that the threshold of sediment initiation is dictated by
the bed shear stress at the center of the seepage zone. As a min-
imum, this thought experiment highlights the limitation of pub-
lished results. Additional experiments and analyses clearly are
necessary.
The experimental results of Ramakrishna Rao and Nagaraj
(1999) showed that injection increases the stability of bed parti-
cles while suction does the opposite. This is in agreement with the
findings of some earlier researchers (Harrison, 1968; Willets and
Drossos, 1975), but contrary to the findings of more recent studies
(Oldenziel and Brink, 1974; Richardson et al., 1985; Cheng and
Chiew, 1999; Cheng, 2003; Dey and Zanke, 2004). A summary
of these contradicting results is shown in Table 1; an attempt to
explain the contradiction is presented below.
In general, the stability of bed particles is controlled by the
effective weight and bed shear stress, of which both are modified
by seepage, as illustrated in Fig. 14. In the figure, a positive (+)
and negative (−) sign denotes an increase and decrease in parti-
cle stability, respectively. The presence of suction can cause an
increase in the effective weight of the bed particles because it
acts downwards, while injection has the opposite effect. As dis-
cussed previously, suction or injection, respectively, increases or
decreases the bed shear stress. Figure 14 shows a graphical repre-
sentation of the influence of the effective weight and shear stresses
on bed particles stability. In quadrant 2, the stability of bed par-
ticles is always reduced due to the increase of bed shear stress
yellow black and the decrease of effective weight of bed particles. In quad-
Box
rant 4, on the other hand, the stability of bed particles is always
_
Bed Shear Stress
Suction
Increase
2 1
_ Effective Weight Decrease 45o
Effective Weight Increase
+
Injection Suction
Black
Bed Shear Stress
Yellow
3 4
Injection
Decrease
Sand Sorts
+
Figure 14 Quadrant analysis of bed particle stability due to seepage
 486 Y. Lu et al.
enhanced due to the reduction of bed shear stress and the increase
of effective weight. While the above inference is rational, it has no
real physical meaning because it is impossible to introduce both
injection and suction at the same time. The other two quadrants
are, however, more realistic. Quadrant 1 depicts the condition
where suction increases the effective weight of bed particles,
thus enhancing the stability of bed particles; on the other hand,
suction also has the tendency to increase the bed shear stress, and
as a result reduces the stability of bed particles. With this com-
bined effect, it is not immediately clear whether the stability of
bed particles is finally enhanced or reduced. Similarly, in quad-
rants 3, injection decreases the effective weight resulting in the
reduction of the bed particle stability, and at the same time it also
decreases the bed shear stress resulting in an increase of stability.
From this consideration, it is clear that the stability of bed parti-
cles is dependent on the relative magnitude of these two opposing
effects. For instance, in the experiments of Ramakrishna Rao and
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014
Nagaraj (1999), as a result of the decrease of effective weight due
to injection, the stability of the bed particles is reduced, but it is
offset by the reduction of bed shear stresses. The final stability of
the bed particles is dependent on which effect is more dominant.
If the reduction in weight is more than the decrease in the bed
shear stress, the stability of the sediment particle will reduce; a
condition designated by the shaded portion in quadrant 3. Sim-
ilarly, suction increases the effective weight, thereby leading to
an increase in stability of the bed particles, but it is offset by the
increase of bed shear stresses. If the increase in weight is less than
the increase in the bed shear stress, the stability of the sediment
particle will reduce, as indicated by the shaded portion in quad-
rant 1. The above reasoning may be an explanation to account for
the contradictory findings in the published literature. This hypoth-
esis, including the idea illustrated in Fig. 14, is only conceptual
presently; more experiments are needed to prove the explana-
tion and investigate the imaginative dashed line shown in the
figure.
6 Conclusions
Published works that deal with seepage effects on open-channel
flow and sediment entrainment are limited. From a theoretical
viewpoint, the presence of seepage causes a deviation of the
logarithmic law of the wall. Many attempts have been made
to characterize the velocity profile of the turbulent flow on
a porous bed (e.g., Willetts and Drossos, 1975; Schlichting,
1979; Maclean, 1991a; Antonia and Zhu, 1995; Prinos, 1995;
Cheng and Chiew, 1998a, b; Krogstad and Kourakine, 2000;
Chen and Chiew, 2004a; Bose and Dey, 2007). Notwithstand-
ing the hypotheses and conjectures, what exists depicts well the
alteration of the flow velocity induced by seepage.
Most previous experimental studies of injection or suction
have shown that when injection is applied, turbulence intensi-
ties are enhanced (Watters and Rao, 1971; Nezu, 1977; Cheng,
1997; Krogstad and Kourakine, 2000; Kim and Sung, 2003)
due to an increase in the turbulent kinetic energy production
rate and acceleration of the turbulent energy diffusion. On the
Journal of Hydraulic Research Vol. 46, No. 4 (2008)
other hand, suction diminishes turbulence intensities (Watters
and Rao, 1971; Nezu, 1977; Antonia and Zhu, 1995; Yoda and
Westerweel, 2001; Chen and Chiew, 2004a) due to a reduction
of the production and inhibiting of the diffusion. Ramakr-
ishna Rao and Nagaraj (1999), however, showed the reverse
trend.
Detailed investigations of flow and sediment movement in
alluvial stream require an accurate evaluation of the bed shear
stress or shear velocity. For point-source seepage, it alters the
flow configuration near the bed resulting in a sudden modification
of the bed shear stress directly; for seepage with long streamwise
extent, the effect of mass (introduced by injection or suction) also
needs to be considered (Ramakrishna Rao et al., 1994).
The issues of whether seepage reduces or enhances sand-bed
stability are still matters of considerable debate. Table 1 shows
many of these discrepancies. For example, some studies (Olden-
ziel and Brink, 1974; Richardson et al., 1985) suggested that
injection increases bed erosion, whereas suction tends to inhibit
sediment motion. On the other hand, other researchers suggested
a completely opposite result, stating that suction increases sedi-
ment transport rate (Willetts and Drossos, 1975), while injection
inhibits the motion of bed particles (Watters and Rao, 1971).
It is evident from changes of the flow velocity distribution, bed
shear stresses and turbulence intensities due to the introduction of
seepage that interaction between flow and permeable boundary
is significantly influenced by seepage. Considerable work is still
needed to explore and better understand the phenomenon. The
principal inference of this review is that the complex interac-
tions caused by seepage raise many important issues that require
further research.
Notation
B = Roughness function
C = 2.7 or 2.2
CD = Drag coefficient
CL = Lift coefficient
c = Sediment concentration in ppm by weight
D50 = Median diameter of particles
d = Diameter of sediment particles
FD = Drag force
FL = Lift force
FS = Seepage force acting on a particle
f∗ = A measure of friction
g = Gravitational acceleration
h = Water depth
Is , I0 = Energy slope with and without seepage
i = Hydraulic gradient of seepage
ic = Hydraulic gradient of seepage under quick condition
Jb = Bed slope
k = (y1 − y0x )/yx
ks = Granular roughness of bed surface
L = Length of seepage zone
m = Exponent =1–2
N = Seepage intensity parameter
 Journal of Hydraulic Research Vol. 46, No. 4 (2008) Seepage effects on turbulent open-channel flow and sediment entrainment
p = Pressure
R = Hydraulic radius of the bed after smooth wall
correction
U = Depth-averaged velocity
U0 = Depth-averaged velocity at the leading edge of the
seepage zone
Us = Mean velocity with seepage
U0x = Average velocity within the depth of yxx
u = Time-averaged streamwise velocity of flow
u1 , ux = Velocity at y from bed in cross-section at A and B
ub = Effective velocity acting on particles at the bed
umax = Maximal streamwise velocity
ur0 , urs = Measured velocities at the reference height without
and with suction, respectively
us = Slip velocity due to seepage
u∗ = Local shear velocity
u∗0 = Local shear velocity without seepage
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014
u∗c = Critical shear velocity with seepage
u∗c0 = Critical shear velocity without seepage
v = Time-averaged vertical velocity of flow
vs = Seepage velocity
vsc = Critical seepage velocity under quick condition
W = Submerged weight force
x = Streamwise coordinate or distance from the leading
edge of seepage zone
y = Normal coordinate
y0 = Zero-velocity displacement measured from the
theoretical bed surface
y0x = The depth below which water is removed from the
stream between cross-sections A and B due to
seepage
y1 , yx = Water depth at cross-sections A and B
yxx = Depth below which water is sucked into the bed
across the infinitesimal element
ys = A reference height at which the interaction due to
seepage takes place
β = Momentum correction factor
θ = Angle of repose of submerged sediment particles
ρ = Mass density of fluid
ρs = Density of particles
µ = Dynamic viscosity of fluid
−u v = Reynolds shear stress
τ = Shear stress
τb = Total boundary shear stress
τb0 , τbs , = Total boundary shear stress without and with
seepage, respectively
τc0 = Critical shear stress for sediment entrainment
without seepage
τcs = Critical shear stress for sediment entrainment with
seepage
τe = Excess shear stress due to suction
κ = Von Karman’s constant
ι = Prandtl’s mixing length
ε = Porosity of granular materials
η = A correction factor
487
References
Antonia, R.A., Fulachier, L., Krishnamoorthy, L.V., Benabid,
T., Anselmet, F. (1988). Influence of wall suction on the orga-
nized motion in a turbulent boundary layer. J. Fluid Mech. 190,
217–240.
Antonia, R.A., Spalart, P.R., Mariani, P. (1994). Effect of suction
on the near-wall anisotropy of a turbulent boundary layer. Phys.
Fluids 6(1), 430–432.
Antonia, R.A., Zhu, Y. (1995). Effect of concentrated wall suc-
tion on a turbulent boundary layer. Phys. Fluids 7(10), 2465–
2474.
Bose, S.K., Dey, S. (2007). Theory of free surface flow over
rough seeping beds. Proc. Royal Soc. A 463(2), 369–383.
Brunke, M., Gonser, T. (1997). The ecological significance of
exchange processes between rivers and groundwater. Fresh-
water Biol. 37(1), 1–33.
Chen, X.W., Chiew, Y.M. (2004a). Velocity distribution of tur-
bulent open channel flow with bed suction. J. Hydraul. Eng.
ASCE 130(2), 140–148.
Chen, X.W., Chiew, Y.M. (2004b). Turbulence characteristics
of open channel flow with bed suction. J. Eng. Mech. ASCE
(under review).
Cheng, N.S. (1997). Seepage effect on open-channel flow and
incipient sediment motion. A thesis submitted to Nanyang
Technological University in fulfillment of the requirements for
the degree of Doctor of Philosophy.
Cheng, N.S. (2003). Application of ergun equation to compu-
tation of critical shear velocity subject to seepage. J. Irrigat.
Drainage Eng. 129(4), 278–283.
Cheng, N.S., Chiew, Y.M. (1998a). Turbulent open-channel flow
with upward seepage. J. Hydraul. Res. IAHR 36(3), 415–431.
Cheng, N.S., Chiew, Y.M. (1998b). Modified logarithmic law for
velocity distribution subjected to upward seepage. J. Hydraul.
Eng. ASCE 124(12), 1235–1241.
Cheng, N.S., Chiew, Y.M. (1999). Incipient sediment motion
with upward seepage. J. Hydraul. Res. IAHR 37(5), 665–681.
Chiew, Y.M., Parker, G. (1994). Incipient sediment motion on
non-horizontal slopes. J. Hydraul. Res. IAHR 32(5), 649–660.
Chung, Y.M., Sung, H.J. (2001). Initial relaxation of spatially
evolving turbulent channel flow with blowing and suction.
AIAA J. 39(11), 2091–2099.
Chung, Y.M., Sung, H.J., Krogstad, P.-Å. (2002). Modulation of
near-wall turbulence structure with wall blowing and suction.
AIAA J. 40(8), 1529–1535.
Clarke, J.H., Menkes, H.R., Libby, P.A. (1955). A provi-
sional analysis of turbulent boundary layers with injection.
J. Aerospace Sci 22(4), 255–260.
Dey, S., Cheng, N.S. (2005). Reynolds stress in open channel
flow with upward seepage. J. Eng. Mech. ASCE 131(4), 451–
457.
Dey, S., Zanke, U.C.E. (2004). Sediment threshold with upward
seepage. J. Eng. Mech. ASCE 130(9), 1118–1123.
Harrison, S.S. (1968). The effects of groundwater seepage on
stream regime—a lab study. A thesis presented to University
 488 Y. Lu et al.
of North Dakota, in partial fulfillment of the requirements for
the degree of Doctor of Philosophy.
Jones, J.B., Mulholland, P.J. (eds.) (2000). Streams and Ground-
waters. Academic, San Diego.
Karambas, T.V. (2003). Modelling of infiltration-exfiltration
effects of cross shore sediment transport in the swash zone.
Coast. Eng. J. 45(1), 63–82.
Kim, K., Sung, H.J. (2003). Effects of periodic blowing from
spanwise slot on a turbulent boundary layer. AIAA J. 41(10),
1916–1924.
Krogstad, P.-Å., Kourakine, A. (2000). Some effects of localized
injection on the turbulence structure in a boundary layer. Phys.
Fluids 12(11), 2990–2999.
Maclean, A.G. (1983). Local erosion over a submerged intake
in an alluvial channel. Thesis presented to the University
of Aberdeen, at Aberdeen, Scotland, in fulfillment of the
requirements for the degree of Doctor of Philosophy.
Downloaded by [UOV University of Oviedo] at 11:23 31 October 2014
Maclean, A.G. (1991a). Open channel velocity profiles over a
zone of rapid infiltration. J. Hydraul. Res. IAHR 29(1), 17–27.
Maclean, A.G. (1991b). Bed shear stress and scour over bed-type
river intake. J. Hydraul. Eng. ASCE 117(4), 436–451.
Maclean, A.G., Willetts, B.B. (1986). Measurement of boundary
shear stress in non-uniform open channel flow. J. Hydraul. Res.
IAHR 24(1), 39–51.
Nezu, I. (1977). Turbulent structure in open channel flow. PhD
thesis, Kyoto University, Kyoto, Japan.
Oldenziel, D.M., Brink, W.E. (1974). Influence of suction and
blowing on entrainment of sand particles. J. Hydraul. Div.
ASCE 100(HY7), 935–949.
Oyewola, O., Djenidi, L., Antonia, R.A. (2004). Influence of
localised wall suction on the anisotropy of the reynolds stress
tensor in a turbulent boundary layer. Exp. Fluids 37, 187–193.
Park, J., Choi, H. (1999). Effects of uniform blowing or suction
from a spanwise slot on a turbulent boundary layer flow. Phys.
Fluids 11(10), 3095–3105.
Prinos, P. (1995). Bed-suction effects on structure of turbulent
open-channel flow. J. Hydraul. Eng. ASCE 121(5), 404–412.
Ramakrishna Rao, A., Nagaraj, S. (1999). Stability and mobility
of sand-bed channels affected by seepage. J. Irrigat. Drainage
Eng. 125(6), 370–379.
Journal of Hydraulic Research Vol. 46, No. 4 (2008)
Ramakrishna Rao, A., Subrahmanyam, V., Thayumanavan S.,
Namboodiripad, D. (1994). Seepage effects on sand-bed
channels. J. Irrigat. Drainage Eng. 120(1), 60–79.
Raudkivi, A.J. (1990). Loose Boundary Hydraulics, 3rd edn.
Pergamon Press, Oxford, UK.
Richardson, J.R., Abt, S.R., Richardson, E.V. (1985). Inflow
seepage influence on straight alluvial channels. J. Hydraul.
Eng. ASCE 111(8), 1133–1147.
Rotta, J.C. (1966). Control of turbulent boundary layers
by uniform injection or suction of fluid. ZAMM 46,
213–215.
Ruff, J.F., Gelhar, L.W. (1972). Turbulent shear flow in porous
boundary. J. Eng. Mech. Div. ASCE 98(4), 975–991.
Schlichting, H. (1979). Boundary-Layer Theory, 7th edn.
McGraw-Hill Book Company, New York.
Shimizu, Y., Tsujimoto, T., Nakagawa, H. (1990). Experiment
and macroscopic modeling of flow in highly permeable porous
medium under free-surface flow. J. Hydrosci. Hydraul. Eng.
8(1), 69–78.
Sumitani, Y., Kasagi, N. (1995). Direct numerical simulation of
turbulent transport with uniform wall injection and suction.
AIAA J. 33(7), 1220–1228.
Turner, I.L. (1995). Simulating the influence of groundwater
seepage on sediment transport by the sweep of the swash
zone across macro-tidal beaches. Marine Geol. 15(1–2),
153–174.
Watters, G.Z., Rao, M.V.P. (1971). Hydrodynamic effects of
seepage on bed particles. J. Hydraul. Div. ASCE 101(3),
421–439.
Willetts, B.B., Drossos, M.E. (1975). Local erosion caused by
rapid forced infiltration. J. Hydraul. Div. ASCE 101(12), 1477–
1488.
Yalin, M.S. (1977). Mechanics of Sediment Transport, 2nd edn.
Pergamon Press, Inc., London, England.
Yoda, M., Westerweel, J. (2001). Particle image velocimetry stud-
ies of a boundary layer perturbed by localized suction. Exp.
Fluids 30, 239–245.
Zhou, D., Mendoza, C. (1993). Flow through porous bed of
turbulent stream. J. Eng. Mech. 119(2), 365–383.