Sei sulla pagina 1di 20

Flow, Turbulence and Combustion 70: 6988, 2003.

2003 Kluwer Academic Publishers. Printed in the Netherlands.

69

An Open Channel Flow Experimental and


Theoretical Study of Resistance and Turbulent
Characterization over Flexible Vegetated Linings 
DAVID VELASCO1, ALLEN BATEMAN1, JOSE M. REDONDO2 and
VICENTE DEMEDINA1
1 Hydraulic and Hydrological Section, 2 Applied Physics Department, Polytechnic University of

Catalonia UPC, E-80340 Barcelona, Spain


Received 6 December 2001; accepted in revised form 4 May 2003
Abstract. Hydraulic engineers and scientists working on river restoration recognize the need for
a deeper understanding of natural streams as a complex and dynamic system, which involves not
only abiotic elements (flow, sediments) but also biotic or biological components. From this point of
view, the role played by riverine vegetation in river dynamics and flow conditions becomes essential.
Hydro-mechanic interaction between flow and flexible plants covering a river bed is studied in this
paper and some previous works are discussed. Experimental tests and measurements of turbulence
on the flow in an open channel were performed using plastic plants seeded in a gravel bed. Characterization of flow resistance (friction factors) due to vegetation flexible roughness for different plant
densities was attained; furthermore, measuring detailed turbulent velocity profiles within and above
submerged and flexed stems allowed us to distinguish different turbulent regimes. Some interesting
relationships were obtained between the velocity field and the deflected height of the plants, such as
a linear fit between the non-dimensional flexural parameter and the relative deflection of the plants.
Turbulent stresses were measured showing two different regions: above and inside the vegetation
domain. The spectral interaction between the plant oscillations, their wakes and the turbulence
at different heights, forces strongly anisotropic Reynolds tensors and in order to clarify turbulent
processes and their complex structure, theoretical concepts (Taylor, Kolmogorovs K41) and several
data analysis (autocorrelation functions, integral scales) were applied.
Key words: canopy flows, plant turbulence.

Nomenclature
a
B
d
D50
E
f
h
hp

=
=
=
=
=
=
=
=

interplant length
channel witdh
average plant diameter
percentile 50 in soil particle distribution
stiffness modulus
DarcyWeisbach friction factor
uniform water depth
penetration point

 Presented at the Fluxes and Structures in Fluids Conference 2001, Moscow, Russia.

70
h
I
k
k
K

L
M
Q
q
Re
Rh
T
U
U
V
W
u , v  , w
Uk

xz , xy , yz

D. VELASCO ET AL.

=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=

non-bending plant height


second order geometrical momentum
deflected height of plant
wavenumber
turbulent kinectic energy
von Krmn turbulent diffusion constant
integral scale
density of vegetation
discharge
unit discharge = Q/B
Reynolds number
hydraulic radius
turbulent intensity
longitudinal mean velocity
bed shear velocity
transversal mean velocity
vertical mean velocity
longitudinal, transversal and vertical velocity fluctuation, respectively
slip velocity
water density
kinematic viscosity of water
Reynolds stresses
bed shear stress

1. Introduction
Several studies have examined the influence on and the global hydraulic relationship between flexible roughness due to plants and the consequent flow conditions.
Hydraulic design of grassed irrigation channels lead to the first experimental tests
[18], where a reduction in friction factors above natural vegetative linings for
higher flow rates was reported. The most important contribution was made by
Kouwens studies [79]. Dimensional analysis led him to create a simple model to
evaluate resistance to flow depending on the geometric and mechanical properties
of submerged plants (density, elasticity) and flow conditions. Momentum transfer
mechanics are responsive to vortex organization and flow configuration, as well as
heat and dissolved substances exchange and diffusion between bottom and surface
regions. Specific models were required to accurately solve the basic fluid mechanics interactions between flexible plants in a similar way as with compilant surfaces.
Important works on numerical (ke turbulent schemes) and experimental turbulent
characterization over vegetation have been carried out [4, 13]. Good adjustments
of the results were obtained from rigid and isolated stem covers, but it is clear
that stem flexibility and the heterogeneity of leave density are critical properties
needed to correctly estimate drag forces [11]. Likewise, biological and environmental processes in natural rivers related to the presence of vegetation (nutrient
transport, oxygen rate) and the fact that plankton and larvae growth in the ocean
often take place within Poseidonia fields were of major interest in some stud-

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

71

Figure 1. Longitudinal flume scheme and transversal section. Photograph of a plastic plant.

ies concerning turbulent and contaminant diffusion over submerged and emergent
vegetation [14, 15].
Vegetation induces biological depuration processes, so it is a very effective
environmental measure for reducing nitrates and phosphates poured into rivers.
Moreover, most rivers in populated areas have supported a great antrophic stress
during the last decades, so they have become totally degradated. Revegetation
techniques must be introduced and public opinion demands effective restoration
proceedings, as well as new forms of flood control. As a consequence, a deeper
knowledge of hydromechanics over vegetation is obviously required to guarantee
appropriate and safe hydraulic designs.
In the present study, some experimental tests were conducted under controlled
conditions. In order to simplify the inherent complexity of vegetation properties,
scaled plastic strips were used to model plants. Different plant densities were tested
as well as different flow conditions such as the flow and the ratio of the plant height
to the water depth. The velocity field was measured directly in the experimental
flume with a Acoustic-Doppler sensor (ADV) which is described in Section 2. The
results from the velocity measurements are presented in Sections 3 and 4 and show
a reduction in friction factors due to plant bending for an increasing Reynolds number as expected. The characteristics of the turbulent flow and the turbulent velocity
profiles as a function of the plant/flow characteristics and an introduction to the
analysis of turbulent structures that may explain some of the turbulent velocity
data, including autocorrelations, integral length-scale distributions and spectra is
included in the discussion and, finally, conclusions are drawn.
2. Experimental Setup
The experiments were performed in the hydraulic laboratory of the Polytechnical
University of Catalonia (UPC) in Barcelona. A 20 m long concrete flume was
used. The cross-section was rectangular, 1 m wide and 0.9 m deep. The gravel
bed was extended in a constant slope of 0.64 (D50 = 2.05 cm). Gravel particles
define a friction factor of ManningStrickler n = 0.025, which reproduces the

Discharge (l/s)
160-20
250-20
160-20
250-20
160-40
250-20
160-50

Density M
(plants/m2 )

170
156
130
100
85
70
25

0.520.12
0.820.17
0.710.18
0.860.19
0.620.25
0.910.23
0.300.153

Velocity (m/s)
0.2790.133
0.3050.117
0.2250.124
0.2900.105
0.2170.128
0.2750.086
0.2530.14

Water depth
h (m)
0.059 0.15
0.0530.15
0.0630.15
0.0620.15
0.0580.15
0.0620.15
0.1130.15

Plant deflected
height k (m)
4.721.0
5.751.0
3.581.0
4.371.0
3.751.0
4.431.0
2.241.0

Submergence
h/k

Table I. Experimental test conditions.

1.2E51.4E4
1.7E51.4E4
1.3E51.5E4
1.7E51.4E4
1.1E51.9E4
1.7E51.4E4
4.3E41.8E4

Re

Gravel (D = 2 cm)
Gravel (D = 2 cm)
Gravel (D = 2 cm)
Gravel (D = 2 cm)
Gravel (D = 2 cm)
Gravel (D = 2 cm)
Sand (D = 0.1 cm)

Bed roughness

72
D. VELASCO ET AL.

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

73

natural roughness of many local rivers. In Table I the basic flow and geometrical
parameters related to the plant models are described for the seven different configurations used. The ranges used for each relevant parameter are indicated. Only one
set of experiments (with a plant density of 25 plants per square meter) used sand to
generate a much smaller bed roughness. Figure 2 describes the overall conditions
of the experimental flume and also shows an example of the plastic plant model
used.
One of the most important features in the investigation was the realistic physical
modelization of natural plant species through plastic, artificial strips. These were
designed to reproduce the mechanical behavior of real local river plants. A PVC
plastic bunch made of 0.15 m long (h = 0.15 m), thin strips approximates some
autochthonous macrophytes shapes (Phragmites Australis-Common Reed), where
aerial parts of the plant are concentrated with an average of 20 leaves per plant.
Also, it was confirmed from independent tests that the stiffness modulus of plastic
(E = 2.38 109 N m2 ) was adequate to the used hydraulic regimes and flow
actions. Kouwens dimensionless analysis shows that a so-called MEI parameter
can be considered as being mainly responsible for the vegetation behavior and thus
to exhibit resistance to flow conditions. The MEI factor (after M = density of
plants per unit area, E = stiffness modulus and I = inertia momentum) represents the global mechanical and geometrical properties of a group of plants, so we
can modify individual stiffness modulus and inertia or the global plant density to
achieve similar roughness effects.
A simulated vegetated zone was set in a 7 m central zone along the channel and
plastic plants were directly fixed on the gravel bed. Initially, three different densities of plants were studied (M = 156, 100 and 70 plants/m2 ). Another parameter
that defines the density of plants is a, an averaged interplant distance, so Ma 2 = 1.
According to previous densities, a = 0.08, 0.10 and 0.12 m, respectively. Later
three intermediate new densities were tested (M = 170, 130 and 85 plants/m2 with
a = 0.076, 0.087 and 0.108 m, respectively). Additionally, a sand bed channel
(2.5 m width) was used in some tests. Sand was very uniform (D50 = 1 mm) and a
very low density of plants was set (25 plants/m2 , a = 0.20 m). Spatial distribution
of obstacles has a great influence on friction factors (preferable streamlines and velocity gradients) and a staggered plantation pattern was used (natural distribution),
but the plant seeding was performed as uniform as possible. A pumping system into
the channel set up unitary discharge flows q ranging from 0.0205 to 0.250 m2 /s and
submergences of plant h/k from 1 to 5.75, where h is the normal average water
depth and k is the deflected plant height. Obviously the higher the flow Q = qB,
the smaller the deflected plant height.
A movable weir located downstream the channel was used to obtain uniform
regime conditions along the vegetated zone.
Measurement equipment included graduated rubbers and mechanic limnimeters of high precision to register water depths and the deflection of plants. Image analysis of video tapes of some experiments were also useful in detecting

74

D. VELASCO ET AL.

Figure 2. Variability of friction factor f versus Re.

variations in the flow and in the model plant canopy. The discharges were controlled in an upstream triangular weir. Two different types of velocity sensors
were used to measure temporal series of velocity in different positions. A 2D
electromagnetic velocimeter took data of average velocity along a cross-section.
In the detailed study of turbulent structures, a 2D and a 3D acoustic Doppler
velocimeter (SONTEK-ADV) were used after extensive calibration. Analysis of
the range of turbulent length scales (until viscous micro-scales) requires a high
resolution and measurement frequency. Temporal series of 5000 data were taken
per point, with a 25 Hz measurement frequency. Higher measurement frequencies
up to 75 Hz were also tested but they did not produce more accurate data, due to
spatial non-homogeneities in the seeding of the sampling volume, typically 210
cubic milimeters.
Some vertical profiles in the center streamline of the channel were registered,
in order to recognize vertical distribution of velocities, but, because of the geometry of the sensors, only the experiments with deepest flows were analyzed. The
relative position of the sensor inside the vegetation pattern is critical because of
the structures of the wakes of the plants, the local effects of stems in the surrounding velocity field, pressure gradients, interaction of the turbulent wakes, and other
sources of coherent structures that had to be treated in an statistical way. In order
to reduce local effects, one or two plants were removed to create a spatialy uniform
flow area, where the sensor was located. We consider that this does not affect the
statistical turbulent properties of the flow.

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

75

Figure 3. Comparison of friction factors between our data and Kouwens artificial roughness
data (extract from [7]).

3. Results
3.1. VARIABILITY ON FRICTION FACTORS
One of the main objectives of this work is to evaluate the global resistance to
flow under different conditions of flow and density of plants. The estimation of
these friction factors, depending on bed roughness, is a very critical point in open
hydraulic flumes. The flexible properties of plants cause a rapidly reduction of relative roughness (k/ h) for increasing Reynolds numbers or hydraulic power (U.Rh ).
This fact is clearly observed in the retardance curves shown in Figure 2, where
the friction factor (DarcyWeisbach f ) falls progressively. These kind of curves
were first plotted by Ree and Palmer [18], who used a wide range of natural grasses
in their tests. In our study, a minimal value of f is obtained for totally deflected
plants (prone condition where the relative deflection of plant k/ h = 0.40.5) and
this value is very similar to the non-vegetated friction factor (fgravel = 0.15) also
measured. In consequence, the influence of flexible vegetation is attenuated in high
velocity fields, such as in a natural river flooding. Also it may be observed that
the variation of plant density is a factor that becomes minimized for increasing
velocities. In Figure 3, our data is compared with Kouwens retardance curves for

76

D. VELASCO ET AL.

Figure 4. Linear adjustment between dimensionless flexural parameter and relative deflection
of plant k/ h .

plastic strips and, regarding the differencies in the MEI parameter, the evolution of
friction factors becomes really well adjusted. The relationship between the friction
factor and the Reynolds number defined in the same way as done by Kouwen [7],
also shows that the different plant densities do not have much affect on the overall
behavior but there may be differences with a factor of 2.
3.2. M ECHANICAL RESPONSE OF VEGETATION TO THE FLOW
Another subject of this study is the deformation response of the flexible element to
flow stresses. Thus it was experimentally measured as the progressive decrease of
the vertical height of the deflected plants, k, as the higher flows produced increasing
bed stresses ( = U 2 ) and forced the submergence of the plants. The elastic
behavior of stems is confirmed according to the mechanical and dimensional model
proposed by Kouwen, in spite of some experimental data dispersion.
Figure 4 shows the relative plant deflection k/ h , where h is the non-bended
plant height, versus a non-dimensional parameter, similar but modified slightly
with respect to the well-known Kouwen parameter. This modified parameter, which
takes into account plant and distribution geometry, mechanical properties and flow
conditions, is non-dimensional and may be defined by using k/a instead of M as


(k/a) E I 1/4 1
,
Ko =
U 2
h

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

77

where the variables indicate, respectively, k = deflected plant height, a = interplant length, E = stiffness modulus of the plant, I = inertial modulus, U = bed
shear velocity and h = non-bending plant height. The k/a ratio that appears
in this parameter describes the relationship that exists between the characteristic
roughness scales as k represents the vertical roughness of the canopy, whereas the
interplant length scale defines a lateral spatial distribution and density of roughness elements. Combination of both scales is important for resistance to flow. An
excellent linear fit is observed in Figure 4, where all data was included. Anyway,
new tests are necessary to confirm that the substitution of M for k/a in Kouwens
MEI model, produces a better predictive relationship but it is important to remark
that all tests were set up under a wide range of submergence conditions and it is
assumed that emerging plant conditions do not comply with the law
k/ h = m Ko + c,
with m, c the linear fit constants.
3.3. V ELOCITY PROFILES
It is known that the longitudinal mean velocity (U ) profile differs from the typical
boundary-layer model (law of the wall) because of the presence of vertical elements
(flexible stems) into the flow. A different momentum diffusion mechanism is developed. As a first macroscopic result, the average velocity U decreases and water
depth increases. This effect of longitudinal velocity reduction (retardance) is very
important inside the vegetated region, and very low values of U were registered.
Figure 5 shows an example of velocity profiles for many submergence conditions
of plants. The well-known logarithmic profile (von Krmns universal law) is no
longer valid in this roughness condition. Two different regions can be distinguished
in all the profiles; the first is the outer region that spans from the top of the plants
until the free-water surface (z > k), and the second region is the inner, deeper zone
from the bed to the extreme of the plant, where drag forces are present. Figure 6
shows experimental data as non-dimensional velocity U/Uk versus the ln(z/k)
for different densities of plants. The slip velocity, Uk , is defined as longitudinal
velocity just in the top of the canopy, where a boundary layer between the inner
and outer region is registered. The slip velocity represents a relevant characteristic
velocity and it is responsible of the improved data collapse in Figure 6.
In the outer region, the velocity profiles fit well a logarithmic distribution [8,
9], but von Krmn turbulent diffusion constant ( = 0.41) is not confirmed in
our data and lower values were estimated. In the inner region, velocity profiles
are complex but they will be commented upon later. Some authors explained this
region using a uniform and constant velocity scheme, but some vertical gradients
of velocity are registered which are governed by obstruction capability (frontal
areas and deformation of stem and leaf). In this sense, it is important to know
the relation between vertical biomass distribution and velocity profiles. Also, these

78

D. VELASCO ET AL.

Figure 5. Variation of U velocity profiles for different density and discharge. Circles indicate
deflected plant height.

Figure 6. Dimensionless logarithmic velocity profile, where Uk = slip velocity.

regions correspond to two different energy transfer schemes. Experimental studies


by Nepf [14, 15] with semi-rigid plants show that turbulent mechanical production
and von Krmn wakes behind obstacles are dominant in the inner region and a
constant velocity profile is developed. A logarithmic profile is explained above
vegetation domain because of the importance of shear stresses (shear turbulence
production). Our data agrees with that of Nepf, but in our gravel bed tests, particle
roughness increases vertical momentum diffusion and velocity gradients near the

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

79

bed. So, in the inner region, both pressure turbulent production and shear turbulent
production are strongly present.
3.4. T URBULENT REYNOLDS STRESSES
An intensive analysis of local velocity fluctuations (u = U + u ) from temporal
series at 25 Hz measuring frequency, led us to calculate second-order statistics in
vegetated and non-vegetated tests. Vertical distribution of Reynolds stresses were
calculated in order to characterize the momentum diffusion mechanisms. Turbulent
intensity T , turbulent kinetic energy K, and Reynolds stresses xz , xy and yz are
defined as
T =

2
3

K 1/2
U

with the turbulent kinetic energy


K=

1 2
(u + v 2 + w 2 ),
2

xz = u w  ,

xy = u v  ,

yz = v  w  ,

where U is the average longitudinal velocity and u , v  , w  are longitudinal,


transversal and vertical velocity fluctuations, respectively. The turbulent intensity
represents an isotropic velocity fluctuation average, whereas vertical Reynolds
stress (xz ) induces longitudinal to vertical momentum exchange and then it
shows vertical mixing activity. Horizontal Reynolds stress (xy ) conversely defines
momentum diffusion in an horizontal plane.
The turbulent measurements clearly show that the flow regime is highly anisotropic and that Reynolds stresses are very useful for analyzing the velocity field.
Plant deflected height has a great influence on turbulent intensity T . In erected
plant conditions, showing no significant deformation, a local maximum of T is
registered in the extreme of plants which is more noticeable in higher plant density
experiments, but it was shown that the maximum moves downwards for increasing
flows and the maximum in turbulent intensity is located near the bed in prone plant
conditions.
The top of canopies (z = k) locates the highest velocity gradients and turbulent
stresses. This zone is a geometrical discontinuity to roughness and supposes a
boundary between both very different turbulent production schemes. Transversal
Reynolds stresses (xy ) confirm this point. In non-vegetated tests (only with a
gravel bed), vertical stresses dominate transversal stresses, which are negligible.
But in vegetated conditions, the tendency changes and, inside the plants, transversal
stresses increase and overcome vertical ones. Development of pressure gradients
in the form of turbulent wakes behind obstacles generates this important horizontal
diffusion of momentum. Once above top of the canopies, where no obstruction

80

D. VELASCO ET AL.

Figure 7. Reynolds stresses profiles for (a) vegetated channel, (b) non-vegetated channel.

Figure 8. Dimensionless Reynolds stresses profiles for vegetated channel.

exists, anisotropy governs and transversal effects gradually become negligible


(Figure 7).
In Figure 8, we plot two different Reynolds stress profiles which are nondimensional using gravity bed shear stress . Measured values are lower than a
2D flow theorical distribution. The answer is the influence upon xz of secondary
currents in a rectangular flume [16]. In the gravel test, ratio h/B = [0.10.3] is
high enough to involve wall shear effects and then large length-scale structures
are developed across the channel. They contribute to the gravity shear stress
absortion. On the other hand, Figure 9 shows an accurate fit between measured
Reynolds vertical stresses and gravity shear stress profile in the free zone. These
test were carried out in a 2.5 m width flume, so the ratio h/B < 0.1. Secondary
currents become very weak and then vertical measurements at the center of the
channel are not affected. Apart from secondary currents, it is very important to
assume that advection of v and w components in the velocity field and transversal momentum diffusion are not negligible during the presence of vegetation. In
addition, highly anisotropic turbulence is generated inside the vegetation domain.

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

81

Figure 9. Vertical profiles of three different tests for M = 25 plants/m2 (sand bed). Dimensionless plots correspond to longitudinal velocity U , Reynolds turbulent stresses and turbulent
intensity for each component.

As mentioned above, it is very interesting to characterize the amount of stress


that is absorbed by a single plant because it represents a basic approach to understand resistance to flow due to these complex elements. A clear relationship
exists between the vertical distribution of biomass (frontal area) and longitudinal
momentum absorption. Reynolds x-equation in the outer region (above plants, no
obstruction) shows turbulent stresses dissipating gravity forces, which are proportional to energy slope. But inside the vegetated zone, flexible stems absorb
momentum, and bend. Then drag forces (vegetative drag forces) appear in the
Reynolds equation. Leaves and stems absorb available gravity stresses (normal
and tangential shear stresses to surface), but efficiency depends on vertical biomass structure. Remaining gravity stresses are transformed into turbulent stresses
(mainly in transversal stresses) through pressure gradients and vortex streching
around obstacles [3, 20]. Nepf [14] observed a characteristic depth inside the
canopy where vertical turbulent stresses xz becomes zero. Momentum absorbing
capability of vegetation and gravity stress (i.e., submergence condition) defines
this region near the bed where no vertical momentum is diffused, and the velocity
profile is very uniform. In our flume tests, gravel roughness (macroroughness)
was responsible for shear stress production near the bed, so that region was never
observed. To avoid this matter, a 2.5 m width sand bed flume was used to investigate sediment transport and shear stresses. Figure 9 shows three different tests and
velocity, turbulent stresses and turbulent intensity profiles are plotted, respectively.

82

D. VELASCO ET AL.

The submergence ratio is low, gravity stresses are absorbed efficiently by the plant,
which is bending, and vertical turbulent stress comes to zero near the top of the
canopy. This position, where xz = 0, is called the penetration depth, z = hp . No
vertical exchange has been reported inside this region [15], which is called the longitudinal exchange zone (z < hp ). In our experiments, negative vertical turbulent
stresses have been measured there, and then negative velocity gradients are developed. Notice the importance of xy and yz in this region. Vertical distribution of
biomass is responsible of the presence of a secondary vortex which locally reverses
the longitudinal to vertical momentum exchange mechanism (xz < 0). The palm
shape or typology of the plastic elements used in our tests accumulates its main
absorbing area at the upper zones where the leaves become totally separated. This
characteristic distribution of palm shape was registered even for totally prone,
bending conditions, quite distinct from the streamlined shapes that are typically
observed in natural riverine species.
Two instances of coupling between the velocity profiles and the horizontal transport of vertical momentum, reflected by the xz profiles are apparent for most of
the experiments. The zero values of vertical turbulent stress are also apparent as
local maxima or minima on the S-shape velocity profiles. We also observe that
at the heights where there is a minima of the xz profile, most of the time, at the
same height there is an inflection point or change or curvature in the corresponding
velocity profile.
In consequence, any attempt to numerically fit a velocity profile in the presence
of vegetation should take into account turbulent stress distribution, which governs
velocity gradients and inflection points.
As a result of the budget ot momentum, mean effective bed shear stresses
are substantially reduced, but an important activity of horseshoe swirls is noticed
(but not yet directly measured). Local scour induces load sediment transport, which
modifies the geometry around the canopy.
4. Integral Scales
The autocorrelation functions Re( ) were calculated at different depths, according
to the expression
Re( ) =

u (t).u (t + )
u2

where t is a time scale and u the velocity fluctuation in the flow direction x.
This normalized parameter represents a correlation factor of a recurrent phenomenon that is acting in the flow at different time scales. Assuming Taylors
frozen whirl hypothesis, l = U . , it is possible to find a direct relationship
between time and space scales l, through the mean longitudinal velocity U ,
for homogeneous turbulent conditions. It is necessary to note that homogeneous
turbulence conditions are not present in our tests, but they are accepted as a first

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

83

Figure 10. Definition of geometrical interplant length a.

approximation to investigate the general turbulence state, for lack of direct Lagrangian measures in the flume. As a generic concept, these derived spatial scales
represent the length or the dimension of turbulent structures (whirls) acting and
developing in the streamline direction. Some incoherence could be noticed between
a theoretical Re( ) function and the experimental one, because the tangent at = 0
should be horizontal (symmetrical function). Our results do not show this typical
shape of function because of an insufficient measurement frequency (25 Hz). Apart
from this instrumental limitation, the calculation of integral scales is very interesting and sheds new light on the statistical structure of the turbulent interplant flows.
The integral scale Lx is considered to be a characteristic length scale which is
able to transfer an equivalent momentum by the local turbulence. This parameter
is obtained through integration of Re( ) as

Lx =

Re(l) dl.
0

Some interesting relations are obtained from the analysis of integral length scale
vertical profiles for vegetated conditions and densities. The data shows a dependence between the geometrical-spatial distribution of roughness (plants) and the
turbulent structures. For this purpose, we defined and used a geometrical length
scale, a, as the interplant transverse length, see Figure 10. This value represents
an average transversal distance between plants in the bed. If a very uniform distribution of canopies is established, a is expressed in terms of the density M
as a = 1/M 1/2 . Relationships found between the integral scale Lx and length
a change from the emergent plant condition to the totally submerged condition.
In the first case, when stems are emerging from surface, results show a constant
ratio L/a = 0.3 along the vertical profile and then characteristic vortices are very
uniform and do not merely develop towards the surface. Horizontal whirls (vertical
rotation axis) governs the flow and secondary currents become very important. But,
in totally submerged plant conditions (h > k) there is a tendency, a distribution in
the L/a ratio, inside and above the vegetation (Figure 11). In spite of dispersion

84

D. VELASCO ET AL.

Figure 11. Dimensionless plot of the vertical evolution of integral scales for different plant
density tests. z/k = 1 denotes the top of the deflected canopy.

of experimental data, integral scales increase just inside the vegetation (z < k)
as a consequence of high momentum exchange, but ratios L/a are smaller than
1. Above the top of the plant, it is in the outer region (z > k), where L/a are
greater than 1, but no distribution has been fitted accurately. These results confirm
the idea that the spatial distribution of roughness elements sets a limit and controls
development of large-scale turbulent structures in very concise ranges.
4.1. D ESCRIPTION OF THE FLOW
The top of the deflected plant canopy is seen as the transition point between to
different turbulent production mechanisms but inside the plant canopy we may also
distinguish the region close to the bed affected by shear. The Reynolds stresses at
different heights exhibit a very strong anisotropy, as seen in Figure 9. The different
mechanisms, such as the vertical flapping of the very deflected leaves, may also
produce some resonance for certain flow parameters. The same is true for resonant coupling of von Krmn vortices among the stems. For example, at the level
z/ h = 0.6 of Figure 9b) the longitudinal rms velocities u are much larger than v 
and w  and the Reynolds tensors would be very elongated. On the other hand, at
heights z/ h = 0.20.4 the rms turbulent velocities are more isotropic. This is also
appreciated, for example, at z/ h = 0.3 for experiment (c) where a maximum in
U/U is also apparent.
Very near the ground the strong vertical shear detected also produces elongated
anisotropic Reynolds stresses. The spectra from data at different heights is also

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

85

Figure 12. Frequency spectra for three different heights within the plant canopy (h/k = 2.57),
corresponding to heights (z/k = 0.12) near the gravel bed, (z/k = 1) at the deflected plant
top and (z/k = 1.59) corresponding to a point above the canopy.

instructive. For example, in Figure 12 a comparison of the frequency spectra from


velocity measurements u at three heights, above the canopy, at the level k, and very
near the ground, show that the turbulence is completely developed. The same data
converted to a wavenumber spectra may be compared with the three forcing scales
imposed by the dynamic flow geometry; namely: the deflected plant height k, the
interplant distance a, and the much smaller stem effective diameter d. The fact that
near the ground there is much less energy available for turbulence production in
spite of the strong vertical shear is shown by the area below the inertial subrange.
At the same time, the higher velocities above the canopy do not produce more
turbulence possibly because the lack of the stem flapping.
5. Discussion and Conclusions
Reynolds stresses distribution, turbulent production and turbulent integral length
scales were calculated for the available data showing very detailed vertical profiles of mean and turbulent relevant parameters. The realization of the strong
anisotropy from the ADV measurements and some visual observations of the flow
structure shows that there are many possible coupled length scales and coherent
structure mechanisms. Thus, a future investigation should also include PIV or flow
visualization for a wide range of parameters.

86

D. VELASCO ET AL.

Figure 13. The same data plotted in Figure 12 for the cited three levels converted to wavenumber spectra. The dashed lines correspond to the wavenumbers due to the deflected plant height
Kk = 2/k, to the plant separation Ka = 2/a and due to an average plant diameter
Kd = 2/d .

The ultimate objective in the future will be to connect this complex hydrodynamic turbulent model with sediment transport phenomena, load or suspended
sediment, that takes place inside vegetated regions or wetlands at natural streams.
Several authors have reported that the turbulent structures, which are generated
inside and above vegetation, induces general sedimentation and reduces local
scouring effects.
The reduction of resistance to flow in flexible vegetation has been confirmed
for increasing flows and low relative roughness (k/ h). Under totally prone plant
conditions, variations in friction factors for different plant densities are reduced
and their values fall to a nearly asymptotic friction factor, equivalent to the nonvegetated one.
There is a direct relationship between vegetative effective roughness and local
plant deformation (deflected plant height). In that sense, the dimensionless function
based on Kouwens investigations is confirmed. Density, stiffness, and geometric
moment of the vegetative model, combined with flow conditions, causes resistance
to flow. The use of the average interplant length scale a, and its use in the presented
modified Kouwen parameter seems a better predictive indicator.
Two different momentum exchange mechanism were analyzed: inside the vegetation, where transversal turbulent stresses dominate vertical components (spatial

TURBULENT CHARACTERIZATION OVER FLEXIBLE VEGETATED LININGS

87

gradients of pressure in wakes), and above plants, where turbulent production by


Reynolds vertical stresses develops a logarithmic velocity profile. The anisotropy
of the turbulence is apparent at different heights and due to several different mechanisms. The Reynolds stress ellipsoid is elongated in the u axis both near the
bed and near the deflected canopy height. Vegetation geometry (vertical biomass
distribution) and deformation capabilities defines absorbed stresses due to plants
and reconfigures Reynolds stress tensor. Pressure gradients and secondary currents
govern the lowest zone of vegetation domain, reducing the amount of bed shear
stress.
The calculations of integral length scales clearly show the distribution of vortical structures with height. Large-scale turbulent structures are limited by spatial
and geometrical roughness scales and the spectral information at different heights
is very helpful. In the future, a more detailed flow visualization of the vortical
structures enhanced by the stems and the conditions for resonant interaction will
be attempted.
Acknowledgements
We thank S. Raffaeli for performing some of the experiments and Alexei Platonov
for technical assistance. Acknowledgement of the support of the Spanish Ministry of Science and Technology (REN2000-1013 HID) and the Network of Fluid
Dynamics and Geophysical Turbulence (XT-2000-016) is due.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

Ben Mahjoub, O., Non-local dynamics and intermittency in non-homogeneous flows. Ph.D.
Thesis, Universidad Politecnica de Catalunya (2000).
El-Hakim, O. and Salama, M., Velocity distribution inside and above branched flexible
roughness. ASCE, J. Irrigation Drainage Engrg. 118(6) (1992) 914927.
Finnigan, J., Turbulence in plant canopies. Annual Rev. Fluid Mech. 32 (2000) 519571.
Ikeda, S. and Kanazawa, M., Three-dimensional organized vortices above flexible water plants.
ASCE J. Hydraulic Engrg. 122(11) (1996) 634640.
Knight, D.W., Boundary Shear in smooth and rough channels. ASCE J. Hydraulics Div.
107(HY7) (1981) 839851.
Kouwen, N., Unny, T.E. and Hill, H.M., Flow retardance in vegetated channels. ASCE J.
Irrigation Drainage Div. 95(IR2) (1969) 329342.
Kouwen, N. and Unny, T.E., Flexible roughness in open channels. ASCE J. Hydraulics Div.
99(HY5) (1973) 713727.
Kouwen, N. and Unny, T.E., Flexible roughness in open channels. ASCE J. Hydraulics Div.
(HY1) (1975) 194196.
Kouwen, N. and Li, R., Biomechanics of vegetative channel linings. ASCE J. Hydraulics Div.
106(HY6) (1980) 10851103.
Kouwen, N., Flow resistance in vegetated waterways. ASCE J. Irrigation Drainage Engrg. 5
(1992) 733743.
Kouwen, N. and Fathi-Moghadam, M., Friction factors for coniferous trees along rivers. ASCE
J. Hydraulic Engrg. 126(10) (2000) 732740.

88
12.
13.
14.
15.
16.
17.

18.
19.
20.

D. VELASCO ET AL.

Lopez, F. and Garcia, M.H., Mean Flow and turbulence structure of open-channel flow trough
non-emergent vegetation. ASCE J. Hydraulics Div. 127(HY5) (2001) 392402.
Naot, D., Nezu, I. and Nakagawa, H., Hydrodynamic behaviour of partly vegetated open
channels. ASCE J. Hydraulic Engrg. 122(11) (1996) 625633.
Nepf, H.M., Drag, turbulence and diffusion in flow through emergent vegetation. Water
Resources Res. 35(2) (1999) 479489.
Nepf, H.M. and Vivoni, E., Flow structure in depth-limited,vegetated flow. J. Geophys. Res.
105 (2000) 2854728557.
Nezu, I. and Nakagawa, H., Turbulence in Open Channel Flows, Monograph Series IAHR.
Balkema, Rotterdam (1993).
Raffaelli, S. and Domenichini, F., Reistenza al moto in un alveo vegetato: Indagine sperimentale
di laboratorio. In: Proceedings of the 28th Convention on Idraulica e Costruzioni Idrauliche
Potenza (2002).
Ree, W.O. and Palmer V.J., Flow of water in channels protected by vegetative linings. Soil
Conservation Service, US Department of Agriculture 967 (1949).
Rouse, H., Elementary Mechanics of Fluids. Dover Publications, New York (1946).
Simpson, R.L., Junction flows. Annual Rev. Fluid Mech. 33 (2001) 415445.

Potrebbero piacerti anche