Vous êtes sur la page 1sur 9

Proceedings of the 19th IAHR-APD Congress 2014, Hanoi, Vietnam

ISBN 978604821338-1

STOCHASTIC LAGRANGIAN MODELLING OF VERTICALLY UPWARD SEDIMENT-LADEN BUOYANT JETS

S. N. CHAN(1) & J. H. W. LEE(2)


(1)
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore
email: snchan@ntu.edu.sg
(2)
Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Hong Kong SAR, China
email: jhwlee@ust.hk

ABSTRACT
A vertically upward sediment-laden buoyant jet is an important flow phenomenon in geophysical science and
engineering as it resembles the dynamics of volcanic eruptions, deep sea hydrothermal vents and wastewater discharge.
In an upward buoyant jet, sediment particles are carried by the rising jet fluid until they reach the maximum height of
rise or the water surface. Sediment fall-out occurs at the jet edge and/or from the radial spreading current and re-
entrainment of sediment occurs. For the first time, a three-dimensional stochastic Lagrangian particle tracking model for
predicting the deposition and dynamics of a vertical sediment-laden buoyant jet is developed. The model solves the
governing equation of particle motion in a predetermined fluid velocity field superimposed with stochastic turbulent
fluctuations. The three flow regimes involved in a buoyant jet - the turbulent jet flow, jet entrainment-induced external
flow and surface spreading current, are modeled using validated semi-analytical methods: (1) the jet mean flow velocity
is determined using a jet integral model; (2) jet-induced external irrotational flow field is computed by the point sink
approach; (3) surface spreading current is predicted using an integral model accounting for the interfacial shear.
Turbulent velocity fluctuations are modelled by an autocorrelation function that mimics the trapping of sediment
particles in turbulent eddies. Root-mean-square (RMS) turbulence velocity and turbulent time scale are estimated from
best-fitted self-similar profiles of turbulent kinetic energy and dissipation rate derived from a computational fluid
dynamics solution of vertical buoyant jets. Model prediction shows excellent agreement with previous experimental
data of bottom sediment deposition. Model prediction reveals the increase in sediment concentration due to sediment
reentrainment which has important significance for the dynamics of vertical sediment jets.

Keywords: buoyant jets, sediment transport, gravity current, particle tracking model

1. INTRODUCTION substances. Many emerging pollutants such as endocrine


disrupting compounds and heavy metals tend to attach to
Vertical particle-laden jet is a two-phase flow that is
small suspended particles in sewage. The suspended
commonly found in natural environment and
sediment can also be transported to areas far from the
engineering processes. Volcanic eruption and black
sewage outfall, adversely impact on the aquatic
smokers released from hydrothermal vents in a deep
environment and benthic fauna. Under the increasing
ocean are some examples in natural environment. In a
environmental impact awareness of these trace
volcanic eruption, ash and solid particles are carried by
pollutants, it is also necessary to assess the transport and
the heated gas flow to a height of neutral buoyancy and
fate of sediment discharged from marine outfalls.
spread laterally (Veitch and Woods, 2000). Particle with
various sizes falls out from the ash plume column and Vertical upward particle-laden buoyant jets have been
the ash cloud to form layered deposit on the ground. studied experimentally (e.g. Carey et al. 1988; Sparks et
Hydrothermal vents release hot buoyant mineral rich al. 1991; Ernst et al. 1996; Zarrebini & Cardoso, 2000).
fluids as high as 400 oC in a deep ocean of 1000-3000m. Neves and Fernando (1995) developed empirical relations
Mixing of the hot vent fluid with cold sea water (2-4oC) for predicting bottom sediment deposition of vertical
triggers chemical reactions, resulting precipitates momentum jets. Ernst et al. (1996) developed theoretical
consisting of minerals which eventually deposits on the models for sediment deposition around a vertical
seabed (Dissanayake et al. 2014). The seabed around upward sediment jet, accounting for the sediment fall out
hydrothermal vents is a potential source of minerals and and reentrainment at jet margins. Theoretical models
metals. Partially-treated municipal or industrial have also been developed for predicting the deposition
wastewater is often discharged in the form of multiple from spreading current (Sparks et al. 1991; Zarrebini &
vertical upward buoyant jets. The effluent often contains Cardoso, 2000; Cardoso & Zarrebini, 2002). However
inorganic and organic solids that may settle close to the these theoretical models usually account for a single
source, giving rise to the formation of sludge banks. The settling mechanism only and lacks generality in
bottom sediment can be a source of oxygen demand and predicting combinations of settling mechanisms.
reservoir of micro-organisms, nutrients and toxic

1
This paper presents the development of a 3D stochastic carried upwards but reaches its maximum height of rise.
particle tracking model for vertical sediment-laden Particles near the edge of the plume, will first settle out
buoyant jets, based on the previous work of Chan along the jet boundary while those near the centerline is
(2013a,b) and Chan et al. (2014) on horizontal buoyant pushed out to the edge. Particles fall to the bottom near
jets. The formulation of the numerical particle tracking the plume edge with attraction by jet entrainment.
model is first presented, especially on the unique and Particles are re-entrained to the plume and some are
simple semi-analytical prediction of the entire mean flow recycled several times before they deposit (Neves and
field and the generation of turbulent fluctuations. The Fernando, 1995; Ernst et al. 1996).
model prediction is validated with laboratory data of
The second mechanism is fall out from spreading layer
sediment bottom deposition of vertical buoyant jets. The
(Fig. 1b). Particles reach the water surface are transported
influence of reentrainment on the jet particle
laterally in the horizontal spreading current. Turbulence
concentration is discussed based on model prediction.
maintains the particles in suspension, until they fall out
from the interface of the spreading current and the
2. PROBLEM DEFINITION
ambient water. The falling particles follow a curved
In a vertically upward sediment jet, sediment particles trajectory towards the nozzle due to the ambient
are directed by the jet flow upwards. The height of rise of entrainment flow. Particles are re-entrained back to the jet
particles depends on the jet volumetric, momentum and when they are close to the jet edge. Inside a critical radius
buoyancy flux of Q0, M0 and B0 and an initial sediment of the spreading layer particles must be re-entrained. The
concentration of C0. When the jet velocity is sufficiently size of this region depends on the strength of the gravity
higher than the particle settling velocity ws, the sediment current and the settling velocity of particles (Sparks et al.
particles are carried upwards. It is observed when the 1991; Zarrebini and Cardoso, 2000; Cardoso and
sediment concentration is less than 0.1% by volume, the Zarrebini, 2002). It is now to develop a model that is
reduction in jet buoyancy and influence on jet dynamics capable of predicting the sediment dynamics and
by the sediment are negligible (Veitch and Woods, 2000; deposition in both fall-out regimes.
Cuthbertson and Davies, 2008). Two regimes of particle
fall-out are observed. 3. NUMERICAL MODEL

0.6
3.1 Equation of particle motion
0.55

0.5 The present modelling approach is to track the position


0.45 evolution of a large number of particles released from the
0.4 jet nozzle in the flow field based on the equation of
0.35
motion of a spherical particle:
z (m)

0.3

0.25
du p 1
0.2 r pV p = ( r f - r p ) Vp g - r f CD Ap up - uf ( up - uf )
dt 1 44 2 4 43 1 2 4 4 4 44 2 4 4 4 4 43
0.15 gravity
drag
0.1

du
du du
+ r f V p f + r f CM V p p - f
0.05

dt dt dt [1]
1 44 2 4 43f 1 4 4 4
0
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
x (m) 4 2 4 4 4 43f
fluid acceleration added mass
(a) Settling out from jet margin - Case 710 (Ernst et al. 1996):
(
t d up - uf )
u0 = 0.75m/s, Fr = 32, ws = 5.3 cm/s 3
+ d 2 pr f m dt dt

0.6 2 t -t
1 4 4 44 2 04 4 4 43
0.55 Basset history
0.5

0.45
where up = (up, vp, wp) is the particle velocity; uf = (uf, vf,
0.4 wf) is the fluid velocity; Vp = d3/6 is the particle volume;
0.35 Ap = d2/4 is the particle projected area; p is particle
z (m)

0.3 density; f is fluid density; g = (0, 0, g) is gravitational


0.25
acceleration (g = 9.81 m/s 2); CD is drag coefficient taken
0.2
as a function of the particle Reynolds number Rep = |
0.15
upuf|d/ using the empirical equation of Clift et al.
0.1

0.05
(1978):
0
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
24 0.42
x (m)

(b) Settling out from spreading current - Case 710 (Sparks et al.
CD =
Re p
( 1 + 0.15 Re p 0.687 ) +
1 + 42500 Re p -1.16
[2]
1996): u0 = 0.0215m/s, Fr = 0.5, ws = 0.37 mm/s
CM = 0.5 is the added-mass coefficient (Lamb, 1932); d is
Figure 1. Visualization of two sediment fall-out mechanisms of
vertical sediment buoyant jet as predicted by the particle model in particle diameter; = /f is kinematic viscosity of water
the present study. The dash lines represent the jet and spreading (10-6 m2/s); and is dynamic viscosity of fluid (10 -3
current boundaries. Water surface is at z = 0.5m. kg/m/s). t is the time from the start of computation and
is a dummy variable for integration. The subscripts f
The first mechanism is fall-out from jet margin (jet near and p represents the fluid and solid phases respectively.
field) (Fig. 1a). As jet velocity reduces to a level similar to
the settling velocity of particles, the sediment is no longer
2
The left hand side of the equation of motion [1] the entrainment hypothesis. The unknown jet trajectory is
represents the acceleration of the sphere and the right viewed as a sequential series of non-overlapping plume
hand side represents the forces acting on the spherical elements which increase in mass due to entrainment (Fig.
particle: body, drag, fluid acceleration, added mass and 2). JETLAG tracks the average properties of a plume
Basset. The fluid acceleration term represents the gradient element by conservation of horizontal and vertical
along the trajectory of a fluid particle and determined momentum, water and tracer masses.
based on the mean flow. The Basset history force
represents the temporally changing viscous shear force
For a buoyant jet in stagnant ambient, the increase in jet
acting on the particle due to the velocity gradient
volume flux dQ along the jet trajectory s is expressed in
between the moving particle and the ambient. Sensitivity
tests (Chan, 2013) have shown that the Basset force is terms of the local jet velocity and width via a shear
negligible in the prediction of sediment deposition entrainment coefficient g. It assumes that the local jet
profiles for sediment sizes of 100-700 m, as in the size of radial entrainment velocity ve is proportional to the local
natural sands. streamwise maximum jet velocity uc.

The fluid velocity uf is composed of the mean flow dQ


velocity and the turbulent fluctuation u. = 2 b g v e = 2 b g g uc [7]
ds
u f = u + u' [3] The jet entrainment coefficient g is dependent on the
is determined based on the analytical mean flow local jet densimetric Froude number (Fox, 1970; Lai and
velocity of buoyant jet, jet-induced external flow and Lee, 2012b):
surface spreading current (Session 3.2). u is determined
by a stochastic approach from computation fluid 0.554 sin k F = uk
dynamic (CFD) derived turbulent kinetic energy and g = 0.057 + , l k
dissipation rate (Session 3.3). The particle velocity up is to Fl 2 a gb g , k
be solved by numerical integration together with the
[8]
particle position xp using a second-order predictor-
corrector scheme (Eq. [4]). where is the jet orientation angle with respect to the
vertical. Fl is the local densimetric Froude number; where
dx p the subscript k denotes the kth jet element; k is the
up = [4]
dt maximum density difference between jet fluid k and
ambient fluid a. The entrainment coefficient has the
In turn, the particle position provides the mean flow
asymptotic values of 0.057 for a pure jet (Fl ) and
velocity and the turbulent properties (turbulent kinetic
0.088 for that of a pure vertical plume (Fl = 4.2). The
energy and dissipation rate).
equivalent Gaussian profiles (Eqs. [5] & [6]) of all
quantities can be readily transformed from the top-hat
3.2 Semi-analytical mean flow models
profiles in JETLAG prediction by conservation of mass
3.2.1 Turbulent jet mean flow and momentum (Lee and Chu, 2003).

For a vertical buoyant jet, the streamwise mean velocity u The jet potential core is separately determined. The
can be described as a Gaussian profile trajectory of the potential core is predicted with the
balance in horizontal and vertical momentum (Lee and
r2 Chu, 2003). JETLAG computation starts at the end of the
u(s , r )
= exp - 2 [5] potential core and ends as the upper jet boundary hits the
uc (s ) bg water surface.

where uc is the centerline maximum velocity and bg is the mi Jet element


Gaussian half-width defined as u(bg)/uc = exp(-1). By
continuity, the radial velocity distribution vr can be found uc(r)
ve = uc
as (Lee and Chu, 2003):
g
(xp , yp , zp )

2
2 2
1 - exp(- r ) - r exp(- r )
v r (r ) b g b g
2
bg 2 db g
= , = z
g uc r /bg ds m2
m1 y
[6]
D x
where g is the jet shear entrainment coefficient.

In the present study, the mean buoyant jet flow is Q0, M 0, B0


modelled using the well-validated integral Lagrangian Figure 2. The JETLAG model (Lee and Cheung, 1990; Lee and Chu,
model JETLAG (Lee and Cheung, 1990; Lee and Chu, 2003) for a vertical jet and the point sink approach for
2003). JETLAG predicts the trajectory and mixing of determining the external flow (Lai, 2009; Lai and Lee, 2012a).
buoyant jets in a wide range of ambient conditions using

3
3.2.2 External entrainment flow dQs d( 2 rhs ur )
= = 2 w e r [12]
The entrainment by the buoyant jet acts as a sink to the dr dr
surrounding fluid and creates an external irrotational
flow towards the jet. The jet entrainment induced dM s d( 2 rh s ur 2 )
=
1 d rg s ' hs 2
= 2 -
(
- i ur 2 r
)
external flow is determined by the point sink method
introduced by Lai (2009) and Lai and Lee (2012a). The
dr dr 2 dr
[13]
buoyant jet is regarded as a number of point sinks along
its trajectory. The strength of point sink i at position (xi, yi, dBs d( 2 rhs ur g s ')
zi) is the entrainment per unit length (Fig. 2) = =0 [14]
dr dr
dQ i
mi = - = -2 b g g uc [9] where Qs, Ms, Bs are the volumetric, kinematic
ds momentum and specific buoyancy fluxes of the surface
spreading layer at radial distance r from the center of
where Qi is the entrainment flow. The velocity potential
impingement respectively. i = 0.003 is the interfacial
i induced by this jet element at an arbitrary location (x,
friction coefficient (Akar and Jirka, 1994). we is the
y, z) is given by
entrainment velocity from the interface between the
spreading current and the ambient fluid, given as a
mi
i = - s [10] function of a Richardson number for the spreading layer
4 r (Akar and Jirka, 1994).

where r = [(x - xi)2 + (y - yi)2 + (z - zi)2]1/2 is the distance gs ' hs


from the sink i (Fig. 2). we = 0.0015 urRi-2, Ri = [15]
ur 2
The three-dimensional flow field (ui, vi, wi) induced by
The variables hs, ur and gs' in governing equations can be
this single jet element is given by differentiating the
solved numerically with the initial conditions specified
velocity potential i with respect to x, y and z directions.
according to the JETLAG prediction at the jet terminal
By summing up the induced velocities at a point (x, y, z)
level, i.e. the upper jet boundary hits the water surface.
outside the jet by all the sinks (Fig. 2), the total induced
The initial flow rate Qs(r0) and buoyancy gs'(r0) of the
velocity (u, v, w) at that point by the jet entrainment can
spreading current are taken as the jet volumetric flow rate
be found as
and buoyancy at the terminal level. The initial radius r0 is
N m i ( x - x i ) s the jet top-hat width at the impingement. The initial
u( x , y , z) = spreading layer thickness hs(r0) is given by 0.08 times
i =1 4 r3 water depth H (Lee and Jirka, 1981). Initial radial velocity
N m ( y - y ) s is given by ur(r0) = Qs(r0)/(2r0hs).
v( x , y , z) = i 3
i
[11]
r0
i =1 4 r
N m ( z - z ) s h0 = 0.08H hs(r), ur(r), gs'(r)
w( x , y , z) = i 3
i
Equivalent
i =1 4
Gaussian wi
r profile Q0 top-hat profile
Jet top-hat
boundary z

where N is the number of jet elements. To account for the


r
free surface and bottom boundary, the method of images
Figure 3. Surface spreading current. H is the distance from jet
is used (Lai, 2009). The external flow computation is nozzle to water surface.
applied to locations greater than three Gaussian jet
widths from the jet centerline. Inside the jet (r < 3bg) the To validate the integral model, a CFD model is developed
mean jet velocity is calculated by Eqs. [5] and [6]. to predict the axisymmetrical surface spreading current
induced by a vertical buoyant jet. The axisymmetrical
3.2.3 Surface spreading current
Reynolds-averaged Navier-Stokes equations are solved
using the FLUENT code with the realizable k- turbulence
As the jet impinges the water surface, the jet fluid spreads
model (Shih et al., 1995). Two buoyant jet conditions are
horizontally in a layer. A simple integral approach is
simulated (see Fig. 4). The model covers an area of 0.53m
adopted to predict the spreading layer flow. Assuming
(r) x 0.3m (z) with 100 x 100 cells. The half-jet inlet (D =
the horizontal jet momentum flux has negligible
7mm) is resolved by 5 cells. Grid resolution increases
influence on the spreading layer dynamics, the governing
towards the jet axis and the surface layer. The free surface
equations (continuity [12], radial momentum [13] and
is modeled by zero velocity gradient and zero mass flux
buoyancy [14]) for the steady spreading layer can be
conditions. The bottom boundary is modeled as a no-slip
formulated using the layer thickness hs(r), the radial
wall and the remaining side is modeled as an open
velocity ur(r) and buoyancy gs'(r) = (/a)g (Fig. 3):
boundary of zero excess pressure. By comparing with the
CFD solution, the integral model successfully captures

4
the velocity and reduced gravity distributions of the For application in turbulent jets, the RMS turbulent
spreading current for both cases (Fig. 4). fluctuation is obtained from the turbulent kinetic
energy k using a realizable k- turbulence closure model
(Shih et al., 1995) with CFD simulation of a buoyant jet.
LE and TE can be obtained from Eqs. [18] & [19] using
and . The CFD predicted and are normalized with the
jet mean flow properties, and fitted with an empirical
equation (Eqs. [20] and [21]) to provide their spatial
(a) Case A: u0 = 0.351 m/s; D = 7mm; Fr = 10.0
functions (Fig. 5). The empirical coefficients C1-C6 are
obtained using least-square best-fitting.

r
2
r
2
= C 1 exp - C 2 - C 3 + exp- C 2 + C 3
uc b b

(b) Case B: u0 = 0.146 m/s, D = 7mm and Fr = 4.0 [20]
Figure 4. Comparison of integral model and CFD predicted mean
spreading current velocity ur and mean reduced gravity gs. ( b ) 1 / 3 r
2
r
2
= C 4 exp- C 5 - C 6 + exp- C 5 + C 6
uc b b

3.3 Stochastic modeling of turbulent fluctuations
[21]
Turbulent fluctuations u' are modeled using a stochastic
approach described by Nielsen (1992). Nielsen postulated
C 1 0.2006 C 4 0.2458

the loitering effect for which particles are trapped in
C 2 = 1.4147 , C 5 = 1.2498
turbulent eddies and delayed from settling. The
C 3 0.6647 C 6 0.6594
autocorrelation function has been successfully applied for
the prediction of horizontal sediment-laden momentum It is observed experimentally (e.g. Papanicolaou and List;
and buoyant jets (Chan, 2013b; Chan et al. 2013). Chan Wang and Law, 2002) and numerically (Fig. 5) that the
(2013a) generalized Nielsens autocorrelation function Ri RMS turbulent fluctuation and dissipation rate for jets
for use in governing equation of particle motion for the and plumes are similar thus the same empirical equations
generation of turbulent velocity fluctuation along the (Eqs. [20] & [21]) are adopted for the entire trajectory of a
particle path. general buoyant jet. The turbulent quantities (, LE and
TE) in the spreading layer are determined using a similar
AE 2 u p, i - u i
2
approach as in the turbulent jet.
t
Ri = exp - 1 + [16]
TE 2

With CFD simulation (see Section 3.2.3) of a vertical
buoyant jet impinging the surface the predicted and LE
(obtained from Eqs. [17] & [19]) are normalized with the
In the expression, is the root-mean-square (RMS)
maximum spreading layer velocity usm and Gaussian half
velocity fluctuation which can be determined from the
width hs respectively and shows self-similarity (Fig. 6).
turbulent kinetic energy k.
They are fitted with the following equations in order to
obtain semi-analytical profiles for the spreading current:
2
= k [17]
3 s z / h s - 0.673 2
The subscript i denotes the values in the current time step = 0.251 exp - [22]
usm 1.011
and t is the time-step size; AE = LE / TE. LE and TE are the
Eulerian spatial and time scale of the turbulence
LE z z
2

respectively, and can be estimated as = 0.435 exp - 0.483 [23]
hs hs hs
k 3 /2
LE = C 3 / 4 [18] With the autocorrelation function, the turbulence

fluctuation can be generated by
3 3/4 k
TE = C [19]
u' i + 1 = i u' i + 1 - i 2
2
is the turbulent dissipation rate; C = 0.09. It is of is randomly generated numbers (in x, y, z-directions)
interest to note that the autocorrelation function depends following a Gaussian distribution with zero mean and
on the instantaneous turbulent velocity (u'i, v'i, w'i) and unit variance.
can be viewed as a heuristic local correlation that
captures the vortex trapping of sediment particles. It can
be shown that this form of autocorrelation can generate
the reduction in settling velocity in turbulence (Nielsen,
1992; Chan, 2013a).

5
The experiments of Ernst et al. (1996) represent the
sediment fall-out from jet boundary. They were carried
out in a tank of size 1.2m x 1.2m and water depth of 0.5m
with a jet nozzle diameter of 4mm. Freshwater jets with
different flow rate are directed to the tank filled with salt
water to produce jet initial Froude number of 3-34.
Silicon carbide particles (p = 3.21 g/cm 3) are used as
sediment. Sediment concentration ranges from 0.2 g/L to
2.5 g/L (<0.08% by volume) to avoid influence of
Figure 5. CFD predicted turbulent intensity and dissipation rate in
the jet and plume regime (u0 = 1m/s, D = 6mm, Fr = 50). Frl is the
sediment-induced buoyancy. Eight of the 17 experiments
local densimetric Froude number defined as Eq. [8]. Solid reported are used for the present study and their
symbols: plume regime; open symbols: jet regime. Fitted equation: parameters are shown in Table 1. The first four are jet-
(a) Eq. [20]; (b) Eq. [21]. like cases with high Fr while the other four are plume-
like cases. All bottom deposition data presented in the
paper is normalized with the maximum accumulated
mass deposition near the nozzle (F/Fmax).

Table 1. Experiments of upward vertical sediment-laden


buoyant jets by Ernst et al. (1996). The letter after the case
number: J - jet-like case; P - plume-like case. D = 4mm.

u 0 Sediment
Case
Jet vel. Fr = Set. vel.
(m/s) g0 'D
dia. (m) ws (cm/s)

Figure 6. Turbulent quantities for the spreading layer, predicted 406 (J) 0.478 21.4 275 4.3
by CFD model and fitted with empirical equations, (a) RMS 407 (J) 0.748 34.2 275 4.3
turbulent fluctuation Eq. [22], (b) turbulent length scale, Eq. [23]. 709 (J) 0.748 32.6 135 2.1
710 (J) 0.748 32.4 328 5.3
327 (P) 0.151 6.6 194 2.9
3.4 Numerical Implementation 402 (P) 0.080 3.5 194 2.9
404 (P) 0.080 3.8 115 1.6
The stochastic particle tracking is performed solving the 805 (P) 0.151 3.0 194 2.9
governing equations of particle motion (Eqs. [1] & [4]).
The mean jet flow velocity is determined from the jet
Comparisons of model prediction and experimental data
integral model for turbulent jet, the potential flow model
are shown in Fig. 7 for the four jet-like cases. The
for external flow and the integral model for spreading
comparison is excellent for all the cases, showing a peak
layer (Section 3.2). Turbulent quantities are determined
deposition close to the jet nozzle consistent with the
from Eqs. [17]-[24]. Np = 10,000 particles are used for each
observation. The higher the jet flow, the larger the extent
jet simulation to obtain the deposition profile. Each
of the deposition profile is (Case 406 vs 407). With the
particle shares an equal fraction of the total sediment
same jet flow, the extent of the deposition profile is
mass used in the experiment (Mi = QjC0Texp/Np). Particle's
smaller for larger settling velocity (Cases 709 vs 710). The
are released at equal intervals in a numerical experiment
lighter particle and/or stronger jet flow result in the
of total duration Texp = 10-20 min. The particles are
expelling of particles further away from the jet and
released at the end of the zone of flow establishment
particles deposit in a wider region. A visualization of
according to a Gaussian distribution and tracked until
case 710 is shown in Fig. 1a.
they fall back to the level of the bottom tray (z = 0). Fluid
density f is determined based on the JETLAG and Results for the plume-like cases are shown in Fig.
spreading layer integral model prediction. The radial Compared with the jet like cases, it is observed that the
deposition rate (in the unit of g/m 2/s) profiles are maximum deposition is much closer to the jet nozzle and
obtained by summing all particles in a region bounded the bottom deposition profile decays much faster,
by two concentric circles with separation r = 0.05-0.1 m, because of the higher entrainment velocity induced by
and divided by the area and experiment time Texp. Particle plumes (entrainment coefficient g = 0.057 for a pure jet
tracking calculations have been performed using a time- vs 0.088 for a pure plume). This is consistent with the
step of 0.5-1ms which is much less than the typical time conclusion of Ernst et al. (1996) that the decay of
of turbulence of jets and plumes. The model predictions deposition rate in radial direction follows r-1 for jet and r-
are validated using a number of reported laboratory data 1/3
for plume.
in previous studies.
The theoretical model by Ernst et al. (1996) accounts for
4. RESULT AND DISCUSSIONS the reentrainment of sediment fallen out from the jet
margin by an empirically determined reentrainment
4.1 Sediment bottom deposition coefficient. This theoretical treatment oversimplified the
complex reentrainment mechanism and the interaction
4.1.1 Sediment fall-out from jet boundary between the turbulent eddies and the ambient jet-induced
entrainment flow. Thus different values of reentrainment

6
coefficient are required for different cases, determined ws is estimated using Hallermeier (1981)'s formula, same
based on fitting of experimental results. Unlike Ernsts as that in Sparks et al. (1991).
model, our model does not require any empirical
The experiments of Zarrebini and Cardoso (2000) were
adjustment for the reentrainment as the mechanism is
carried out in a tank of 0.75m x 0.75m x 0.3m depth with
already embedded in the equation of particle motion [1].
a jet nozzle diameter of 7mm. Freshwater jets with
various flow rates are directed to salt water in the tank to
produce different buoyancy flux. The densimetric Froude
number ranges from 4.0 to 10.5. Spherical Ballotini
particles with a density of 2.47 g/cm 3 are used with
sediment concentration 5-6 g/L (0.2% by volume).
Sediment sizes ranges from 49m to 81m. The
characteristics of the one experiment used in the present
study are shown in Table 2.

Table 2. Vertical buoyant jet experiments (fall out from surface


current) of Sparks et al. (1991) (D = 8mm) and Zarrebini and
Cardoso (ZC) (2000) (D = 7mm).
Jet Vel. Sediment Set. Vel.
Case Fr Source
(m/s) dia. (um) (mm/s)
83-2 0.0215 0.53 58 4.0
Figure 7. Comparison of predicted and measured radial sediment Sparks et
810-2 0.0215 0.53 67 5.4
deposition profile of the four vertical sediment-laden jet cases of al. (1991)
86-1 0.0215 0.53 96 11.1
Ernst et al. (1996). F/Fmax is the radial deposition (g/m2/s)
13 0.3513 10 65 3.3 ZC (2000)
normalized against the maximum predicted/measured values.

The comparison between predicted and measured


deposition profiles are shown in Fig. 9. The deposition is
a cone-like shape with a maximum close to the jet nozzle.
Compared with the fall out from jet margin, the sediment
fall out from the spreading current produces a much
larger region of deposition, extent to over 0.1-0.4m,
depending on the particle size and the height of jet
impingement. Sparks et al. (1991) derived that the decay
of the deposition follows the r-2 law, representing a much
larger deposition extent than the r-1 law for jet and r-1/3
law for plume. The model predicted deposition agrees
satisfactorily with the data under a wide range of particle
size (57-96 m). A visualization of the particle field of
Case 83-2 is shown in Fig. 1b.

Figure 8. Comparison of predicted and measured radial sediment 4.2 Sediment concentration
deposition profile of the four vertical sediment-laden plume cases
of Ernst et al. (1996). F/Fmax is the radial deposition (g/m2/s) The developed model is applied to study the sediment
normalized against the maximum predicted/measured values. concentration in a vertical sediment-laden buoyant jet.
Sediment concentration is determined from the average
number of particles inside a concentric control volume of
4.1.2 Sediment fall-out from spreading current
V = 2rrz, where z = 2.5D, r = 0.1bT and bT is the
The experiments of Sparks et al. (1991) and Zarrebini and top-hat half width of the jet.
Cardoso (2000) represent the sediment fall-out from
spreading current. The experiments were carried out in a
tank of size 1.2m x 1.2m x 0.5m depth with a jet nozzle
diameter of 8mm. Freshwater jets with u0 = 0.0215 m/s
(Q0 = 1.08 cm 3/s) are directed to salt water with density
1021 kg/m 3 in the tank. The densimetric Froude number
is about 0.5, representing a pure plume. The present
model independently computes the potential core
development and accounts for the contraction and
velocity acceleration due to buoyancy (Lee and Chu,
2003). Non-spherical silicon carbide particles with a
density of 3.21 g/cm 3 are used as sediment seedings with
initial concentration 10 g/L (0.3% by volume). Sediment
diameter ranges from 28m to 131m. Settling velocity

7
Figure 11. Cross sectional sediment concentration for a vertical
upward sediment jet at z = 40D. Sediment concentration is
normalized by that of maximum tracer concentration in a simple
jet. Case 710: u0 = 0.75m/s, Fr = 32, ws = 5.3 cm/s.
Figure 9. Comparison of predicted and measured radial sediment
deposition profile of the vertical sediment-laden jet cases of
5. CONCLUSIONS
Sparks et al. (1991), and Zarrebini and Cardoso (ZC, 2000). F/Fmax
is the radial deposition (g/m2/s) normalized against the A 3D stochastic particle tracking model for the prediction
maximum predicted/measured values. of sediment deposition from vertical upward buoyant jets
in stagnant water is developed. The model is unique in
Fig. 10 shows the centerline sediment concentration for a solving the full governing equation of particle motion,
vertical sediment jet with dominant fall-out from the jet using simple semi-analytical models to predict the three
boundary. The centerline sediment concentration is flow regimes of a buoyant jet impinging a free surface -
significant higher than that of a simple jet, due to the and the generation of turbulent fluctuation using an
reenentrainment and recycling of particles fallen out from autocorrelation function that realistically simulates the
the jet margin back to the jet. Fig. 11 shows the cross- trapping of particles in turbulent eddies. The model is
sectional sediment concentration (normalized by the validated against previously reported experimental data
centerline concentration of a simple jet) for the same jet. It for vertical upward buoyant jets in stagnant water and
is predicted that the maximum sediment concentration is predicts well the two regimes of sediment fall out: from
about 70-80% higher than that of the simple pure jet, with jet boundary and from spreading current. Model
prediction reveals the increase in jet sediment
increased jet width. The prediction is comparable to the
concentration caused by the re-entrainment process.
measurement by Carey et al. (1988) which shows similar
effect for a plume with sediment fall out from surface
ACKNOWLEDGMENTS
current. The increase in sediment concentration may
reduce the buoyancy of the plume and cause the This research in supported by a grant from the Hong
instability and collapse of the stable plume structure. Kong Research Grants Council (RGC HKU719408) and in
Veitch and Woods (2000) found that owing to the part by a grant from the University Grants Committee of
recycling of particles, the plume becomes unstable if the Hong Kong (Project No. AoE/P-04/04) to the Area of
initial particle loading exceeds a critical value of about Excellence (AoE) in Marine Environment Research and
Innovative Technology (MERIT).
0.1% by volume. The instability leads to the collapse of
the sediment plume and forms a density current as a
REFERENCES
mixture of fluid and particles close to the ground. An
example is the collapse of volcanic plumes, forming the Akar, P.J. and Jirka, G.H., (1994). Buoyant spreading
pyroclastic flow which is much more devastating than processes in pollutant transport and mixing. Part 1:
the volcanic ash itself. Lateral spreading with ambient current advection.
Journal of Hydraulic Research, 32(6), 815-831.
Cardoso, S.S.S. and Zarrebini, M., (2002). Sedimentation
from surface currents generated by particle-laden jets.
Chemical Engineering Science, 57(8), 14251437.
Carey, S.N., Sigurdsson, H., and Sparks, R.S.J. (1988).
Experimental studies of particle-laden plumes. Journal
of Geophysical Research, 93(B12), 15314-15328.
Chan, S.N. (2013a) Mixing and deposition of sediment-laden
buoyant jets. PhD thesis. The University of Hong
Kong.
Chan, S.N. (2013b). Modelling the Bottom Deposition of
Sediment-Laden Buoyant Jets. Proceedings of 2013
IAHR World Congress, Chengdu, China, 8-13 Sep.
Chan, S.N., Lee, K.W.Y. and Lee, J.H.W. (2014).
Figure 10. Centerline maximum sediment/tracer concentration Numerical modelling of horizontal sediment-laden
(Cmax/C0) for a vertically upward sediment buoyant jet and a jets. Environmental Fluid Mechanics, 14, 173200.
simple jet. Case 710: u0 = 0.75m/s, Fr = 32, ws = 5.3 cm/s.

8
Clift, R., Grace, J.R. & Weber, M.E., (1978) Bubbles, Drops
and Particles, Dover Publications.
Cuthbertson A.J.S. and Davies P.A., (2008). Deposition
from particle-laden, round, turbulent, horizontal,
buoyant jets in stationary and coflowing receiving
fluids. Journal of Hydraulic Engineering ASCE, 134(4),
390-402.
Dissanayake, A.L., Yapa, P.D. and Nakata, K. (2014).
Modelling of hydrothermal vent plumes to assess the
mineral particle distribution. Journal of Hydraulic
Research, 52(1), pp. 4966.
Ernst, G.G.J., Sparks, R.S.J., Carey, S.N. and Bursik, M.I.,
(1996). Sedimentation from turbulent jets and plumes.
Journal of Geophysical Research B: Solid Earth, 101(3),
5575-5589.
Fox, D.G., (1970). Forced plume in a stratified fluid.
Journal of Geophysical Research, 75, 6818-6835.
Hallermeier, R.J., (1981). Terminal settling velocity of
commonly occurring sand grains. Sedimentology, 28(6),
859865.
Lai, A.C.H., (2009). Mixing of a Rosette Buoyant Jet Group.
Ph.D. thesis, The University of Hong Kong.
Lai, A.C.H. and Lee, J.H.W., (2012a). Dynamic interaction
of multiple buoyant jets. Journal of Fluid Mechanics,
708, 539-575.
Lai, C.C.K. and Lee, J.H.W. (2012b). Mixing of inclined
dense jets in stationary ambient. Journal of Hydro-
environment Research, 6(1), 9-28.
Lamb, H. (1932). Hydrodynamics. Cambridge University
Press.
Lee, J.H.W. and Cheung, V., (1990). Generalized
lagrangian model for buoyant jets in current. Journal
of Environmental Engineering, ASCE, 116(6), 1085-1106.
Lee, J.H.W. and Chu V.H., (2003). Turbulent Jets and
Plumes: A Lagrangian Approach. Kluwer Academic
Publishers.
Lee, J. H. W. and Jirka, G.H. (1981). Vertical round
buoyant jet in Shallow Water. Journal of Hydraulic
Engineering, ASCE, 107(HY12), 16511675.
Neves, M.J. and Fernando, H.J.S. (1995). Sedimentation of
particles from jets discharged by ocean outfalls: a
theoretical and laboratory study. Water Science and
Technology, 32(2), 133-139.
Nielsen, P., (1992). Coastal Bottom Boundary Layers and
Sediment Transport. Advanced Series on Ocean
Engineering - Vol.4, World Scientific.
Papanicolaou, P.N. and List, E.J., (1988). Investigations of
round vertical turbulent buoyant jets. Journal of Fluid
Mechanics, 195, 341391.
Shih, T.H., Liou, W.W., Shabbir, A., Yang, Z. and Zhu, J.,
(1995). A new k- eddy viscosity model for high
Reynolds number turbulent flows. Computers and
Fluids, 24(3), 227-238.
Sparks, R. S. J., Carey, S. N., and Sigurdsson, H. (1991).
Sedimentation from gravity currents generated by
turbulent plumes. Sedimentology, 38(5), 839856.
Veitch, G. and Woods, A.W., (2000). Particle recycling
and oscillations of volcanic eruption columns. Journal
of Geophysical Research, 105(B2), 2829-2842.
Wang, H. and Law, A.W.K., (2002). Second-order integral
model for a round turbulent buoyant jet. Journal of
Fluid Mechanics, 459, 397428.
Zarrebini, M. and Cardoso, S.S.S., (2000). Patterns of
sedimentation from surface currents generated by
turbulent plumes. AIChE Journal, 46(10), 1947-1956.

Vous aimerez peut-être aussi