You are on page 1of 18

Environ Fluid Mech (2016) 16:503–520

DOI 10.1007/s10652-015-9435-y

ORIGINAL ARTICLE

Velocity measurements in inclined negatively buoyant


jets

A. T. Crowe1 • M. J. Davidson1 • R. I. Nokes1

Received: 14 May 2015 / Accepted: 3 November 2015 / Published online: 19 November 2015
Ó Springer Science+Business Media Dordrecht 2015

Abstract Experiments were performed with a particle tracking velocimetry system to


investigate the behaviour of inclined negatively buoyant jets with source angles of 15°,
30°, 45°, 60°, 65°, 70°, and 75° in stationary ambient conditions. Velocities were measured
in a plane aligned with the central axis of the flow and the experiments were designed such
that the flow did not interact with boundaries in the region were the flow behaviour was
measured. The results of this study complement previous research, which has largely
focused on the mean geometric characteristics and the mean dilution of the discharged
fluid. Geometric characteristics, spreading rates, and time-averaged (mean) centreline
velocity results are compared with relevant experimental results from previous studies and
integral model predictions. Axial and transverse mean velocity profiles at maximum height
and the return point provide additional insights into the detrainment of discharged fluid due
to the unstable density gradient on the inner side of the flow.

Keywords Particle tracking velocimetry  Dense  Jet  Desalination  Experimental


velocity data

1 Introduction

The depletion of natural potable water sources combined with increased water consump-
tion has resulted in water demand deficits for an increasing number of communities
worldwide. The quantity and quality of available natural potable water sources has
degraded due to overuse, pollution, or salinization [1]. Desalination involves desalting
saline water such as brackish water or seawater to produce potable water. Large-scale

& M. J. Davidson
mark.davidson@canterbury.ac.nz
1
Department of Civil and Natural Resources Engineering, University of Canterbury, Canterbury,
New Zealand.

123
504 Environ Fluid Mech (2016) 16:503–520

desalination plants are the preferred option for meeting water demand deficits in many
countries such as Saudi Arabia, United Arab Emirates, Oman, Qatar, Israel, USA, Spain,
and Australia [2].
The effluent from reverse osmosis desalination plants is a hypersaline brine that is
typically 1.5–2 times the salinity of the ambient seawater [3]. The higher density of the
brine discharge causes it to sink to the sea floor and it has the potential to damage marine
fauna and flora near the disposal site [4]. Benthic ecosystems, such as seagrass meadows or
macroalgae stands and related species, are most susceptible to brine discharges as they
have limited movement [1]. Marine organisms that are unable to regulate their osmotic
pressure are particularly sensitive to changes in salinity. The increased salinity causes
water to leave the cells of these organisms resulting in cell dehydration and possibly death
[3]. The effluent brine discharge also contains chemicals used in the pretreatment of intake
feedwater such as coagulants, antiscalants, or disinfectants [5–8]. However, the majority of
chemicals are considered non-toxic to marine organisms or are discharged at non-critical
concentrations [9].
The most common effluent disposal option in countries with discharge point effluent
regulations is disposal via an offshore pipeline and diffuser system. It is required that the
effluent mix rapidly with the ambient to ensure salinity levels drop below levels specified
by regulatory authorities within the mixing zone [10, 11]. The brine discharge is inclined
upwards so that it initially rises towards the surface of the ambient fluid as an inclined
negatively buoyant jet (INBJ). The negative buoyancy of the brine reduces the vertical
momentum flux of the flow until it reaches a maximum height in the water column. The
flow then falls until it passes the original source height at the return point and impinges the
seabed (lower boundary). A gravity current type flow develops as the discharged fluid
spreads across the seabed [12].
Zeitoun [13] carried out the first significant study into INBJs. Concentrations and flow
trajectories were measured for discharges released at 30°, 45°, and 60° to the horizontal.
While the concentration measurements were difficult to interpret, the source angle of 60°
produced the maximum discharge trajectory and this implied it would also produce the
maximum dilution. Subsequent studies by Roberts and Toms [14] and Roberts et al. [12]
made use of this conclusion in defining the experimental configuration for their investi-
gations of geometric parameters and dilution. Lane-Serff et al. [15] conducted shadow-
graph experiments for source angles between 15° and 75° to determine the maximum
height of the outside edge of the flow. Lane-Serff et al. observed asymmetry of flow
behaviour between the outer side that was well defined and the inner side that had no
distinct edge due to the unstable density gradient. Lindberg [16] and Bloomfield and Kerr
[17] also used shadowgraph techniques to determine geometric parameters of discharges
with source angles between 30°and 90°. Cipollina et al. [18] presented dilution and geo-
metric results from dye tracer flow visualization experiments for the source angles of 30°,
45°, and 60°. Changes in fluid viscosity were found to have no significant effect on the
flow, which indicates that the behaviour of the INBJs studied was independent of Reynolds
number. Nemlioglu and Roberts [19] conducted experiments at source angles between 15°
and 90° for a similar Froude number using three-dimensional laser induced fluorescence
(LIF) to obtain geometric and dilution results.
Physical experiments were conducted by Kikkert et al. [20] to verify an analytical
predictive model. Source angles between 0° and 75° were investigated using the light
attenuation (LA) and LIF flow visualization techniques. Measured concentration profiles
from these experiments demonstrated an asymmetry at the maximum height due to
detrainment on the inner side of the flow, where the density gradient is unstable. Ferrari

123
Environ Fluid Mech (2016) 16:503–520 505

et al. [21] used slide projectors to illuminate fluorescent dye to measure geometric
parameters and dilution for INBJs. Kelvin–Helmholtz instabilities were observed to form
just beyond the source for experiments with a Reynolds number (Re) [ 500. Ferrari et al.
also found that concentration fluctuations were distributed over a wider area and smaller on
the inner side of the discharge due to the detrainment process. Additionally, it was noted
that significant interaction occurred between rising and falling sides of INBJs for angles
above 75° due to the Coanda effect.
Shao and Law [22] investigated the influence of the lower boundary on the initial jet
region of INBJs at source angles of 30° and 45° using a combined particle image
velocimetry (PIV) and laser induced fluorescence (LIF) experimental system to simulta-
neously measure the velocity and concentration fields. Papakonstantis et al. [23, 24]
published results for geometric parameters and dilution from an experimental investigation
across two companion papers for six source angles between 45° and 90°. Lai and Lee [25]
investigated INBJs discharged into a stationary ambient at source angles between 15° and
60°. LIF was used to obtain dilution and geometric parameters, while velocity data was
obtained using PIV. Oliver et al. [26] investigated the behaviour of INBJs where influences
of the lower boundary at the return point had been removed. It was proposed that the
varying lower boundary conditions used in previous studies could, in part, explain some of
the considerable variation between the measured dilution values at the return point. LIF
experiments were conducted for source angles between 15° and 75°. Mean and temporal
dilution profiles were presented along with geometric parameters.
Predictive integral models, such as CORJET [10] and VISJET [27], are essential tools
for diffuser design as they provide an inexpensive method of simulating mean flow
behaviour for a range of alternative design parameters. The CORJET and VISJET models
were initially developed for positively buoyant jets and verification is lacking for INBJs in
stationary and moving ambient conditions [11]. In previous INBJ research, large dis-
crepancies between model predictions and experimental results have been noted [28, 29].
INBJs have additional flow features not seen for typical positively buoyant discharge
configurations, including detrainment of discharged fluid due to the unstable density
gradients [15, 20, 23, 24, 26] and re-entrainment for high source angles [17, 19, 21].
Additionally, these predictive models do not consider the influences of lower boundary
interactions [11, 30] making comparisons with the majority of experimental studies where
the lower boundary influences the observed flow behaviour problematic.
The idealised discharge configuration for the problem under consideration is shown in
Fig. 1. The negatively buoyant brine is initially discharged upwards at a source angle, ho,
through a circular pipe with an initial diameter, d, and initial discharged fluid velocity, Uo.
The important scaling parameter for these discharges is the Froude number, Fo, (Eq. 1).
The initial reduced gravity, ĝo, relates the initial discharge density, qo, with the ambient
fluid density, qa, and gravity, g (Eq. 2).
Uo
Fo ¼ pffiffiffiffiffiffiffi ð1Þ
g^o d

ðqo  qa Þ
g^o ¼ g ð2Þ
qa

Here, the centreline of the flow is defined as the physical location of local maximum
velocity, uc, for cross-sectional profiles through the flow. The flow rises through the water
column until it reaches centreline maximum height xm, zm before descending to the

123
506 Environ Fluid Mech (2016) 16:503–520

Fig. 1 Important parameters of INBJ trajectories depicted in this diagram include: horizontal and vertical
distance to maximum centreline height xm, zm, vertical distance to the outside edge zme, horizontal distance
to return point and to outside edge xr, xre, centreline path length to maximum height and return point sm, sr,
mean centreline velocity at maximum height and return point um, ur

elevation of the source, where the coordinates of the centreline define the location of the
return point xr, 0. The inside and outside edges of the flow define its width, b, where the
flow ‘‘edge’’ is defined by the location of the component of velocity in the direction of the
cross-sectional maximum velocity that is equal to e-1uc.
Geometric parameters, non-dimensionalised by d, and dilution have been found to be
directly proportional to Froude number [12, 14, 18, 20, 22, 26]. The k-notation for
experimental coefficients, dependent on the initial source angle, used in previous studies
and here is shown in Eqs. 3, 4, 5.
xm
¼ kxm ðho Þ ð3Þ
Fo d
zm
¼ kzm ðho Þ ð4Þ
Fo d
xr
¼ kxr ðho Þ ð5Þ
Fo d

Dimensional relationships for mean centreline velocity at maximum height, um, and the
return point, ur, can be derived in a similar way (Eqs. 6, 7), where the relevant coefficients
are kum and kur respectively. Note these equations represent the inverse of a Froude number
based on the local velocity at the maximum height and the return point of the flow.
Uo
¼ kum ðho Þ ð6Þ
Fo um
Uo
¼ kur ðho Þ ð7Þ
Fo ur

Complementing earlier studies, in particular the study of Oliver et al. [26], the present
investigation provides high quality velocity field information for INBJs where the lower

123
Environ Fluid Mech (2016) 16:503–520 507

boundary has no influence on the measured flow behaviour. This approach provides a more
complete picture of the flow behaviour and it enables integral model predictions, prior to
interaction with the lower boundary, to be assessed more effectively. Velocities are
measured using the Particle Tracking Velocimetry (PTV) flow visualization technique. The
PTV system involves seeding the discharge with tracer particles that are illuminated with a
laser, thus allowing the velocity of the particles to be determined from images captured
with a video camera. Geometric parameters were determined from velocity field data,
which differs from the majority of previous studies where geometric parameters have been
determined from concentration-based experiments. Velocity results are presented and
compared to predictive models and the new data provides additional insights into the
unique features of INBJs.

2 Experimental setup

A schematic of the experimental tank configuration is shown in Fig. 2. The glass exper-
imental tank is 2.30 m wide, 1.23 m deep, and 1.78 m high. Three sources with internal
diameters, d, of 2.43, 4.40, and 7.19 mm were used in the experiments to provide a wide
range of Froude numbers from 10.1 to 81.0 (Table 1). The source diameter was constant
for a minimum of 24 diameters before the outlet. Discharge Reynolds numbers ranged
from 2700 to 6100 resulting in turbulent flow at the outlet. The source was elevated at least
655 mm above the lower boundary of the experimental tank, which was a similar height
employed by [26]. The source angle was set to an accuracy of ±1 with an inclinometer and
the initial angle was progressively adjusted from 15° to 75° in this study.
The density of the discharge fluid was approximately 3 % greater than that of the
ambient fluid and was created by dissolving salt (NaCl) in tap water filtered to 5 lm.
Ambient and discharge fluid densities were measured using an Anton Paar DMA 5000
density meter. The discharge fluid header system shown in Fig. 2 supplied dense fluid to
the source. The system provided a constant head of at least 2.7 m above the exit of the
source to ensure flow was constant for each experiment. The pressurized tank inside the

Fig. 2 Schematic diagram of experimental system

123
508 Environ Fluid Mech (2016) 16:503–520

Table 1 Initial conditions of experiments in present study


ho Number of d (mm) Qo (L/min) Fo
experiments

15° 8 2.43, 4.40, 7.19 0.46–2.01 10.1–80.4


30° 8 2.43, 4.40, 7.19 0.45–1.99 10.3–81.0
45° 8 2.43, 4.40, 7.19 0.45–2.03 10.2–78.8
60° 9 2.43, 4.40, 7.19 0.37–2.11 10.4–80.2
65° 4 2.43, 4.40 0.44–1.29 30.0–80.4
70° 3 2.43, 4.40 0.58–1.29 30.8–78.4
75° 5 2.43, 4.40, 7.19 0.38–1.98 10.1–50.7

experimental tank kept the temperature of discharge fluid at a similar temperature to the
ambient fluid. Two identical Krohne IFC 010D electromagnetic flow meters were used to
measure the flow rates in the system. The flow to the constant head tank was 50 % greater
than the flow out of the source, which ensured that fluid overflowed the constant head tank
throughout each experiment. A calibrated data logger was used to record the source flow
rate.
The discharge fluid and the ambient fluid were seeded with Pliolite resin tracer particles
that had been sifted to diameters between 125 and 180 lm. The tracer particles were
illuminated by a 2 W Spectra-Physics Millennia II laser with a wavelength of 532 nm and
mirror system that produced a light sheet that was orientated along the central plane of the
flow. The laser light beam was directed at an 8-sided glass scanning mirror, rotating at
16,000 RPM, which scanned the beam across a parabolic mirror to produce a vertical laser
sheet approximately 5 mm thick (Fig. 2).
The motion of illuminated particles was recorded through the front glass wall of the
tank, which had maximum viewable dimensions of 1.25 m by 1.25 m. The camera was
placed at a perpendicular distance of 3.04 or 4.26 m from the laser sheet, depending on
the scale of the flow. A JAI Pulnix TM-2030CL single CCD camera, with a GoYo 100
50 mm f/0.95 lens, created grayscale images of the flow at 32 Hz. The typical duration
for an experiment was 5 min. Raw images from the camera were transferred to the
computer via a CameraLink interface and written to a high speed hard drive. Raw
images were converted to lossless bitmap images before being analysed with the Streams
software [31]. There were no detectable barrel or pincushion effects in the images
captured. The camera was located such that parallax was insignificant for these
experiments.
The creation of velocity fields from the captured images was a three-stage process. In
the first stage, the precise location of tracer particles within each image was found by
examining pixel intensities. If a pixel had sufficiently high intensity, relative to the
background pixel intensity, it was considered part of an illuminated tracer particle. Typ-
ically, 6000–8000 tracer particles were identified in each image. The second stage matched
particles between consecutive images, which is an ‘‘assignment’’ problem. Streams soft-
ware utilises an auctioning algorithm to match tracer particles between consecutive pairs of
images. Costing algorithms assign a cost to each possible match of tracer particles, with the
goal of minimizing the overall cost of matches.
A search window limited the number of possible matches to reduce the computation
time of the matching process. Two correlation-costing strategies with different correlation

123
Environ Fluid Mech (2016) 16:503–520 509

window sizes were used to match particles initially. Additional costing algorithms,
including local velocity, recent velocity, time average velocity, and space average accel-
eration [31], made use of previously successful matches to create new particles matches
between images. Each costing strategy was iterated through all images five times in for-
ward and reverse directions to maximize the number of particle matches [32].
The velocity fields were created using an Eulerian specification of the two-dimen-
sional flow field, where the motion of the fluid was evaluated at fixed locations on a
regular grid over time. The instantaneous velocity of individually matched particles was
determined using a central difference approximation from the positions of each tracer
particle in the previous and following images. If the position of the particle was known
only in the preceding and subsequent image, a backward or forward difference
approximation was used. Binning and triangulation were utilized for interpolation onto
the regular grid. The selection of bin size and triangle limits influenced the velocity
field created for the flow field. It was necessary to balance the accuracy of measured
velocity at grid points with the coverage of the flow field in the time domain. Further
information about the PTV costing algorithm and analysis method can be found in
Crowe [33].
Eight fully turbulent pure jet experiments were conducted to assess the accuracy of
the PTV system adopted for this study [33]. Pure jets have been studied extensively and
there are well-defined experimental coefficients for velocity decay and spread [34–38].
Velocity and spread coefficients for the PTV system were consistent with the values
from previous pure jet studies. In addition, the turbulent intensity profiles were self-
similar and of similar form to those that had been measured previously. Centreline axial
and radial turbulent intensities were approximately 8 % lower than the mean of previous
studies due to the PTV system’s limited ability to capture the full range of eddy scales.
This issue is discussed further in Crowe [33], where discrepancies associated with the
mean velocity measurements are also discussed and noted to be less than 2 % of the
measured value.

Fig. 3 Contours of mean absolute velocity for one 45° experiment: Fo = 78.8, Uo = 2110 mm/s, source
coorindates are xo/d = 39.3, zo/d = 76.3

123
510 Environ Fluid Mech (2016) 16:503–520

3 Results

3.1 Flow characteristics

Mean velocity contours for an INBJ with a source angle of 45° are shown in Fig. 3,
superimposed with the discharge centreline determined from a curve fit through the
locations of local maximum velocity. The difference between flow behaviour on the outer
and inner side of the centreline is clearly illustrated by the narrowly spaced contour lines
on the outer side near maximum height compared to the broadly spaced contour lines on
the inner side in the same region. This difference is due to the stable density gradient on the
outer side and the unstable density gradient on the inner side of flow. Detrainment was
observed during the physical experiments on the inner side of flow, which is consistent
with previous observations [15, 26]. The amount and direction of detrained fluid was
dependent on the source angle and path length from the source. Less fluid was observed to
detrain from the flow near maximum height for lower source angles, presumably because
of differences in the relative sizes of momentum and buoyancy fluxes. The distinct for-
mation of eddies of varying size at irregular time intervals was clearly evident on the outer
side of the flow in the maximum height region and beyond. In contrast, on the inner side of
the flow individual eddies were difficult to identify and mixing appeared to be dominated
by relatively small-scale motions. A continuous progression of detrained fluid away from
the main flow was also evident in this region. Re-entrainment of the falling plume by the
rising arm of the flow occurred near the source for the higher source angles of 70° and 75°,
where the horizontal separation between rising and falling arms was small, which is again
consistent with previous observations.

3.2 Geometric parameters

Geometric parameters are important in the design of desalination discharges to determine


the flow location in the context of the relevant regulatory mixing zone. As expected
geometric parameters determined from experiments in this study, non-dimensionalised by
source diameter, d, were found to be linearly proportional to Froude number for all source
angles and locations [33]. Experimental coefficients for geometric parameters were found
by applying a linear regression through data at each source angle (Table 2). The coeffi-
cients for the horizontal and vertical distance to maximum height kxm and kzm respectively
from the present study are in agreement with previous studies for all source angles (Figs. 4,
5) [21]. The scatter of kxm coefficients is more significant for all studies when compared to
kzm coefficients across all source angles. The vertical distance to maximum height is

Table 2 Geometric experimen-


ho kxm kzm kxr
tal coefficients with standard
deviations found from linear
regressions. Further experimental 15° 1.45 ± 0.04 0.25 ± 0.01 2.51 ± 0.06
coefficients for other geometric 30° 1.87 ± 0.03 0.69 ± 0.01 3.56 ± 0.05
locations can be found in Crowe 45° 1.96 ± 0.07 1.22 ± 0.04 3.43 ± 0.09
[33]
60° 1.69 ± 0.04 1.71 ± 0.03 2.93 ± 0.06
65° 1.50 ± 0.03 1.82 ± 0.01 2.53 ± 0.06
70° 1.34 ± 0.02 1.96 ± 0.01 2.28 ± 0.03
75° 1.10 ± 0.04 2.02 ± 0.04 1.87 ± 0.06

123
Environ Fluid Mech (2016) 16:503–520 511

Fig. 4 Experimental coefficients for the horizontal location of maximum height kxm for all source angles

Fig. 5 Experimental coefficients for the vertical location of maximum height kzm for all source angles

relatively simple to define accurately when compared to the horizontal location due to the
shallow curvature of the centreline near maximum height. Predictions by the integral
models, CORJET and VISJET, follow the general trends across the range of source angles.

123
512 Environ Fluid Mech (2016) 16:503–520

Fig. 6 Experimental coefficients for horizontal location of the return point kxr for all source angles

However, experimental geometric coefficients are underestimated with model predictions


located outside the scatter of data.
Experimental coefficients for the horizontal distance to the centreline return point kxr
from the present study are also in agreement with previous studies (Fig. 6). It is worth
noting that there is a variation in the physical location of the experimental coefficients
reported in the return point region. The location where the discharge impacts a lower
boundary is often used [12, 19] or return point values are reported with [22, 24] or without
[26] a lower boundary in close proximity to the return point measurements. These dif-
ferences are not accounted for in Fig. 6 and this will, in part, account for some of the
variation between studies as the influence of lower boundary on INBJ behaviour is not well
defined. CORJET and VISJET predictions again follow the trend of experimental coeffi-
cients, but underestimate coefficients for all source angles.

3.3 Spread

Discharge width was determined independently on the inner and outer sides due to the
asymmetrical nature of INBJs. The spread values then provide a consistent definition of the
inner and outer edges of the flow. The spreading rate is the gradient of discharge width to
path length. The constant spreading rates of radially symmetric pure jets and vertical
buoyant jets have been experimentally measured in numerous studies [38]. The spreading
rates of concentration measurements are greater than those for velocity [34]. Previously
measured velocity spread rates for jets include 0.106 [34], 0.104 [35], 0.115 [36], 0.113
[37], 0.111 [33] and in the plume region of vertical buoyant jets include 0.105 [34], 0.105
[35], 0.131 [39].

123
Environ Fluid Mech (2016) 16:503–520 513

For a given discharge angle, the non-dimensionalised discharge spread measurements


collapse with non-dimensionalised path length on both the inner and outer sides of the flow
(Figs. 7, 8, 9). Discharge spread on the outer side of INBJs are linearly dependent on the
path length. The outer spreading rates for the present study were determined using linear
regression, which yielded values of 0.114, 0.117, and 0.114 for the 30°, 45°, and 60° source
angles respectively. These values indicate that outer spreading rate of INBJs is not
dependent on source angle and that spreading rates are similar to previous measured values
for jets and plumes noted above. In [26] it was noted that concentration width increased
sporadically with a non-linear spreading rate for some experiments near the return point;
however this was not evident in the velocity spread measurements presented here.
Discharge spread has a non-linear relationship with path length on the inner side of
INBJs (Figs. 7, 8, 9), with significant dependence on source angle. Discharge width on the
inner side before maximum height is similar to that for the outer side of 30° discharges
(Fig. 7). Subsequently, discharge width on the inner side increases due to the unsta-
ble density gradient driving detrainment, which creates elongated concentration profiles on
the inner side of the flow near maximum height [40]. Lai and Lee [25] reported that the
ratio between the discharge width on the inner and outer sides was approximately one near
the source before increasing to about two at maximum height for 30° discharges, which is
consistent with the present study. The inner spreading rate reduces near the return point
before becoming constant (Fig. 7). The spreading rate on the inner side of the flow was
found to be 0.249 for s/Fod [ 3.5, which is much greater than the spreading rate on the
outer side 0.114. In contrast Lai and Lee noted that the ratio between discharge width on
the inner and outer sides remained constant after maximum height for 30° discharges. The
difference in observed behaviour is possibly the result of the close proximity of the lower
boundary in the experiments of Lai and Lee, leading to a reduction in discharge width on
the inner side of the flow.

Fig. 7 Non-dimensionalised discharge width against non-dimensionalised path length for all 30°
experiments. Unfilled symbols: outer side, filled symbols: inner side

123
514 Environ Fluid Mech (2016) 16:503–520

Fig. 8 Non-dimensionalised discharge width against non-dimensionalised path length for all 45°
experiments. Unfilled symbols: outer side, filled symbols: inner side

Fig. 9 Non-dimensionalised discharge width against non-dimensionalised path length for all 60°
experiments. Unfilled symbols: outer side, filled symbols: inner side

The increase in discharge width on the inner side occurs more rapidly for 45° and 60°
discharges and the reduction in spreading rate to a constant value occurs closer to maxi-
mum height (Figs. 8, 9). The inner spreading rate values after maximum height for 45° and

123
Environ Fluid Mech (2016) 16:503–520 515

60° discharges are 0.255 for s/Fod [ 3.5 and 0.108 for s/Fod [ 3.2, respectively. The inner
spread rate constant for 60° discharges is very similar to the outer spread rate constant of
0.114, which indicates the falling flow behaviour is similar to a plume and hence
detrainment is no longer having a significant influence on the flow.

3.4 Velocity results

The inverse of the centreline velocity (in non-dimensional form) at maximum height and
the return point is linearly dependent on Froude number for all source angles (Figs. 10, 11),
confirming the relationships derived in Eqs. 6, 7. Experimental coefficients for centreline
velocity were determined by applying a linear regression to the data for each source angle
(Table 3). The inverse of the kum values, a Froude number based on local maximum height
velocity, decrease with increasing source angle at maximum height (Fig. 12). This is
consistent with the momentum flux at maximum height being equal to the initial horizontal
momentum flux. The velocity angles at maximum height (hm) are also listed in Table 3.
These values represent an average across all Froude numbers for a given discharge angle
and are expected to be approximately 0. The location of the maximum height was defined
to be where the centreline of the flow reaches the highest vertical geometric distance above
the source. While the magnitude of the centreline velocity was relatively insensitive to this
location, the sensitivity of the velocity angle increased for higher initial discharge angles,
where the curvature associated with the flow reversal was more severe. Thus, the angle
measurements at maximum height should be treated with some caution. There is limited
experimental velocity data available for comparison; however centreline velocities at
maximum height for the source angle of 60° were interpolated and averaged from Lai and

Fig. 10 Centreline velocity at maximum height um, non-dimensionalised by initial source velocity Uo, for
all source angles against Froude number Fo

123
516 Environ Fluid Mech (2016) 16:503–520

Fig. 11 Centreline velocity at the return point ur, non-dimensionalised by initial source velocity Uo, for all
source angles against Froude number Fo

Table 3 Centreline velocity experimental coefficients and local angle at maximum height kum, hm and the
return point kur, hr with standard deviations found from linear regressions
ho kum hm, rad kur hr, rad

15° 0.251 ± 0.015 0.014 ± 0.011 0.404 ± 0.015 -0.399 ± 0.023


30° 0.356 ± 0.010 0.042 ± 0.013 0.581 ± 0.020 -0.745 ± 0.027
45° 0.481 ± 0.014 0.076 ± 0.028 0.693 ± 0.031 -1.1018 ± 0.028
60° 0.645 ± 0.010 0.081 ± 0.064 0.732 ± 0.012 -1.238 ± 0.022
65° 0.704 ± 0.016 0.109 ± 0.029 0.742 ± 0.037 -1.311 ± 0.048
70° 0.768 ± 0.013 0.106 ± 0.017 0.703 ± 0.008 -1.315 ± 0.007
75° 0.872 ± 0.033 0.155 ± 0.044 0.702 ± 0.015 -1.381 ± 0.015

Lee [25]. The experimental coefficient determined from Lai and Lee is located between the
values from the present study and the predictions from CORJET and VISJET. The inverse
kur values (Froude numbers based on local return point velocity) decrease with increasing
source angle, but become reasonably constant at a value of 1.40 between initial discharge
angles of 45 and 75 degrees (Fig. 12). hr values become increasingly negative with
increasing source angle (Table 3), which is consistent the flow approaching a vertical
orientation at the return point as the discharge angle increases. CORJET and VISJET
predictions substantially overestimate inverse kur values at the return point and these
predicted values increase after 30°, whereas the experimental values continue to decrease
(by approximately 20 %). VISJET predictions are more consistent with the experimental
data than those from CORJET at the return point; however the average difference between
experimental coefficients and VISJET predictions across all source angles is 68 %. This is

123
Environ Fluid Mech (2016) 16:503–520 517

Fig. 12 Variation of Froude number based on the local velocity at maximum height 1/kum and the return
point 1/kur with source angle

Fig. 13 Axial (unfilled symbols) and transverse (filled symbols) components of velocity at maximum height
for 60° experiments. Profiles for other source angles can be found in Crowe [33]

123
518 Environ Fluid Mech (2016) 16:503–520

similar to the 50–65 % difference between model predictions and experimental coefficients
for dilution at the return point [29].
Profiles of non-dimensionalised velocity components collapse at maximum height when
plotted against the radial distance from the centreline, r, non-dimensionalised by the outer
discharge width, b (Fig. 13). The transverse velocities were measured in the vertical plane,
perpendicular to the axial velocities. Profiles demonstrate the asymmetric nature of dis-
charges due to the unstable density gradient on the inner side of the flow. Axial velocities
on the outer side negative r, where the density gradient is stable compare well with a
Gaussian distribution, consistent with pure jet and plume discharges. Positive transverse
velocities on the outer side of the flow show the entrainment of ambient fluid, as a local co-
ordinate system referencing the flow centreline was employed when defining the velocity
component profiles. Therefore, the much larger positive transverse velocities on the inner
side of the flow positive r demonstrate the flow is moving away from the centreline
(detrainment) previously observed in dilution studies [19, 25, 29]. Axial velocities are also
substantially distorted on the inner side of the flow, indicating the detraining fluid velocity
has comparable transverse and axial components as it leaves the main flow at maximum
height. Axial velocity profiles on the outer side of the flow at the return point have a
Gaussian distribution and transverse velocities show entrainment is occurring (Fig. 13).
Transverse velocities on the inner side of the flow are of similar magnitude to the outer side
at the return point; however detrainment is still occurring. The decay of axial velocity is
approximately linear with radial distance from the centreline on the inner side of these
discharges at the return point.

4 Conclusions

Velocity measurements from a detailed laboratory study into the behaviour of inclined
negatively buoyant discharges in a stationary ambient have been presented. The experi-
ments were conducted using a PTV system for initial source angles ranging from 15° to
75°. Froude numbers ranged between 10.1 and 81.0. The experimental system was
designed such that the measured flow did not interact with a lower boundary, allowing for
direct comparison with predictions from integral models that do not incorporate lower
boundary interactions.
The experimental results show that the form of the velocity profiles is independent of
the initial Froude number at a given discharge angle and path length. The mean axial
velocity profiles at maximum height and the return point on the outer side of the flow
remain self-similar with a Gaussian form. In contrast the inner side of the flow transitions
from a Gaussian form near the source to linear decay from the centreline at the return point.
The transverse velocities measured on the inner side of the flow at maximum velocity show
the substantial detrainment that occurs, rather than the typical entrainment that is evident
on the outer side of the flow.
The experimental results illustrate the complex flow behaviour of INBJs due to
unstable density gradients on the inner side of these flows, which result in substantial
detrainment from the main flow. The effect of this flow feature is particularly evident in
measurements of discharge spread, defined by e-1uc, between the inner and outer sides of
the flow. The discharge spread on the outer side remains linear for the full path length,
whereas on inner side the discharge spread undergoes a rapid expansion starting before
maximum height and ending before the return point. A linear growth in discharge spread

123
Environ Fluid Mech (2016) 16:503–520 519

then becomes evident; however the growth rate remains elevated for initial discharge
angles of 30 and 45 degrees.
Geometric parameters determined from velocity measurements were found to be con-
sistent with the values determined by previous studies, which generally measured flow
concentrations. Small differences are possibly due to inconsistencies in definitions between
studies. In addition, the presence of a lower boundary in the experimental setup of previous
studies will have some influence on conditions at the return point.
The non-dimensional centreline velocities at maximum height and the return point were
found to be directly proportional to initial Froude number, confirming the relationships
developed using dimensional arguments. The inverse of the kum values (a Froude number
based on the local centreline velocity) reach an approximately constant value of 1.4 at the
return point for initial discharge angles ranging from 45 to 75 degrees. Values of the same
Froude number at maximum height (inverse of kum) decline in a manner that is consistent
with the reductions in the horizontal momentum flux at this location as the initial discharge
angle increases.
Integral model predictions (CORJET and VISJET) of geometric parameters are similar
to experimentally measured values; however centreline velocities are overestimated by
these models. Centreline velocity predictions at maximum height follow the trend evident
in the experimental values across the full range of source angles, but the differences at the
return point are more significant. The distortion of velocity profiles and measured
detraining velocities show that unstable density gradients on the inner side of the flow have
a significant influence on mean flow characteristics. Existing numerical models do not
adequately take account of these influences and it is important to recognise these potential
weaknesses when interpreting predictions from such models.

References
1. Lattemann S, Kennedy MD, Schippers JC, Amy G (2010) Chapter 2 Global Desalination Situation.
Sustain Sci Eng
2. Bleninger T, Jirka GH (2008) Modelling and environmentally sound management of brine discharges
from desalination plants. Desalination 221:585–597
3. Voutchkov N (2011) Overview of seawater concentrate disposal alternatives. Desalination 273:205–219
4. Einav R, Lokiecb F, Lokiec F (2003) Environmental aspects of a desalination plant in Ashkelon.
Desalination 156:79–85
5. Cooley H, Gleick PH, Wolff G (2006) Desalination, with a grain of salt: a California perspective. Stud.
Dev. Environ. Secur, Oakland, CA Pacific Inst
6. Dolnicar S, Schäfer AI (2009) Desalinated versus recycled water: public perceptions and profiles of the
accepters. J Environ Manag 90:888–900
7. Hodgkiess T, Oldfield J, Höpner T (2003) Assessment of the composition of desalination plant disposal
brines. Middle East Desalin. Res, Cent
8. WHO (2007) Desalination for safe water supply: Guidance for the health and environmental aspects
applicable to desalination
9. Bleninger T, Jirka GH, Report P (2009) Environmental planning, prediction and management of brine
discharges from desalination plants
10. Doneker RL, Jirka GH (2001) CORMIX-GI systems for mixing zone analysis of brine wastewater
disposal. Desalination 139:263–274
11. Palomar P, Lara J, Losada I et al (2012) Near field brine discharge modelling part 1: analysis of
commercial tools. Desalination 290:28–42
12. Roberts PJW, Ferrier A, Daviero G (1997) Mixing in inclined dense jets. J Hydraul Eng 123:693–699
13. Zeitoun MA (1970) Conceptual Designs of Outfall Systems for Desalting Plants
14. Roberts PJW, Toms G (1987) Inclined dense jets in flowing current. J Hydraul Eng 113:323–341
15. Lane-Serff GF, Linden PF, Hillel M (1993) Forced, angled plumes. J Hazard Mater 33:75–99

123
520 Environ Fluid Mech (2016) 16:503–520

16. Lindberg WR (1994) Experiments on negatively buoyant jets, with and without cross-flow. NATO ASI
Ser E Appl Sci Study Inst 255:131–146
17. Bloomfield LJ, Kerr RC (2002) Inclined turbulent fountains. J Fluid Mech 451:283–294
18. Cipollina A, Brucato A, Grisafi F, Nicosia S (2005) Bench-scale investigation of inclined dense jets.
J Hydraul Eng 131:1017–1022
19. Nemlioglu S, Roberts PJ (2006) Experiments on dense jets using three-dimensional laser-induced
fluorescence (3D LIF). Int. Conf. Mar. Waster Water Discharges Coast, Environ
20. Kikkert GA, Davidson MJ, Nokes RI (2007) Inclined negatively buoyant discharges. J Hydraul Eng
133:545–554
21. Ferrari S, Querzoli G (2010) Mixing and re-entrainment in a negatively buoyant jet. J Hydraul Res
48:632–640
22. Shao D, Law AWKW (2010) Mixing and boundary interactions of 30° and 45° inclined dense jets.
Environ Fluid Mech 10:521–553
23. Papakonstantis IG, Christodoulou GC, Papanicolaou PN (2011) Inclined negatively buoyant jets 1:
geometrical characteristics. J Hydraul Res 49:3–12
24. Papakonstantis IG, Christodoulou GC, Papanicolaou PN (2011) Inclined negatively buoyant jets 2:
concentration measurements. J Hydraul Res 49:13–22
25. Lai CCKK, Lee JHWW (2012) Mixing of inclined dense jets in stationary ambient. J Hydro Environ
Res 6:9–28
26. Oliver C, Davidson M, Nokes R (2013) Removing the boundary influence on negatively buoyant jets.
Environ Fluid Mech 13:625–648
27. Cheung SKB, Leung DYL, Wang W, et al. (2000) VISJET - a computer ocean outfall modelling system.
Proc. Comput. Graph. Int. Conf. CGI. IEEE Comput. Soc, pp 75–80
28. Palomar P, Lara JL, Losada IJ (2012) Near field brine discharge modeling part 2: validation of com-
mercial tools. Desalination 290:28–42
29. Oliver C, Davidson M, Nokes R (2013) Predicting the near-field mixing of desalination discharges in a
stationary environment. Desalination 309:148–155
30. Loya-Fernández Á, Ferrero-Vicente LM, Marco-Méndez C et al (2012) Comparing four mixing zone
models with brine discharge measurements from a reverse osmosis desalination plant in Spain.
Desalination 286:217–224
31. Nokes R (2012) Streams Version 2.00 - System Theory and Design. University of Canterbury. Tech-
nical manual. Available: http://www.civil.canterbury.ac.nz/streams.shtml
32. Blackett SA (1994) Particle Tracking Velocimetry. University of Auckland. Masters thesis
33. Crowe AT (2013) Inclined negatively buoyant jets and boundary interaction. University of Canterbury.
Doctoral thesis
34. Wang H, Law A (2002) Second-order integral model for a round turbulent buoyant jet. J Fluid Mech
459:397–428
35. Papanicolaou PN, List EJ (1988) Investigations of round vertical turbulent buoyant jets. J Fluid Mech
195:341–391
36. Panchapakesan NR, Lumley JL, Panchapakesan NR (1993) Turbulence measurements in axisymmetric
jets of air and helium. Part 1. Air Jet J Fluid Mech 246:197–223
37. Hussein HJ, Capp SP, George WK (1994) Velocity measurements in a high-Reynolds-number,
momentum-conserving, axisymmetric, turbulent jet. J Fluid Mech 258:31–75
38. Fischer HB, List JE, Koh RC, et al. (1979) Mixing in Inland and Coastal Waters. Academic Press, Inc
39. Shabbir A, George WK (1992) Experiments on a round turbulent buoyant plume. J Fluid Mech
275:1–32
40. Kikkert GA (2006) Buoyant jets with two and three-dimensional trajectories. University of Canterbury.
Doctoral thesis

123

You might also like