J. Fluid Mech. (2017), vol. 833, R2, doi:10.1017/jfm.2017.

762

journals.cambridge.org/rapids
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

On the dynamics of starting plumes

N. Bhamidipati1 and Andrew W. Woods1, †

1 BP Institute, Department of Earth Sciences, University of Cambridge, Madingley Road,
Cambridge CB3 0EZ, UK

(Received 23 August 2017; revised 6 October 2017; accepted 18 October 2017)

We explore the dynamics of starting plumes by analysis of a series of new small-scale

laboratory experiments combined with a theoretical model for mass, momentum, and
buoyancy conservation. We find that the head of the plume ascends with a speed
which is approximately 0.6 times the characteristic speed of the fluid in the following
steady plume, in accord with Turner (J. Fluid Mech., vol. 13 (03), 1962, pp. 356–368),
and so the fluid released from the source eventually catches the head of the flow.
On reaching the top of the plume it recirculates and mixes in the plume head. We
estimate that approximately 0.61 ± 0.04 of the total buoyancy released from the source
accumulates in the plume head, with the remainder in the following steady plume.
Using measurements of the volume of the head, we estimate that a fraction 0.16 ± 0.08
of the volume of the head is entrained directly from the ambient, with the remainder
of the fluid in the head being supplied by the following steady plume. These results
imply that the buoyancy force exerted on the plume head plus the momentum flux
supplied by the following plume exceeds the rate of change of momentum of the
plume head even including the added mass of the plume head. We propose that the
difference is associated with a drag force resulting from the displacement of ambient
fluid around the plume head. Using our experimental data, we estimate that the drag
coefficient Cd has a value 4.2 ± 1.4, with the range in values associated with the
uncertainty in our estimate of entrainment of fluid directly into the plume head. As
a test, the proposed model is shown to provide a reasonable description of a starting
plume rising through a stratified environment in the region below the maximum height
of rise of the associated steady plume, although, above this point, the shape of the
plume head changes and the model breaks down.
Key words: plumes/thermals, turbulent mixing
https://doi.org/10.1017/jfm.2017.762

1. Introduction
The study of turbulent plumes in uniform and stratified ambient has been of
interest for many decades owing to their importance in numerous industrial and

† Email address for correspondence: [email protected]

c Cambridge University Press 2017
833 R2-1
 N. Bhamidipati and A. W. Woods
environmental processes (Morton, Taylor & Turner 1956; Woods 2010). Morton et al.
(1956) showed that the rate of entrainment of ambient fluid at a given height is
proportional to the characteristic plume velocity at that height, and developed a series
of self-similar solutions based on this assumption. These solutions have proved robust
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

to many experimental observations and theoretical analyses. Turner (1962) explored

the transient nature of buoyant plumes with an experimental investigation of the
initial stages of formation of a turbulent buoyant plume, measuring the properties of
the plume head which develops ahead of the steady plume. The key observation of
his work was that the plume head rises at a speed that is approximately 0.6 times the
characteristic velocity of the steady plume which develops behind the head. Turner
(1962), Middleton (1975), and more recently Scase, Aspden & Caulfield (2009) have
proposed theoretical models for the dynamics of the head of the plume building on
the vortex-ring theory for a discrete buoyant thermal proposed by Turner (1957),
although there remains some uncertainty about the internal dynamics of the plume
head (Scase et al. 2009).
There has been some debate about the speed of the plume head relative to the
speed in the steady plume which develops behind the head, and there is little
experimental evidence supporting the model that with highly turbulent flow, the
plume head behaves as a vortex ring. The purpose of this contribution is to reassess
the dynamics of a starting plume with a series of new experiments, exploring the
balance of mass, momentum and buoyancy between the plume and the plume head.
As well as the starting plume, the experiments may provide some new insights into
the class of problems in which the source buoyancy flux of an established plume is
rapidly increased from one value to another. In this case, it is likely that a plume
head type structure, akin to the starting plume, may develop between the original and
new plume (Scase et al. 2006, 2009).
We commence by describing some observations from a series of new laboratory
experiments of starting plumes, which complement previously published data. We then
use conservation laws to constrain the fraction of source buoyancy which accumulates
in the head. We find that the associated buoyancy force plus the momentum flux
supplied by the following plume exceed the rate of change of momentum of the
starting plume. We interpret this imbalance as evidence of a drag force associated with
the displacement of ambient fluid originally ahead of the starting plume. We explore
the implications of our model for the dynamics of a starting plume rising in a stratified
ambient.

2. Experimental observations
A series of six experiments were carried out in a tank of dimensions 0.8 m ×
0.6 m × 0.6 m, initially filled with fresh water. Salt water plumes of source salinity, so ,
varying between 2.5 and 10 wt.% were supplied through a nozzle of diameter 0.001 m
placed at the top of the tank. A peristaltic pump was calibrated to supply plume fluid
to the nozzle at a constant volume flow rate of 1.33 cm3 s−1 . The plumes have a
source Reynolds number of O(103 ), and become fully turbulent within 1–2 cm from
https://doi.org/10.1017/jfm.2017.762

the source. The plume fluid was dyed in order to distinguish it from the fluid in the
tank, and the tank was backlit by an electroluminescent light sheet (W&Co LED panel
0.8 m × 0.4 m). The experiments were carried out in an otherwise dark room, and
each experiment was recorded by video at 30 f.p.s.
First a series of experiments were carried out in order to measure velocity of the
front of the plume and the average shape of the plume head. We define the edge of
833 R2-2
 On the dynamics of starting plumes

(a) 3.0 (b) 2.5

2.5
2.0
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

1.5 2.0
1.0
0.5
1.5
0 5 10 15 20 25 30 35 1 2 3 4 5 6
t (s) Experiment number

F IGURE 1. (a) Variation of the non-dimensional velocity ξf with time for each of the six
experiments. The legend shows the source salinity, so , associated with each experiment.
The average value of ξf across each experiment and the corresponding error bars are
represented in (b), retaining the same colour scheme as in (a), and the horizontal black
line in (b) indicates the mean value of ξf across the experiments.

the plume head as the point where the intensity of the dye, integrated through the
plume, drops to 3 % of the maximum dye intensity at the source. For smaller values
of dye intensity, the background noise becomes as significant as the signal. Using a
larger threshold of 4 % leads to a reduction in the radius of the head by less than 8 %.
From dimensional analysis we expect that for a plume with a source buoyancy flux
πBo , the velocity of the plume head will scale as B1/3 −1/3
o zf , where zf is the position
of the leading edge of the plume head,

uf = ξf Bo 1/3 zf −1/3 . (2.1)

Here ξf may be taken to be the non-dimensional velocity obtained by integrating (2.1)

with uf = dzf /dt, and is given by

zf4/3
ξf = 4 1/3 . (2.2)
B (t + to )
3 o

The constant to relates to the time taken by the plume to develop a plume head after
issuing from the source. Figure 1(a) illustrates the variation of ξf , using measured
values of zf at each time t, with different lines corresponding to different experiments.
It is seen that the data collapse to a constant value after an initial adjustment above the
source. The value of to was estimated for each experiment to optimize the asymptote
of ξf to a constant value, and, in all cases, this has a value in the range 0.5–2.5 s,
which is considerably shorter than the 20–30 s typical of the experiments.
Figure 1(b) shows the average value of ξf for each experiment, with the colours
corresponding to the legend in figure 1(a), and the error bars represent the variation
of ξf in each experiment after an initial adjustment time (see figure 1a). The mean
value of ξf across experiments was found to be 1.99 ± 0.06.
https://doi.org/10.1017/jfm.2017.762

The equations for the averaged top-hat velocity up , radius b, and reduced gravity
g0p for an established steady plume in an unstratified ambient are given by classical
plume theory (Morton et al. 1956),

dup b2 dup 2 b2 dup b2 g0p

= b2 g0p ,
2
= 2αup b, = 0, (2.3a−c)
dz dz dz
833 R2-3
 N. Bhamidipati and A. W. Woods

(a) 0.7 (b) 1.2 (c) 1.2 (d) 1.2

0.6
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

1.0 1.0 1.0

0.5
0.8 0.8 0.8

0.4
z (m)

0.6 0.6 0.6

0.3

0.4 0.4 0.4

0.2

0.2 0.2 0.2

0.1

0 0.1 0 0.1 0.2 0 0.1 0.2 0 0.1 0.2

r (m)

F IGURE 2. Variation of (a) the radius r as a function of height z, and (b) the rescaled
radius r/zf as a function of rescaled height z/zf of the starting plume. Data are shown
from 25 frames, recorded with a time interval of 1 s between frames. In (b), the black
line represents the standard deviation of the rescaled data calculated for the experiment
with so = 2.5 % and Qo = 1.33 cm3 s−1 . (c) Time average of the rescaled radius of the
starting plume as a function of rescaled height for each of the six experiments, with the
same colour scheme as in figure 1, with the black dashed line representing the standard
deviation of the profiles relative to the mean. (d) Averaged shape of the starting plume
obtained by averaging the profiles in (c). The blue dashed line represents the best fit for
the radius of the steady plume which follows the plume head.

where α is the entrainment coefficient, which has a value of 0.13 ± 0.01 (Morton et al.
1956). Solving the steady-state plume equations given by (2.3) gives the average top-
hat velocity within the plume up as
 1/3
5 9αBo
up = z−1/3 = ξp Bo 1/3 z−1/3 . (2.4)
6α 10

Figure 2(a) illustrates the radius of the plume as a function of height in the plume
at a series of times. By scaling the height zf and radius r at a time t by the position of
the leading front of the plume zf (t), we find that the starting plume can be mapped to
https://doi.org/10.1017/jfm.2017.762

a universal shape, suggesting that both the plume head and plume are self-similar, as
shown in figure 2(b). In this figure, the black line represents the standard deviation of
the plume radius for the 25 frames shown. The relatively large standard deviation is a
result of turbulent fluctuations in the position of the edge of the plume with time. The
plume profiles obtained in each experiment were time-averaged, and the mean profiles
for the six experiments are shown in figure 2(c). In this figure, the black dashed line
833 R2-4
 On the dynamics of starting plumes
represents the standard deviation of these averaged experimental profiles. Figure 2(d)
illustrates the average shape for the six experiments. The radius of the steady plume
behind the plume head, shown in figure 2(d), increases linearly with height, as shown
by the dashed line rf = λ(z + zo ), where λ = 0.15 and zo is the virtual origin. The
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

intercept on the line rf = 0 occurs at z ≈ −4.5 ± 1.5 cm suggesting the virtual origin
has value zo ≈ 4.5 cm, which is small compared to the distance of order 60–70 cm
travelled by the plumes. The ratio of the velocity of the front of the plume head (2.1)
to the velocity of a steady plume at that height (2.4) (ξf /ξp ) lies in the range 0.63 ±
0.04, which is consistent with the results of Turner (1962) (0.61 ± 0.05) and Scase
et al. (2009) (0.65).
The linear increase in plume radius with height shown in figure 2(d) is observed
to be valid only below the height z = 0.71zf , after which the difference between the
best-fit straight line modelling the radius rf and the average of the experimental data r
is observed to increase more rapidly (figure 2d). We define this point to be the lower
boundary of the plume head zb . By comparing the data with the model linear increase
in radius with height, we find that in all experiments, the height of the lower boundary
remains nearly constant, zb = (0.70 ± 0.02)zf . Given the near constant ratio between zb
and zf and the self-similar shape of the plume, we estimate the velocity of the lower
boundary of the plume head
 
zb
ub = uf = ξb Bo 1/3 zb −1/3 . (2.5)
zf
The ratio of the velocity of the lower boundary of the plume head to the velocity
of a steady plume at that height (ξb /ξp ) is found to be 0.39 ± 0.04. This is consistent
with Turner’s (1962) estimate of 0.38 and the maximum theoretical prediction of 0.42
given by Delichatsios (1979).
In order to help understand the interaction between the plume and plume head,
pulses of dye were injected into the starting plume to enable the visualization of the
path followed by individual fluid elements in the plume. The colour of the plume
fluid was changed by injecting a small amount of dye (less than 0.2 cm3 ) through the
nozzle using a syringe at a volume flux of 0.2 cm3 s−1 , so that the volume flux of dye
injected does not produce a significant change in the volume flux exiting the nozzle
(which has value Qo = 1.33 cm3 s−1 ). In experiment A, only one pulse of dye was
injected at the time t ≈ 2.5 s to visualize the propagation of dye through the plume.
Figure 3 illustrates the evolution of the pulse of dye in the plume. In this figure, the
black dashed lines represent the positions of the front and back of the plume head as
defined by the relation zb (t) = 0.71zf (t). The series of photographs show, firstly, the
accumulation of the pulse of dye within the region defined by zb < z < zf , and secondly,
the irregular turbulent structure of the plume head as illustrated by the motion of the
packets of dyed fluid which are visible approaching the leading edge of the plume. In
experiment B, a pulse of dye was added to the source fluid at five successive times
during the experiment in order to visualize the motion of the plume fluid as it rises
towards the leading edge of the plume and then mixes into the plume head. In figure 4,
https://doi.org/10.1017/jfm.2017.762

we show a time series of a vertical line of pixels passing through the source. This
figure illustrates the successive waves of dyed fluid which migrate to the front of the
flow. Here they accumulate and mix into the head. In this figure two sets of lines are
shown. First the two dashed lines follow the evolution of the front and back of the
plume head. These lines are given by
zf 4/3 = 43 ξf Bo 1/3 (t + to ) and zb 4/3 = 43 ξb Bo 1/3 (t + to ) (2.6a,b)
833 R2-5
 N. Bhamidipati and A. W. Woods

0.6

0.4
z (m)
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

0.2

0
2 4 6 8 10 12 14 16 18 20
t (s)

F IGURE 3. Experiment A. Propagation of a pulse of dye through the plume and

accumulation of dye in the plume head. The dashed lines indicate the positions of the
front (zf ) and back (zb ) of the plume head, with zb = 0.71zf . In this experiment, so = 5 %,
Qo = 1.33 cm3 s−1 . The time interval between frames is 2 s.

0.6

0.5

0.4
z (m)

0.3

0.2

0.1

0 5 10 15 20 25
t (s)

F IGURE 4. Experiment B. Time series of the vertical centreline of the plume. The dashed
lines show the t3/4 characteristics for the position of the front (black) and back (yellow) of
the head, and of the pulses of dye (white). In this experiment, the initial salinity, so = 5 %,
and the volume flux at the source, Qo = 1.33 cm3 s−1 .

consistent with (2.1) and (2.5). Secondly a series of dashed lines zi (t) are shown, each
of which follows the ascent of one of the successive streaks of dye. These lines have
the form
zi 4/3 = 43 ξs Bo 1/3 (t − ti ), (2.7)
where ti is the time of release of the ith pulse of dye. Tracking the pulses of dye
that reach the front of the plume gives an approximate estimate for ξs , which has a
value of approximately 3.82 ± 0.7. The error bar is found by using (2.7) to model
the leading and trailing edge of each pulse of dye, which move at somewhat different
speeds due in part to the effects of dispersion (cf. Rocco & Woods 2015). The
estimate for ξs is larger than the estimates of ξf (1.99) and ξb (1.25), and shows
https://doi.org/10.1017/jfm.2017.762

that the fluid in the steady plume which develops behind the plume head continually
supplies fluid to the plume head, as is clear from the mean path followed by the dye.

3. Modelling the plume head

We now build a model of the mass, momentum and buoyancy conservation in the
plume head in order to estimate the mixing of ambient fluid directly into the head,
833 R2-6
 On the dynamics of starting plumes

(a) Displacement of (b)

ambient fluid
around the head
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

Head
Supply of Recirculation
fluid to the of fluid in
head from the head of

Steady plume
plume below the plume
Turbulent
entrainment of
ambient fluid into
the plume below
the head

F IGURE 5. (a) Cartoon illustrating the motion of fluid within the starting plume and
the motion of ambient fluid around the plume head. The motion of fluid within the
plume head shown is in the frame of reference of the plume head. (b) Illustrating various
dimensions and velocities.

and also to test the momentum balance of the head. The model describes an ensemble
average of the flow averaged over several realizations of the flow so that fluctuations
associated with the detailed distribution of individual eddies in the flow are averaged.
The plume behind the plume head is modelled as being steady, while the plume head
slowly evolves in time and space. We define the plume head to correspond to the
region zb < z < zf as defined by figure 2(d). Figure 5(a) captures the observations from
figures 3 and 4 in a schematic, and figure 5(b) illustrates the shape of the starting
plume and the various parameters involved in the modelling of the plume head.

3.1. Conservation of volume

The volume of the plume head Vc is found by evaluating the integral
Z zf
Vc = πr2 dz, (3.1)
zb

where r = r(z) is the radius of the starting plume as given by the shape profiles shown
in figure 2(a–d). Similarly, the vertical momentum of the plume head is defined as
Z zf
Mc = πr2 u dz, (3.2)
zb
https://doi.org/10.1017/jfm.2017.762

where u, the horizontally averaged vertical velocity at height z in the plume head, is
given by u = uf (z/zf ) in order that the shape of the plume head remains similar at all
times. The velocity of the centre of mass of the plume head uc = dzc /dt is found by
combining (3.2) with (3.1) (uc = Mc /Vc ). The estimate for the position of the centre
of mass of the plume head, zc , was found to have a near constant value over all the
833 R2-7
 N. Bhamidipati and A. W. Woods
experiments, zc = (0.83 ± 0.01)zf . We can define a shape factor Ω for the plume head
as
Vc
Ω= , (3.3)
(zc − zb )R2
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

where R is the maximum radius of the plume head (see figure 5b). The average shape
factor for the plume head was found to be 5.41, which is a larger value than the
shape factor of an equivalent ellipsoid (= 4π/3 ≈ 4.19). Therefore the volume of the
plume head as defined in (3.1), and using the 3 % dye threshold, which is the limit
of our experimental method, was found to be approximately 23 % larger than the
volume estimates of Turner (1962) and Scase et al. (2009), where the plume head
was assumed to have a spherical shape.
The plume head is supplied with fluid by the steady plume. The volume flux of
fluid supplied by the steady plume to the plume head can be found by solving the
steady plume equations for the radius and the velocity as defined earlier (Morton et al.
1956). The flux entering the plume head, evaluated at z = zb , is given by Qp = πb2 (up −
ub )|z=zb , where the expression (up − ub ) accounts for the fact that the plume head is
advancing. The equation for conservation of volume of the plume head is given by
dVc
= Qp + Vc 2/3 uc , (3.4)
dt
where the second term on the right hand side gives the flux due to entrainment of
ambient fluid into the plume head. Solving (3.4) using the estimates of Vc , Qp , uc
and zb , suggests that the ratio Qp /(dVc /dt) = 0.88 ± 0.06. This implies that the time-
averaged volume flux entrained directly into the plume head is small compared to the
flux supplied from the steady plume. The range of values of Qp /(dVc /dt) is consistent
with the error in our measurement of dVc /dt, and arises in part owing to the turbulent
fluctuations of the flow and in part owing to the uncertainty in the precise value of
the entrainment coefficient of the steady plume, α. Based on our data, we estimate
that  R= 0.14 ± 0.08. If the plume head were to entrain over its entire surface area
z
(Sc = zbf 2πr dz), then the equivalent entrainment coefficient per unit surface area
would be 0.04 ± 0.02, which is smaller than for a steady plume, and consistent with
the entrainment coefficient for the plume head, 0.04–0.05, as proposed by Scase et al.
(2009).

3.2. Conservation of buoyancy

The conservation of buoyancy for the plume head is given in terms of the flux of
buoyancy entering the head from the steady plume. Since the flux of buoyancy
is defined as πg0p b2 up in the top-hat model, then since the rear of the plume
head advances with speed ub , the flux entering the plume head is given by
πb2 (up − ub )g0p = Qp g0p . This leads to the equation for the conservation of buoyancy,
dg0c Vc
= Qp g0p |z=z . (3.5)
dt b

Here g0c is the reduced gravity of the plume head averaged throughout the plume
https://doi.org/10.1017/jfm.2017.762

head, and g0p is the reduced gravity of the following steady plume, averaged across
a horizontal surface at height z, obtained by solving the steady-state plume equations
(Morton et al. 1956)
5Bo 9αBo −1/3
 
gp (z) =
0
(z + zo )−5/3 , (3.6)
6α 10
833 R2-8
 On the dynamics of starting plumes
where zo is the virtual origin of the plume. The total buoyancy in a steady plume in
the region 0 < z < zb (t) is given by
Z zb      −1/3
3π 6α 9α
π gp (z)b(z) dz =
0 2
Bo 2/3 [(zb + zo )4/3 − zo 4/3 ].
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

(3.7)
0 4 5 10
In the limit zo  zb , appropriate for our experiments with zo ∼ 4.5 cm and zb ∼
60–70 cm, we may approximate the right-hand side of (3.7) by the expression

ξb
     −1/3  
3π 6α 9α 2/3 4/3
Bo zb = πBo t, (3.8)
4 5 10 ξp
where πBo t is the total buoyancy supplied from the source up to a time t, and ξp and
ξb are defined in (2.4) and (2.6), respectively. Here (ξb /ξp ) = 0.39 ± 0.04 (see § 2).
The remainder of the buoyancy accumulates in the plume head and is given by
ξb
 
1− πBo t = (0.61 ± 0.04)πBo t. (3.9)
ξp

3.3. Conservation of momentum

We propose that the equation for conservation of momentum of the plume head has
the form
dMc
(1 + A) = Qp up |z=zb + g0c Vc − Cd u2c (πR2 ), (3.10)
dt
where A is the added mass of the plume head, and the first two terms on the right
hand side denote the momentum flux supplied to the plume head by the fluid entering
from the following steady plume and the buoyancy force on the head. We have also
included a drag force, with drag coefficient Cd , which is similar in effect to form drag
on a moving body. The added mass, A, arises due to the acceleration of ambient fluid
around the head, thus contributing an increased force on the plume head (Ohl, Tijink
& Prosperetti 2003).
We note that this model differs from some of the earlier literature in which the
plume head has been modelled as a spherical vortex (Turner 1962; Scase et al.
2009). Our observations of the structure and motion of fluid in the plume head
(figures 3 and 4) suggest that the flow does not develop a coherent vortex-ring-type
structure. We observe a more irregular motion as turbulent pulses of buoyant fluid
supplied to the plume head reach the plume front and spread radially, with successive
parcels spreading in different directions. Instead, our approach follows from the work
of Scorer (1997) who proposed a similar momentum conservation equation for a
discrete thermal, although he did not include the turbulent drag term since he did not
measure the shape of the thermal independently of the momentum equation.
We have solved (3.4) and (3.5) with the modified momentum equation for the plume
head given by (3.10) to find the drag coefficient Cd , which has a value 4.2 ± 1.4. The
range of values of Cd correspond to the uncertainty in our estimate of , and hence
https://doi.org/10.1017/jfm.2017.762

we establish a relationship between the entrainment coefficient  and drag coefficient

Cd of the plume head by combining (3.4) and (3.10),
Cd = (6.0 ± 0.3) − (11.8 ± 1.2). (3.11)
The added mass of the plume head was assumed to be equal to the added mass of a
sphere, A = 0.5, consistent with Scorer (1997) and Ohl et al. (2003).
833 R2-9
 N. Bhamidipati and A. W. Woods

(a) 1.2 (b) 1.5 (c) 1.5

1.0
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

0.8 1.0 1.0

0.6

0.4 0.5 0.5

0.2

0 0.1 0.2 0 2 4 6 8 0 0.2 0.4 0.6 0.8 1.0

tN

F IGURE 6. (a) Rescaled radius r/zf as a function of the rescaled height z/zf in a linearly
stratified ambient. The figure shows the radius averaged across five experiments when the
plume is at different fractions of the maximum top height zT (zf /zT = 0.5–1.0). The red
profile shows the profile of radius when the plume reaches its top height (z = zT ). The
black profile shows the equivalent profile in an unstratified ambient (see figure 2d). In
the five experiments, N 2 = 0.35, Q = 1.01 cm3 s−1 , so = 10 %, 12 %, 13 %, 14 %, 15 %.
(b) Plot of dimensionless height of rise (z/zs ) versus time (tN), where zs is the maximum
height of the steady plume in a stratified ambient. The graph illustrates the height of rise
as predicted by the new model (blue), and is compared against experimental data (black
circles) and the model proposed by Turner (1962) using the equivalent top-hat coefficients
estimated by Scase et al. (2009) (black). In (c), the instantaneous neutral height of the
plume head (black), which is the height in the ambient at which the density of the plume
head equals the density of the ambient, and the positions of the centre (yellow) and
top (blue) of the plume head are shown. The horizontal lines in (b) and (c) show the
maximum height of the continuing steady plume (red) and the maximum height of rise
of the centre of the plume head (green) as predicted by the model.

4. The starting plume in a stratified ambient

In this section, we describe how a starting plume rises through a linearly stratified
ambient, characterized by the buoyancy frequency N, where N 2 = −(g/ρo )(dρa /dz), in
which ρa is the density of the ambient at height z and ρo is a reference density.
We have carried out a series of experiments in which we set up a linear stratification
in the experimental tank with N 2 = 0.35 s−2 . We then explored the ascent of series
of starting plumes with volume flux Qo = 1.01 cm3 s−1 and salinity so = 10, 12, 13,
14, 15 %. In figure 6(a) we illustrate the variation of the rescaled radius (r/zf ) as a
function of the rescaled height (z/zf ) at the different times when the plume height zf
was a fraction 0.5 (grey), 0.6 (yellow), 0.7 (blue), 0.8 (green), 0.9 (purple), and 1.0
(red) of the maximum height of rise of the starting plume zT . In each case, we have
https://doi.org/10.1017/jfm.2017.762

averaged the profile over the five experiments. The figure shows that as the plume
approaches the maximum height, the plume radius begins to increase as it slows down
and intrudes laterally. For reference, the black profile represents the equivalent profile
of the plume in an unstratified environment (figure 2). In figure 6(b), we illustrate the
average of the experimental data for the height of the top of the plume head as a
function of dimensionless time, tN (black circles).
833 R2-10
 On the dynamics of starting plumes
To include stratification in the model of a starting plume, we modify the buoyancy
conservation equation (3.5) to the form (cf. Morton et al. 1956)
dg0c Vc
= Qp g0p |z=z − N 2 Vc uc . (4.1)
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

dt b

We now solve (3.4), (3.10), and (4.1), for mass, momentum, and buoyancy
conservation of the plume head, coupled with the solutions of the steady plume
equations in a stratified environment (cf. Morton et al. 1956) to calculate the supply
of volume, momentum, and buoyancy into the plume head. To complete the model,
we need to specify the shape of the plume head in order to relate the position of the
base of the head, zb , to the position of the centre of mass of the head, zc , and to
estimate the radius of the head, R, according to (3.3).
Below the maximum height of rise, before the shape of the plume is influenced
by the stratification (figure 6a), we might expect that the relationships between zb , zc ,
zf , and Ω for the unstratified case (zb /zf = 0.71, zc /zf = 0.83, Ω = 5.41) provide a
reasonable approximation. In these calculations, we also need a specific value of the
drag coefficient, Cd , and the entrainment coefficient, , which were only constrained
by the relation Cd = (6.0 ± 0.3) − (11.8 ± 1.2) from our experiments in an unstratified
ambient. We find that the error between the model and experimental data is less than
5 % for the range of values of Cd given by the relation (3.11). The predictions of the
model are shown in figure 6(b) (blue line), where the mean value of (= 0.14) was
used to constrain the value of the drag coefficient Cd . Figure 6(b) also shows the
predictions using the vortex-ring model proposed by Turner (1962) and Scase et al.
(2009) (black dashed line), in which the parameters were taken from those given by
Scase et al. (2009). We used the parameters from Scase et al. (2009), since in the
original work of Turner (1962), the speed of the back of the plume head was assumed
to equal a fraction 0.6 of the maximum centreline speed of the steady plume rather
than a fraction 0.6 of the top-hat speed as used by Scase et al. (2009), and which
is consistent with our experimental results. Equation (4.1) was used to describe the
conservation of buoyancy. The figure shows that the new model predicts the height
of rise somewhat more accurately until the plume reaches the maximum height of
the steady plume, where the shape of the plume head and its dynamics change as a
result of the stratification (figure 6a). Figure 6(c) illustrates the variation of the height
at which the plume head would be neutrally buoyant (black line) as a function of the
height of the centre of the plume head, as well as the height of the top (blue dashed
line) and centre (yellow line) of the plume head. The neutral buoyancy height of the
plume head, which occurs where the yellow and black lines meet, is very similar
to the maximum height of the steady plume (red horizontal line). The maximum
overshoot of the centre of the plume head above this height (green horizontal line),
as predicted by the model, is comparable to the experimental measurement of the
maximum height of rise of the plume head as shown in figure 6(b). However, given
the distortion of the shape of the plume head near the maximum height of rise
(figure 6a), we do not expect the model prediction for the height of the leading edge
https://doi.org/10.1017/jfm.2017.762

of the plume head to follow accurately the motion of the leading front of the plume
head beyond this height.

5. Discussion
In this paper we have carried out some new experiments to describe the mixing
and interaction between the plume head, the following steady plume, and the ambient.
833 R2-11
 N. Bhamidipati and A. W. Woods
We found that the buoyancy force on the plume head combined with the momentum
flux supplied to the plume head from the following steady plume exceed the rate of
change of momentum of the plume head. We propose that a turbulent drag on the
plume head develops which accounts for this difference. We have estimated that the
Downloaded from https://www.cambridge.org/core. Purdue University Libraries, on 08 Nov 2017 at 07:46:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.

drag coefficient is given by Cd = 4.2 ± 1.4, with the range of values corresponding
to the uncertainty in the estimate of the entrainment coefficient  = 0.14 ± 0.08. The
range of values of the entrainment coefficient suggest that the plume head entrains
only a small fraction of its volume from the ambient, with the dominant influx of fluid
supplied by the following steady plume. We also found that a fraction 0.61 ± 0.04
of the total buoyancy supplied by the source lies within the plume head, so that the
plume head is more buoyant than the developing steady plume immediately behind at
z = zb .
In a stratified ambient the assumption that the plume head retains the same shape as
in the unstratified case allows an initial estimate for the height of rise with time, which
compares well with experimental data up to the point at which the plume head reaches
the maximum height of rise of the ensuing steady plume. It would be of interest to
explore the evolution of the plume head beyond the maximum height, perhaps with an
investigation into the dynamics of a negatively buoyant forced thermal in a uniform
and stratified environment, to understand the dynamics and the deformation of the
plume head near its maximum height.
When a steady flux of buoyancy, generating a steady turbulent plume, is suddenly
increased to a substantially larger value, then a mixed zone of fluid may develop and
grow in the transition region, somewhat akin to the starting plume. It would be of
interest to develop the experimental modelling presented herein to estimate the drag
and entrainment coefficients associated with such transition zones in order to build a
model of their evolution.

References
D ELICHATSIOS , M. A. 1979 Time similarity analysis of unsteady buoyant plumes in neutral
surroundings. J. Fluid Mech. 93 (02), 241–250.
M IDDLETON , J. H. 1975 The asymptotic behaviour of a starting plume. J. Fluid Mech. 72 (04),
753–771.
M ORTON , B. R., TAYLOR , G. & T URNER , J. S. 1956 Turbulent gravitational convection from
maintained and instantaneous sources. Proc. R. Soc. Lond. A 234 (1196), 1–23.
O HL , C. D., T IJINK , A. & P ROSPERETTI , A. 2003 The added mass of an expanding bubble. J. Fluid
Mech. 482, 271–290.
ROCCO , S. & W OODS , A. W. 2015 Dispersion in two-dimensional turbulent buoyant plumes. J. Fluid
Mech. 774, R1.
S CASE , M. M., A SPDEN , A. J. & C AULFIELD , C. P. 2009 The effect of sudden source buoyancy
flux increases on turbulent plumes. J. Fluid Mech. 635, 137–169.
S CASE , M. M., C AULFIELD , C. P., DALZIEL , S. B. & H UNT, J. C. R. 2006 Time-dependent plumes
and jets with decreasing source strengths. J. Fluid Mech. 563, 443–461.
S CORER , R. S. 1997 Dynamics of Meteorology and Climate. Wiley.
T URNER , J. S. 1957 Buoyant vortex rings. Proc. R. Soc. Lond. A 239 (1216), 61–75.
https://doi.org/10.1017/jfm.2017.762

T URNER , J. S. 1962 The starting plume in neutral surroundings. J. Fluid Mech. 13 (03), 356–368.
W OODS , A. W. 2010 Turbulent plumes in nature. Annu. Rev. Fluid Mech. 42, 391–412.

833 R2-12