You are on page 1of 9

E-proceedings of the 36th IAHR World Congress

28 June 3 July, 2015, The Hague, the Netherlands

OIL DROPLET SIZE CALCULATION WITH AND WITHOUT DISPERSANTS IN AN UNDERWATER WELL
BLOWOUT
1

INDRAJITH D. NISSANKA AND POOJITHA D. YAPA


(1)

Graduate Research Assistant, Department of Civil & Environmental Engineering, Clarkson University, Potsdam, NY 13699, USA,
e-mail: nissannd@clarkson.edu
(2)

Professor, Department of Civil & Environmental Engineering, Clarkson University, Potsdam, NY 13699, USA,
e-mail: pyapa@clarkson.edu

ABSTRACT
In an accidental underwater oil well blowout, the major concerns of emergency response and contingency planning are to
predict the amount of oil that reaches the surface, how fast it reaches the surface, and the surface spreading. In this
regards, oil droplet sizes play a critical role, as the rise velocities of droplets essentially depend on their size although nonlinearly related due to a variety of factors including the changes in shape. Therefore, accurate calculation of droplet size
distribution is very important in modeling and predicting the fate and transport of oil in the water column and surface
spreading. However, the current knowledge on oil droplet size calculation/estimation is limited with only a few methods
available. None of the above methods are robust enough to calculate different droplet sizes corresponding to wide ranges
of release conditions. Further, the theoretical knowledge behind the computation of small droplet sizes is almost none in
existing methods. In this paper we introduce an improved theoretical method starting with Bandara and Yapa (2011) to
calculate droplet sizes for a wide range of underwater accidental oil releases. Droplet evolution is modeled using
population balance equation. Here, we consider the oil droplet break up and coalescence as death and birth of droplet
sizes. This results in 4 primary components that had to be computed separately together to find the final distribution. In
computing these components we use new improved methods for energy dissipation, droplet breakup rate and
coalescence rate closures. The new model is validated with experimental data from Deepspill field experiments and a
series of recent experiments conducted by SINTEF (Brandvik et al., 2013) which include cases with and without
dispersants. The model results show a very good agreement with experimental data for without dispersants cases and
with dispersants cases model calculations show reasonable initial results with good promise for continuation of the work.
This work is still progressing.
Keywords: Droplet breakup, droplet coalescence, underwater jet/plumes, deepwater oil well blowouts
1.

INTRODUCTION

In oil spill models, accurate calculation of oil Droplet Size Distribution (DSD) is vital in predicting the fate and transport of
oil in the water column and surface spreading from underwater oil well blowouts (Niu et al., 2011). Most of the oil spill
models require providing a DSD as an input or use a simple empirical formula to calculate DSD in simulating fate and
transport of oil. The given DSD plays a critical role in model predicted results such as the amount of oil that reaches the
surface, time it takes oil to first appear in the surface, and surface spreading which are important in emergency response
and contingency planning. The rise velocities of droplets essentially depend on their size although non-linearly related to
the size due to a variety of factors including the changes in shape (Zheng and Yapa, 2000). Further, the time droplets
remain in the water column affect the lateral movement of oil as the velocity currents advect the droplets. In toxicity
studies, the size of the oil droplet and the duration of it in water column will impact the amount of dissolved toxic oil
components. In deepwater oil well blowouts, released oil breaks into small droplets due to the instabilities created by high
turbulence at the release point. As the droplets travel with the jet / plume they further break into smaller droplets or
coalesce to form larger droplets. The addition of dispersants causes changes in DSD resulting smaller droplet sizes.
Previous studies show that DSD stabilizes within few meters from the release point (Bandara and Yapa, 2011).
In literature, there are only a limited number of models available to calculate the oil DSD in a sub sea oil well blowout
(Johansen, 2000; Chen and Yapa, 2007; Bandara and Yapa, 2011; Johansen et al., 2013;Zhao et al., 2014),out of which
most of the models are empirical or semi empirical with few exceptions. Bandara and Yapa (2011) presented a theoretical
formulation based on the Population Balance Equation (PBE) and droplet breakup and coalescence theories. The model
was tested with limited data, and calibrated only for a limited droplet size range (1- 10 mm). Zhao et al. (2014) also
presented a model which considers both surface tension and viscous effect in breakup. Their model was used to simulate
the droplet breakup experiments by SINTEF laboratory experiments and Deepspill experiments. However, their model
results depend on two coefficients for breakup and coalescence which appears to depend on the case. The coefficients
are not well understood for deepwater oil and gas jets/plumes. The above methods are not robust enough to calculate
different droplet sizes corresponding to wide ranges of release conditions that can exist in deepwater oil and gas well
blowouts. The Deepspill experiments had a release velocity of 1.8 m/s. SINTEF laboratory experiments (Brandvik et al.,
1

E-proceedings of the 36th IAHR World Congress,


28 June 3 July, 2015, The Hague, the Netherlands

2013) included velocities as high as 26.5 m/s. In field conditions, such high release conditions can happen if the opening
is small or release volume rate is large. This will result in ultra-small droplet sizes. Also, the injection of dispersants to the
oil jet/ plume causes the oil-water interfacial tension to drop significantly resulting in smaller droplet sizes. The capability of
the model to calculate both mm range as well as m range droplet sizes due to different release conditions is important.
However, the present knowledge behind the computation of small droplet sizes (in the m range) is almost none in
existing methods. In this paper, we introduce an improved theoretical method to calculate droplet sizes for a wide range of
accidental oil releases in underwater oil and gas releases. The model is validated with experimental data from Deepspill
experiments (Johansen et al., 2001) and a series of very recent laboratory experiments conducted by SINTEF (Brandvik et
al., 2013,2014). These comparisons include experiments with dispersants and without dispersants.
2.

MODEL FORMULATION

The model presented in this paper is developed based on the initial framework of the DSD model presented by Bandara
and Yapa (2011). The new model can be used in a wide range of droplet sizes, whereas the Bandara and Yapa (2011)
model is calibrated only for a limited droplet size range (1 mm to 10 mm) and does not work for the small droplet sizes.
Model considers the droplet breakup and coalescence as source and sinks terms respectively in calculating the DSD. In
this formulation, theoretical modifications were made to the droplet breakup and coalescence closures such that they can
be used to calculate different droplet sizes corresponding to wide ranges of possible release conditions in under water oil
well blowouts. The bubble/droplet breakup and coalescence models are initially developed in the chemical engineering
and pharmaceutical industry. (e.g. Coulaloglou and Tavlarides, 1977; Prince and Blanch, 1990; Lehr et al., 2002; Chen et
al., 2005). The theories developed are based on the concepts of isotropic turbulence and predefined probabilistic functions
and can be used in other turbulent flow applications such as turbulent jets. Bandara and Yapa (2011) adapted these
concepts to simulate droplet breakup and coalescence in underwater oil and gas jets/plumes resulting from a well blowout.
This paper presents an extension of the work by Bandara and Yapa (2011) with improved theoretical formulations.
The model is integrated into the Comprehensive Deepwater Oil and Gas model (CDOG). The details of CDOG model can
be found in Zheng et al. (2003). The hydrodynamics and thermodynamics of the plume dynamic stage is modeled using
integral largrangian control volumes (Zheng et al., 2003). The evolution of droplets within a control volume is modeled
using the PBE (Bandara and Yapa, 2011). PBE is the key equation in the model that tie up the four primary process,
described below, and is given by (Hagesaether et al., 2002; Lehr et al., 2002),
( , )

(v v , v ) V(v v , t)V(v , t) dv (v, v ) V(v)V(v )dv dv + (v, v ) (v)V(v , t) dv (v)V(v, t) [1]

where ( , ) = total volume of droplets each having a volume at a given time t. ( , ) = coalescence rate of droplets
having volume and . ( ) = breakup rate of droplet size . ( , ) = probability density of daughter bubbles.
Equation 1 is solved numerically by using the fixed pivot technique proposed by Kumar and Ramkrishna, (1996). In this
method droplet sizes are divided into a finite number of representative size classes (pivotal sizes) and the equation is
integrated over each size class (Wang et al., 2005a). Any droplets that fall between two representative droplet size
classes are represented by redistribution of mass into neighboring classes. The details on the redistribution of mass can
be found in Bandara and Yapa (2011). To close the Equation 1 droplet breakup and coalescence closures need to be
calculated and those formulations are described in subsequent sections.
2.1 Droplet Breakup
The droplets in a fluid flow continuously interact with the continuous phase (sea water in this study) hydrodynamics. When
the hydrodynamic forces exerted on a droplet is larger than the stabilizing forces the droplet breaks up. From previous
studies, four major mechanisms can be identified which can cause breakup of droplets (Prince and Blanch, 1990): (1)
turbulent fluctuation, (2) viscous shear stress, (3) wake effect, and (4) interfacial instabilities. However, previous modeling
studies and experimental evidences (Hesketh et al., 1991) suggest that turbulent fluctuation is the dominant mechanism of
droplet breakup in under water oil jets/plumes and the other effects has negligible contribution. Therefore, only the droplet
breakup due to turbulent pressure fluctuation is included in this study.
In a turbulent flow, droplet breakup occurs due to interaction of oil with turbulent eddies. When a droplet-eddy collision
happens, first the droplet deforms and if the eddy has sufficient energy it will cause the droplet breakup. The breakup rate
of a droplet of size class due to the interaction with an eddy of size
expressed as a product of collision frequency of
and breakup efficiency
(Hagesaether et al., 2002). The turbulent collision
the droplet with the turbulent eddy
rate
can be expressed as,
=

[2]

where ,
are turbulent velocities of droplet size class i and eddy size class j respectively; ,
are the number
concentrations of droplet size class i and eddy size class j respectively; , is the collision cross sectional area of droplets
of size class i and eddies of size class j. Previous studies showed that only eddies in the inertial sub- range have an effect
on the droplet breakup (Prince and Blanch 1990). A finite number of eddy size classes (Ne) are assumed in inertial subrange. The number of eddies in a size class is calculated as (Prince and Blanch 1990).
2

E-proceedings of the 36th IAHR World Congress


28 June 3 July, 2015, The Hague, the Netherlands

=
where,

[3]

= lower limit of the inertial subrange and

(1 +

);

= eddy class number.

The upper and lower limits of inertial subrange is calculated from, =


and =
/0.51 respectively (Bandara and
Yapa 2011). Where the constants c = 20 and
= 0.45 is used in this study. The turbulent velocity of an eddy of size in
/
the inertial subrange of isotropic turbulence is calculated as
= 1.4
(Hinze,1955). Primarily, fluctuations of
droplets are caused by turbulent eddies and therefore the turbulent velocity of a droplet is assumed to be similar to the
turbulent velocity of an eddy of the same size (Prince and Blanch, 1990).
The breakup efficiency expresses the ratio of number of breakup events to the number of collisions. In the literature
different breakup criteria has been used by different authors. Luo and Svendsen (1996a) used the criterion that the energy
of the incoming eddy should be larger than the surface energy increase during the breakup to calculate the breakup
efficiency. Luo and Svendsen's (1996a) model has a weakness that it does not consider the minimum attribute of the
breakup (Hagesaether et al., 2002). Wang et al. (2003) overcome this difficulty in Luo and Svendsen's (1996a) model by
introducing capillary criteria to calculate the minimum attribute of breakup. Bandara and Yapa (2011) converted this
capillary criterion to a minimum daughter droplet as
= / ( ) / and droplets of size less than
were not
generated in their model. In this paper, we used Luo and Sevendsens (1996a) surface energy criteria with the
modifications by Wang et al (2003) to calculate the breakup efficiency. The minimum attribute is calculated similar to
Bandara and Yapa (2011). Luo and Svendsen (1996) expressed the breakup probability as,
= exp( )

= exp( )

[4]

where, = ( )/ ( ) is the normalized kinetic energy of an eddy,


is the critical value of which is expressed as
is the mean turbulent kinetic energy
. Here, ( ) = surface energy increased during the breakup;
= ( )/
of an eddy with size .
is calculated using the formula suggested by Luo and Svendsen (1996a). Hibiki and Ishii
(2002) calculated the energy increase during the breakup as the average value of the energy increase in the breakup of
two equal size droplets and the breakup of smallest and largest drop size.
( )=

where, d is the diameter of parent droplet size.


calculated by volume balance.

[5]

is calculated by using the capillary criterion, and

can be

2.1.1 Daughter Droplet Probability Function


To close the breakup kernel, a daughter droplet size distribution has to be defined. In this paper, the model suggested by
Tsouris and Tavlarides (1994) is used, where the daughter droplet distribution function can be defined as follows,
,

( )]

( )]}

[6]

where,
,
= probability density of producing droplet size class from the breakage of droplets of size class .
=
/
/
/
+

.
= 2
/2
.
=
+

. ,
are the droplet diameters of
size class i and j respectively.
is the minimum droplet size class calculated using capillary criteria described above.
2.2 Droplet Coalescence
In a fluid flow, droplets collide due to the relative motion between them. The relative motion between droplets can happen
due to different mechanisms. In literature, we can find at least five different mechanisms which cause the relative motion
between particles in a fluid flow (Liao and Lucas, 2010): (1) turbulent fluctuations in continuous phase, (2) different droplet
rise velocities, (3) mean velocity gradient/ shear flow, (4) droplet capture in an eddy, and (5) wake interaction. However,
only the collisions due to turbulent fluctuations and different rise velocities will be considered in this paper. Previous
studies showed that other mechanisms are insignificant (Bandara and Yapa 2011). When droplets collide only a fraction of
the droplet collisions will results in coalescence and others will bounce back. Similar to the breakup rate, coalescence rate
is expressed as the product of collision frequency and coalescence efficiency. The collision frequency is calculated by
adding the collisions due to two mechanisms. The turbulent collision rate ( ) is calculated by assuming the droplet
movement to be similar to the gas molecule motion in gas kinetic theory (Kennard, 1938; Prince and Blanch, 1990). The
equation is similar to Equation 2, except that the eddy velocity and eddy concentration are replaced with droplet velocity
and droplet concentration of droplet size dj (Prince and Blanch, 1990). Buoyancy collision rate (
) which is due to the
different buoyant velocities of droplets can be calculated as follows (Prince and Blanch, 1990).
=
,

[7]

are the rise velocities of bubble size classes i and j respectively.

E-proceedings of the 36th IAHR World Congress,


28 June 3 July, 2015, The Hague, the Netherlands

The coalescence efficiency expresses the ratio between the number of successful coalescence events to the total number
of collision events. The most popular criterion used to express the coalescence efficiency is the film drainage theory. The
film drainage theory suggest that droplet coalescence takes place in three steps: (1) two droplets collide, tapping a small
liquid film between them, (2) The liquid film drains to a critical thickness, and (3) film ruptures causing droplet coalescence
(Prince and Blanch, 1990). Two droplets will coalesce only if they remain in contact for sufficient time for film to drain out
to a critical thickness. Based on this, coalescence efficiency is expressed as,
= exp

[8]

where,
and
are the coalescence time and contact time of the droplets of size classes i and j respectively. Using the
parallel film theory Luo and Svendsen (1996b) calculated the contact time and expressed the coalescence efficiency as,
= exp

.
(

[9]

where,
=
/ ,
is the density of continuous phase and
is the relative velocity between two droplets.
= / ,
is the virtual mass coefficient, which is taken as a constant between 0.5 and 0.8 (Liao and Lucas, 2010;
Bizmark et al., 2012). Luo and Svendsen (1996b) used a constant value of = 0.4.
2.3 Energy Dissipation Rate
In the above formulation turbulent energy dissipation rate is a key characteristic that relate the droplet breakup and
coalescence with continuous phase turbulence. Accurate calculation of energy dissipation rate is important in predicting
the DSD. However, there is no well-developed formula to calculate the energy dissipation rate in a jet/plume. Johansen et
al. (2000) proposed to use
=
/ , where = 0.01,
is the centerline velocity. Zhao et al., (2014) suggest a
constant energy dissipation rate of = 0.03 /
close to the origin of the jet and varying energy dissipation rate further
away. Bandara and Yapa (2011) used the following formula to calculate the energy dissipation rate in the jet/plume.
=

| |

[10]

Bandara and Yapa (2011) used a constant value of cd = 2.0. However, in the present study we found that cd varies with the
release condition. The variation of cd with release conditions are discussed later in this paper.
3.

COMPARISON WITH EXPERIMENTAL DATA

The model is used to simulate a series of recent laboratory experiments conducted by SINTEF (Brandvik et al., 2013,
2014) and the large scale field experiment, Deepspill (Johansen et al., 2001). The experiments were simulated by using
the newly modified droplet breakup and coalescence modules described above. All the experiments and data used for the
comparison are listed in Table 1 and comparisons of experimental data and the model computed results are given in the
next two sections.
3.1 SINTEF Experiments
SINTEF conducted a series of experiments to study the oil DSD in an underwater oil well blowout (Brandvik et al., 2013,
2014). The experiments were conducted in a 3 m diameter, 6 m height cylindrical tank. The tank was filled with sea water
0
maintained at 8 C. Norwegian crude oil (Oseberg blend) was used in the experiments. As listed in Table 1, the
experiments consisted of two parts: droplet sizes with release conditions and the effect of dispersants. Release rates
ranged from 0.2 L/min to 5L/s with different release diameters (nozzle diameters) ranging from 0.5 mm to 3 mm being
used. The oil DSD was measured for each experiment at 2m above the release point and the volume fractions are given
for 29 droplet size classes with representative diameter varying from 4.5 to 460 m.
The newly developed model is used to simulate some of those experiments and the results were compared with the
experimental observations. Simulations used 30 droplet size classes with representative diameter varying from 4.5 m to
500 m. An initial seed diameter is required to start the simulation and in these simulations seed diameter of 500 m was
used. As shown by Bandara and Yapa (2011), the initial seed diameter does not impact the final DSD. Figure 1 shows the
comparisons of the model generated DSD (by volume) and the experimentally observed DSD for different release
conditions. The cases shown in the Figure 1 are for oil only release and does not include dispersants. Figure 2 shows the
comparison of DSD for different dispersant to oil ratio (DOR) for the base case of Exp 2 (1.5 L/min release rate and 1.5
mm diameter).

E-proceedings of the 36th IAHR World Congress


28 June 3 July, 2015, The Hague, the Netherlands

100

Exp 1

80

Model
Exp

60
Q = 1.0 L/min
D = 1.5 mm
rel

40

Vrel = 9.4 m/s


cd = 0.1

20

Cumulative volume percentage

Cumulative volume percentage (%)

100

0
0

50

Exp 2
80

Model
Exp

60
Q = 1.5 L/min
Drel = 1.5 mm

40

Vrel = 14.1 m/s


cd = 0.2

20

100 150 200 250 300 350 400 450 500 550

50

100 150 200 250 300 350 400 450 500 550

Droplet diameter (m)

100

Exp 3
Model
Exp

80

60

Q = 5 L/min

Drel = 2 mm

40

Vrel = 26.5 m/s


cd = 0.8

20

Cumulative volume fraction (%)

Cumulative volume percentage (%)

100

Droplet diameter (m)

Exp 4
80

60
Q = 5 L/min

50

100 150 200

250 300 350 400

Drel = 3 mm

40

Vrel = 11.8 m/s


cd = 0.036

20

0
0

Model
Exp

450 500 550

50

100 150 200 250 300 350 400 450 500 550

Droplet diameter (m)

Droplet diameter (m)

Figure 1: Comparison of model calculated DSD (by volume) with experimental observations: SINTEF experiments without dispersants
given in Table 1: (a) 1.0 L/min, 1.5 mm, (b) 1,5 L/min, 1.5 mm, (c) 5L/min, 2mm (d) 5 L/min, 3 mm

Table 1: Experimental data and coefficient c d used for the simulations given in Figure 1 and Figure 2
Experiment
SINTEF
Exp 1
Exp 2
Exp 2a
Exp 2b
Exp 2c
Exp 3
Exp 4
Deepspill
Exp-DS 01

Qrel (L/min)

Drel (mm)

Vrel (m/s)

DOR

IFT (mN/s)

cd

1
1.5
1.5
1.5
1.5
5
5

1.5
1.5
1.5
1.5
1.5
2
3

9.4
14.1
14.1
14.1
14.1
26.5
11.8

1:1000
1:500
1:250
-

15.5
15.5
15.3
15.0
1.7
15.5
15.5

0.1
0.2
0.2
0.2
0.2
0.8
0.4

120

1.8

30

0.1

1.0 (m /s)

E-proceedings of the 36th IAHR World Congress,


28 June 3 July, 2015, The Hague, the Netherlands

100

100

Exp 2b
Cumulative volume percentage

Cumulative volume fraction

Exp 2a
Model
Exp

80

60

Q = 1.5 L/min
Drel = 1.5 mm
Vrel = 14.1m/s
cd = 0.2

40

DOR = 1:1000

20

0
0

50

Model
Exp

80

60

Q = 1.5 L/min
Drel = 1.5 mm
Vrel = 14.1m/s

40

cd = 0.2
DOR = 1:500

20

100 150 200 250 300 350 400 450 500 550

50

100 150

Droplet diameter (m

250 300

350

400 450

500 550

Droplet diameter (m)

100

100

Exp 2c
Model
Exp

80

60

Q = 1.5 L/min
Drel = 1.5 mm
Vrel = 14.1m/s

40

cd = 0.2
DOR = 1:250

20

0
0

50

100

150 200

250

300 350 400

Droplet diameter (m)

450

500 550

Cumulative volume percentage

Cumulative volume percentage

200

Exp 2d

Model
Exp

80

60

Q = 1.5 L/min
Drel = 1.5 mm
Vrel = 14.1m/s

40

c d = 0.2
DOR = 1:100

20

0
0

50

100 150 200

250 300 350

400 450 500

550

Droplet diameter (m)

Figure 2: Comparison of model calculated DSD (by volume) with experimental observations: SINTEF experiments with different DOR
given in Table 1: (a) DOR = 1:1000 (b) DOR =1:500 (c) DOR =1:250 (d) DOR =1:100

As shown in Figure 1, the model calculated DSD for different release rates and release conditions show a good agreement
with experimental data. In Figure 1, Exp 1 and Exp 2 show the variation in DSD with flow rate, while keeping the diameter
constant. Exp 3 and Exp 4 shows the variation of DSD with release diameter, while keeping the release rate constant. In
these cases, the model was able capture the changes in DSD and calculated the DSD with reasonable accuracy. There
are some small deviations observed in the larger droplet sizes for lower release velocity case (Fig. 1 Exp 1). The problem
here is that the experimental data are restricted to the 4.5-460 m bin sizes and with these release conditions droplets
larger than 460 m are possible. Figure 2 shows the comparison of model calculated DSD with experimental observations
for experiments with dispersants. All the dispersants cases given here are for a release rate of 1.5 L/min and 1.5 mm
diameter with the dispersants premixed with oil (Brandvik et al., 2013). The comparisons are given for different
experiments with dispersants varying from DOR (Dispersant to oil ratio) = 1:1000 to 1:100. As shown in Figure 2,
comparisons for the low dispersants cases (DOR =1:1000 and DOR =1:500) show a good agreement with the
experimental data. However, the model calculated DSD for the high dispersants cases shows some deviations. Several
factors can be identified as possible causes for these deviations in the model results. The droplet breakup and
coalescence only consider the surface tension force and neglect the viscous force inside the droplet. This is valid for most
of the droplet breakup applications, where the surface tension effect is dominant. However, when the dispersants is mixed
oil surface tension drops to a low value and the viscous forces become important. The simulations did not include the
dispersant efficiency, because the dispersants are premixed. However, the efficiency can still be less than 100 %, which
can play a key role in deciding the DSD. The viscosity of the oil can change due to the mixing of dispersants, for which we
used a constant viscosity during the simulations. Viscosity data from the experiments are not available. These factors can
affect the calculation of DSD in the model and we are working on improving the model to include those effects.

E-proceedings of the 36th IAHR World Congress


28 June 3 July, 2015, The Hague, the Netherlands

3.2 Deepspill Experiments


Deepspill is a large scale experiment conducted in the North seas to study the behavior of oil and gas in an underwater
well blowout (Johansen et al., 2001). They consist of 3 sets of large scale experiments: Exp-DS1 (Marine Diesel/LNG),
Exp-DS2 (Crude oil/LNG), and Exp-DS3 (LNG/Sea water). The bubbles/droplet sizes were measured only for experiments
DS1 and DS 3. In the present study, our focus is on the droplets sizes and the diesel droplet sizes measured in Exp DS1
was used in the comparison. In Exp-DS1 Diesel and LNG were released from a depth of 844 m for duration of 60 min. The
release rates of diesel and LNG are 1 m 3/s and 0.6 Sm 3/s respectively and the release port diameter of 0.12 m being
used. The oil density of 854.8 kg/m3 and surface tension for oil-water of 0.03 N/m is used for the simulations. The
simulations used 10 droplet size classes ranging from 1mm to 10 mm with a seed diameter of 10 mm. The seed diameter
does not have a significant effect on the final DSD. Figure 3 a and b show the comparison of model simulated DSD with
the experimental data for two different depth ranges: 840 to 839 m and 830-822 m from the sea surface, respectively.
Considering the complexity of the problem and the accuracy of the observed data, model results show a good agreement
with experimental data.

100

a (840m-839m)

b (830 m- 822 m)

80
Model
Exp

60

40

20

Cumulative volume percentage (%)

Cumulative volumePercentage (%)

100

80
Model
Exp

60

40

20

0
0

Droplet diameter (mm)

10

10

Droplet diameter (mm)

Figure 3: Comparison of cumulative volume fraction of diesel droplet sizes Exp-DS1 (Diesel and LNG)

3.3 Model Coefficient cd


In the above simulations, the release conditions vary in a wide range. The release diameter changes from 1.5 mm to 120
mm (Deeepspill) and the release velocity changes from 1.8 m/s to 26.4 m/s. These different release conditions cause
different turbulence behaviors closer to the release point. Bandara and Yapa (2011) showed that the DSD is stabilized
some distance away from the release pint. Hence, in droplet breakup and coalescence in initial stage of the jet/plume is
key in deciding the DSD. The instabilities near the release point cause the oil to break in to small droplets and turbulent
fluctuations are the main cause of further droplet breakup and coalescence. The turbulence characteristics are related with
the droplet breakup and coalescence through turbulent energy dissipation rate. Here, we calculate the energy dissipation
rate using Equation 10. In the simulations we found that the coefficient cd in equation 10 has an influence on the DSD. The
best matched DSD with experimental data was observed by varying cd values. We found the variation in cd has some
relation with the release conditions. Using the available experimental data, we correlated cd with the flow parameters. Our
first attempt was to correlate the cd with a non-dimensional parameter. However, the best relation we could found is with
the release velocity. The variation of cd values with the release velocity is shown in Figure 4. As can be seen from the
Figure 4, for higher release velocities cd shows an increasing trend whereas for lower velocities it levels out around 0.1.
Analysis of more cases is in progress to form a correlation. All the cases with dispersants correspond to a single release
velocity and therefore the cd value kept equal to the base case.

E-proceedings of the 36th IAHR World Congress,


28 June 3 July, 2015, The Hague, the Netherlands

1.5
cd
Linear fit for cd

cd

1.0

0.5

0.0

10

15

20

25

30

Release velocity (m/s)

Figure 4: Correlation of cd with release velocity

4.

SUMMARY AND CONCLUSIONS

This paper introduces an improved model to calculate the Droplet Size Distribution (DSD) in an underwater oil spill. The
model is based on the initial framework by Bandara and Yapa (2011). The new improvements allow the model to calculate
the DSD in a wide range of droplet sizes corresponding to different release conditions. The droplet size evolution in the
plume is modelled using the population balance equation. The model considers the droplet breakup and coalescence as
birth and death terms to calculate the DSD. Turbulent pressure fluctuation is considered as the dominant mechanism of
droplet breakup and coalescence due to both turbulent fluctuations and different buoyant velocities are considered in the
model. The model takes into account the stabilizing and destabilizing forces on the droplet in calculating droplet breakup
and coalescence. New theoretical formulations are used for calculating the droplet breakup rate, coalescence rate and
energy dissipation rate in the plume. The formulation is based on the isotropic turbulence theories and some predefined
probabilistic function and therefore, can be applied in different scales of problems and a wide range of conditions.
The model is validated using field and experimental data for different release conditions. The model was able to calculate
the DSD for very small droplet sizes corresponding to small diameter and high release velocity cases as well as relatively
large droplet sizes corresponding to small release velocity and large release diameter cases. During these simulations we
found that for high release velocity cases the model coefficient cd varies with the release velocity whereas for relatively
small velocities it remains constant. However, this correlation is obtained with only a limited data available and need to be
further strengthened with more experimental data. The model is also used to simulate a few experiments which used
dispersants. The model shows good comparisons for low dispersants cases. However, for high dispersants cases model
calculations deviate from the experimental data. Not accounting for the viscous forces and the change in the viscosity due
to adding dispersants is identified as the causes for those deviations and we are working on those to improve the model.

E-proceedings of the 36th IAHR World Congress


28 June 3 July, 2015, The Hague, the Netherlands

REFERENCES
Bandara, U.C., Yapa, P.D., 2011. Bubble Sizes, Breakup, and Coalescence in Deepwater Gas/Oil Plumes. J. Hydraul.
Eng. 137, 729738. doi:10.1061/(ASCE)HY.1943-7900.0000380
Bizmark, N., Mostoufi, N., Mehrnia, M.-R., Zarringhalam, S.M., Yazdani, A., 2012. Coalescence efficiency of bubbles in
bubble columns. Can. J. Chem. Eng. 90, 15791587. doi:10.1002/cjce.20664
Brandvik, P.J., Johansen, ., Farooq, U., Glen, A., Leirvik, F., 2014. Subsurface oil releases Experimental study of
droplet distributions and different dispersant injection techniques Version 2 (Technical report No. SINTEF
A26122). - SINTEF.
Brandvik, P.J., Johansen, ., Leirvik, F., Farooq, U., Daling, P.S., 2013. Droplet breakup in subsurface oil releases Part
1: Experimental study of droplet breakup and effectiveness of dispersant injection. Mar. Pollut. Bull. 73, 319326.
doi:10.1016/j.marpolbul.2013.05.020
Chen, F., Yapa, P., 2007. Estimating the Oil Droplet Size Distributions in Deepwater Oil Spills. J. Hydraul. Eng. 133, 197
207. doi:10.1061/(ASCE)0733-9429(2007)133:2(197)
Chen, P., Sanyal, J., Dudukovi, M.P., 2005. Numerical simulation of bubble columns flows: effect of different breakup and
coalescence closures. Chem. Eng. Sci. 60, 10851101. doi:10.1016/j.ces.2004.09.070
Coulaloglou, C.A., Tavlarides, L.L., 1977. Description of interaction processes in agitated liquid-liquid dispersions. Chem.
Eng. Sci. 32, 12891297. doi:10.1016/0009-2509(77)85023-9
Hagesaether, L., Jakobsen, H.A., Svendsen, H.F., 2002. A model for turbulent binary breakup of dispersed fluid particles.
Chem. Eng. Sci., Jean-Claude Charpentier Festschrift Issue 57, 32513267. doi:10.1016/S0009-2509(02)001975
Hesketh, R.P., Etchells, A.W., Russell, T.W.F., 1991. Experimental observations of bubble breakage in turbulent flow. Ind.
Eng. Chem. Res. 30, 835841. doi:10.1021/ie00053a005
Hibiki, T., Ishii, M., 2002. Development of one-group interfacial area transport equation in bubbly flow systems. Int. J. Heat
Mass Transf. 45, 23512372. doi:10.1016/S0017-9310(01)00327-1
Hinze, J.O., 1955. Fundamentals of the hydrodynamic mechanism of splitting in dispersion processes. AIChE J. 1, 289
295. doi:10.1002/aic.690010303
Johansen, ., 2000. DeepBlow a Lagrangian Plume Model for Deep Water Blowouts. Spill Sci. Amp Technol. Bull. 103
111. doi:10.1016/S1353-2561(00)00042-6
Johansen, ., Brandvik, P.J., Farooq, U., 2013. Droplet breakup in subsea oil releases Part 2: Predictions of droplet size
distributions with and without injection of chemical dispersants. Mar. Pollut. Bull. 73, 327335.
doi:10.1016/j.marpolbul.2013.04.012
Johansen, ., Ray, H.M.A.G., Jensen, S.H.S., Knutsen, T., 2001. Deep spill jip-experimental discharges of gas and oil at
helland hansen-June 2000 (Technical report No. 661182). SINTEF Applied Chemistry, Trondheim, Norway.
Kumar, S., Ramkrishna, D., 1996. On the solution of population balance equations by discretizationII. A moving pivot
technique. Chem. Eng. Sci. 51, 13331342. doi:10.1016/0009-2509(95)00355-X
Lehr, F., Millies, M., Mewes, D., 2002. Bubble-Size distributions and flow fields in bubble columns. AIChE J. 48, 2426
2443. doi:10.1002/aic.690481103
Liao, Y., Lucas, D., 2010. A literature review on mechanisms and models for the coalescence process of fluid particles.
Chem. Eng. Sci. 65, 28512864. doi:10.1016/j.ces.2010.02.020
Luo, H., Svendsen, H.F., 1996. Theoretical model for drop and bubble breakup in turbulent dispersions. AIChE J. 42,
12251233. doi:10.1002/aic.690420505
Luo, H., Svendsen, H.F., 1996. Modeling and Simulation of Binary Approach by Energy Conservation Analysis. Chem.
Eng. Commun. 145, 145153. doi:10.1080/00986449608936473
Niu, H., Li, Z., Lee, K., Reed, M., Mullin, J., 2011. Sensitivity of a deepwater blowout model on oil droplet size distribution.
Proc. Thirty-Fourth AMOP Tech. Semin. Environ. Contam. Response Banff Alta. Can. 189200.
Prince, M.J., Blanch, H.W., 1990. Bubble coalescence and break-up in air-sparged bubble columns. AIChE J. 36, 1485
1499. doi:10.1002/aic.690361004
Tsouris, C., Tavlarides, L.L., 1994. Breakage and coalescence models for drops in turbulent dispersions. AIChE J. 40,
395406. doi:10.1002/aic.690400303
Wang, T., Wang, J., Jin, Y., 2003. A novel theoretical breakup kernel function for bubbles/droplets in a turbulent flow.
Chem. Eng. Sci. 58, 46294637. doi:10.1016/j.ces.2003.07.009
Wang, T., Wang, J., Jin, Y., 2005a. Population Balance Model for GasLiquid Flows: Influence of Bubble Coalescence
and Breakup Models. Ind. Eng. Chem. Res. 44, 75407549. doi:10.1021/ie0489002
Wang, T., Wang, J., Jin, Y., 2005b. Theoretical prediction of flow regime transition in bubble columns by the population
balance model. Chem. Eng. Sci., 7th International Conference on Gas-Liquid and Gas-Liquid-Solid Reactor
Engineering 7th International Conference on Gas-Liquid and Gas-Liquid-Solid Reactor Engineering 60, 6199
6209. doi:10.1016/j.ces.2005.04.027
Zhao, L., Boufadel, M.C., Socolofsky, S.A., Adams, E., King, T., Lee, K., 2014. Evolution of droplets in subsea oil and gas
blowouts: Development and validation of the numerical model VDROP-J. Mar. Pollut. Bull. 83, 5869.
doi:10.1016/j.marpolbul.2014.04.020
Zheng, L., Yapa, P., 2000. Buoyant Velocity of Spherical and Nonspherical Bubbles/Droplets. J. Hydraul. Eng. 126, 852
854. doi:10.1061/(ASCE)0733-9429(2000)126:11(852)
Zheng, L., Yapa, P.D., Chen, F., 2003. A model for simulating deepwater oil and gas blowouts - Part I: Theory and model
formulation. J. Hydraul. Res. 41, 339351. doi:10.1080/00221680309499980

You might also like