Energy 64 (2014) 887e899

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

A CFD (computational uid dynamic) simulation for oil leakage from

damaged submarine pipeline
Hongjun Zhu a, b, *, Pengzhi Lin b, Qian Pan a
a
b

State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation, Southwest Petroleum University, Chengdu, Sichuan 610500, China
State Key Laboratory of Hydraulics and Mountain River Engineering, Sichuan University, Chengdu, Sichuan 610065, China

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 19 June 2013
Received in revised form
10 October 2013
Accepted 14 October 2013
Available online 7 November 2013

The objective of the present paper is to study the oil ows from damaged submarine pipelines with
different leak sizes. CFD (computational uid dynamic) simulations with FLUENT software are carried out
to investigate the process of oil spill from submarine pipeline to free surface. Effects of oil density, oil
leaking rate, leak size and water velocity on the oil spill process are examined. High density, slow leaking,
small leak size or fast current brings about long time for oil reaching the maximum horizontal migrate
distance when it reaches surface. And this maximum horizontal migrate distance increases with the
increase of leak size or water velocity, while increases with the decrease of leaking rate. Then, the
dimensionless time required for oil droplets which have the longest horizontal migrate distance when
they reach the sea surface and the dimensionless longest horizontal distance the droplets migrate when
they reach the sea surface are analyzed and the tting formulas are obtained. Only the formula for the
dimensionless longest horizontal distance versus dimensionless density meets the polynomial, other ve
formulas meet the natural logarithm distribution. Using the formulas we can obtain when and where to
see oil reaching the sea surface, and conduct rapid response. Finally, the maximum horizontal migration
distance of oil at certain time is predicted, and a forecasting model is proposed. The using methods of
tting formulas and the forecasting model are shown in the paper by examples. These calculated results
provide useful guidance to place the oil containment boom.
2013 Elsevier Ltd. All rights reserved.

Keywords:
Oil spill
Submarine pipeline
VOF (volume of uid)
CFD (computational uid dynamics)

1. Introduction
Washout and perforation failures are usually present in oil
submarine pipelines due to corrosion or ow erosion. Then oil spills
into marine environment from the leak, causing extensive damage
to marine life, human health, and natural resources [1]. As the
accidental oil leakages in Gulf of Mexico, about twenty thousand
barrels of oil were released into the sea everyday, resulting in
ecological devastation in adjacent sea area. Similar oil spill accidents occurred at platform B and C of the Penglai 19-3 oileld
located in Bohai Sea [2]. About 700 barrels of oil and 2500 barrels of
mineral oil-based drilling mud were released. Moreover, Bohai Sea
is a semi-enclosed shallow sea with average water depth of 18 m,
making the risk relatively high. Given the frequent occurrence of oil
spills, it has been a matter of constant concern from the viewpoints
of environmental and social disasters.

* Corresponding author. State Key Laboratory of Oil and Gas Reservoir Geology
and Exploitation, Southwest Petroleum University, Chengdu, Sichuan 610500,
China. Tel.: 86 28 83032206.
E-mail address: ticky863@126.com (H. Zhu).
0360-5442/$ e see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.energy.2013.10.037

Once accidental oil leakages occur, a quick and adequate

response in order to reduce the environmental consequences is
required [3]. By this time, the information about the rise process of
oil droplets and the dispersal path of oil spill under the action of
current is particularly necessary. Besides, laying oil containment
boom, as a basic way to control oil dispersal, also depends on the
rising velocity of oil droplets and the trend of spreading. Therefore,
an exact prediction of oil spill process and dispersal could provide
useful information for setting up oil containment boom and
reducing the damage of future oil spills.
Most previous studies focused on the surface two-dimensional
oil spill [4e6], in which ROMS (regional ocean modeling system)
is employed widely. However, for oil leakage from damaged submarine pipeline, the migration of oil ow along the depth direction
becomes an important issue. An effective attempt has been made
by Li et al. [7] to observe the oil spill under the action of current and
wave. However, the velocity of current in their study was uniform,
which does not match with the actual shear velocity distribution
under sea surface. And the actual hydrostatic pressure distribution
was not used in their modeling. Moreover, the crucial parameter,
the maximum horizontal migration distance of oil, was not
considered in their research.

888

H. Zhu et al. / Energy 64 (2014) 887e899

Without the effect of wave or current, a nearly vertical ascent

of spilled oil droplets would present, and the diffusion radius of oil
lm in the surface is just a function of time. Where to place the oil
containment boom can be easily obtained in this condition.
However, there are waves and currents in ocean all the time.
Under the action of wave and current, the trajectory of spilled oil
runs off the straight line, and oil lm in the surface spreads
rapidly downstream. Therefore, the maximum horizontal migration distance of oil is a key parameter to place the oil containment
boom.
Under the action of current with shear velocity distribution,
the length of time for oil to reach sea surface, and the distance for
oil moving downstream when it reaches the surface are the two
key parameters to guide quick response, including the laying of oil
containment boom. Especially for submarine pipelines near production platforms or shore, once oil spill is observed by ROV
(remote operated vehicle) or acoustic detection, a quick rescue
response can be implemented in a few minutes, even in 1 min. In
this case, the process of oil spill from submarine pipeline to free
surface should be concerned to nd when and where the spilled
oil droplets can reach the free surface. While for submarine
pipelines far away from platforms and shore, the rescue ships
usually take hours to reach the accident location. Thus, the
maximum horizontal migration distance of oil at certain time, the
hours for rescue ships arrival after leakage started, is the major
concern. For the process of oil spill consisting of two processes,
rising process and drifting process, the maximum horizontal
migration distance is still related to the longest horizontal distance the droplets migrate when they reach the sea surface. After
the oil droplets arriving at surface, the horizontal migration distance is a function of surface water velocity. So the process of oil
spill from submarine pipeline to free surface is still needed to be
solved rstly.
Numerical simulation can provide detailed information on the
hydrodynamics of oil ow, which is not easily obtained by physical
experiments [8e10]. Therefore, in this work, CFD (computational
uid dynamic) model coupling with VOF (volume of uid) method
has been used to investigate the process of oil spill from submarine
pipeline to free surface. The actual shear velocity distribution of
current and the actual hydrostatic pressure distribution are
considered in this study. Detailed oil droplet and sea-surface information could be obtained by the VOF model. By conducting a
series of numerical simulations, effects of oil density, oil leaking
rate, leak size and water velocity on the oil spill process are
examined. Then, the dimensionless time required for oil droplets
which have the longest horizontal migrate distance when they
reach the sea surface and the dimensionless longest horizontal
distance the droplets migrate when they reach the sea surface are
analyzed and the tting formulas are obtained. Using the formulas
we can obtain when and where to see oil reaching the sea surface,
and conduct rapid response. Finally, the maximum horizontal
migration distance of oil at certain time is predicted, and a forecasting model is proposed. The results provide useful guidance to
place the oil containment boom.
The remaining part of this paper is organized as follows. In
Section 2, the description of mathematical models and simulation
method are provided; Section 3 presents the simulated results and
discussion; Section 4 is the concluding remarks.
2. Simulation method
2.1. Governing equations
Oil, water and air are treated as incompressible ows. And at
the interface of uids, no phase change and no-slip between uids

are assumed. The ows of uids are governed by the RANS (Reynolds-Averaged-Navier-Stokes) equations, including continuity and
momentum equations written as follows [11e13]:

vui
0
vxi

(1)
0 0

vui uj
vui vui uj
1 vp


y V2 u i 
gi
r vxi
vt
vxj
vxj

(2)

where ui represents instantaneous velocity component in i direction, for example u and v are velocity in x and y direction, respectively, while u0i is uctuation velocity component in i direction, xi is
space coordinate in i direction, gi is gravitational acceleration in i
direction, t is time, p is pressure and r and y are density and kinematic viscosity, respectively.
The VOF approach is based on the solution of one momentum equation for the mixture of the phases, and one
equation for the volume fraction of uid. In this study, volume
of uid functions Fw and Fo are introduced to dene the water
region and the oil region, respectively. The physical meaning of
the F function is the fractional volume of a cell occupied by the
liquid phase [14,15]. For example, a unit vale of Fw corresponds
to a cell full of water, while a zero value indicates that the cell
contains no water. The fraction functions Fw and Fo are described
as follows:

Fw

Vw
Vc

(3)

Fo

Vo
Vc

(4)

where Fo and Fw are oil and water fractional function, respectively,

Vc, Vo and Vw represent volume of a cell, volume of oil inside the cell
and volume of water inside the cell, respectively.
And the two-dimensional transport equations for the fractional
functions are given by:

vFw vuFw vvFw

0
vt
vx
vy

(5)

vFo vuFo vvFo

0
vt
vx
vy

(6)

Then, the density and viscosity can be expressed in following

equations:

r 1  Fw  Fo ra Fw rw Fo ro

(7)

y 1  Fw  Fo ya Fw yw Fo yo

(8)

where subscript a, o and w represent air, oil and water, respectively.

In this study, Reynold number ranges from 181.25 to 4531.25. So
in condition of high Reynold number, realizable k- turbulence
model [16e20] is employed to close the ow governing equations
and describe the turbulent properties:

vrk vrkui
v

vt
vxi
vxj

"

mt vk
Gk Gb  r
sk vxj

(9)

H. Zhu et al. / Energy 64 (2014) 887e899

vr vrui
v

vt
vxi
vxj

"

mt v
2
p
rC1 S  rC2
s vxj
k y

C1 1  C3 Gb
k
(10)
where,


C1 max 0:43;

h S

h5

(12)


1=2
S 2Sij $Sij

Sij

1 vui vuj

2 vxj vxi

Gk ru0i u0j

(13)
!

vuj
vxi

(14)

(15)

m t vr
Prt rvxi

(16)

k2

(17)

Gb gi

mt rCm

(11)

where k and represent turbulent kinetic energy and turbulent

kinetic energy dissipation rate per unit mass, respectively, h is the
relative strain parameter, S is the strain rate, Gk and Gb represent
production term of turbulent kinetic energy due to the average
velocity gradient and production term of turbulent kinetic energy
due to lift, respectively, m ry is dynamic viscosity of uid, mt is
turbulent viscosity, Prt is Prandtl number taken as 0.85, Cm, C1, C2
and C3 are empirical model constants taken as 0.09, 1.44, 1.9 and
0.9, respectively, and st and s are turbulent Prandtl numbers taken
as 1.0 and 1.2, respectively.
2.2. Numerical method
FVM (nite volume method) is employed to discretize above
equations. All the simulations are carried out using a commercial
software package FLUENT 14.0. In calculations, Patankars wellknown SIMPLE algorithm [21], applied well in many similar simulation studies [22e24], is employed to solve the pressureevelocity
coupling to satisfy the conservation law of mass. In order to ensure
the accuracy of calculation, second-order upwind scheme and
second-order central-differencing scheme are used for convective
terms and diffusion terms, respectively. For convective terms,
second-order upwind scheme can save CPU time [25], while for
diffusion terms second-order central-differencing scheme is a good
choice [26]. The convergent criteria for all calculations are set as
that the residual in the control volume for each equation is smaller
than 104.
2.3. Computational domain and mesh
Under the action of current, spilled oil spreading directly
downstream can reach the maximum horizontal migration distance. Therefore, two-dimensional ow simulation is accurate
enough to capture the maximum horizontal migration distance. In

889

addition, three-dimensional simulation needs a higher CPU cost.

Due to time limitations, 2D simulation is applied in this work.
Fig. 1 shows a sketch of the geometry and numerical grid for
computational domain investigated in this study. The average
depth of water (14.5 m) in Kenli oileld located in Bohai Sea is taken
as the model depth in order to facilitate the comparison. The whole
computational domain is a rectangle with a length of 20 m and a
height of 15 m. The length of computational domain is large
enough, which is larger than the longest horizontal distance the oil
droplets migrate when they reach the sea surface. Water occupies
the lower region with height of 14.5 m, while air occupies the upper
region.
In the computational domain, the damaged submarine pipe
with the outer diameter (D) of 0.6 m, the most common
diameter of submarine pipe used in Bohai oileld, is located in
the sea bed, 1.8 m (3D) downstream of the inlet. There is a
leakage hole on the top of pipe, opening upwards. The size of
the leakage hole (d) is a variable ranging from 0.01 m to 0.05 m
with increment of 0.01 m, in order to examine the effect of leak
size.
GAMBIT 2.3 mesh-generator is employed to perform all geometry generation and meshing. As shown in Fig. 1, computational
domain is divided into two blocks. The water occupying region is
discretized with triangular cells, while the upper region is discretized with quadrilateral cells. Progressive mesh is used to capture the near-leak ow properties. A suitable grid density is
reached by repeating computations until a satisfactory independent grid is found. At last, the number of grid cells used in calculation is 9011.
2.4. Boundary conditions
A logarithmic velocity prole is adopted to meet the actual ow
near seabed as vw vwmax{1  [1  y/(H D)]2}, where vwmax is the
maximum velocity (vwmax 0.1 m/s, the most commonly measured
maximum ow rate in Bohai Sea surface) presenting at the free
surface, H is the height from leakage hole to the free surface, y is the
independent variable (0  y  H D) and origin of coordinates is
located in the seabed. This velocity prole is dened for the left
inlet of computational domain. In order to nd the effect of water
velocity on the displacement of oil droplets, the maximum water
velocities are taken as 0.04 m/s and 0.07 m/s in comparing
cases. For the right outlet, a linear static pressure prole is
employed to meet the actual hydrostatic pressure distribution as
pout rg(H D  y).
In Bohai oil eld, due to corrosion or ow erosion, perforations
usually present in submarine pipelines. Only a few pipes have
cracks on them due to mechanical damage. According to a lot of
failure marine pipe tests, the perforation holes mainly show circular shape or roughly circular shape. For roughly circular shape,
we can use a circular instead of it with an equivalent diameter (the
two have the same equal area). By measuring a large number of
perforation holes, the pore sizes mainly range from 0.006 m to
0.08 m. For analyzing the effect of pore size, ve different leak diameters, 0.01 m, 0.02 m, 0.03 m, 0.04 m and 0.05 m, are chosen in
simulations.
Oil leaking rate is related to leak size, hydrostatic pressure of
water above the pipe, the pressure within the pipe and pressure
drop of oil owing through the leak. Therefore, perforations in
different oil pipelines or in different locations at one oil pipeline
have the different oil leaking rates. In the Bohai Sea environment,
oil leaking rate usually ranges from 0.1 m/s to 10 m/s. In order to
facilitate comparative analysis, we have selected ve rates ranging
from 1 m/s to 5 m/s to conduct simulations. Pressure inlet boundary
condition with value of 0 Pa is used for the three edges of air region.

890

H. Zhu et al. / Energy 64 (2014) 887e899

Fig. 1. Sketch of the geometry and numerical grid for computational domain: (a) overall view of the computational domain and boundary conditions; (b) grid distribution of
computational domain.

At the initial time, the lower area is lled with water and the
upper area is lled with air. Still water surface is assumed at the
interface of the two regions. And the pressure in air region is
dened as 0 Pa, meaning the atmospheric pressure.
In simulations, densities of air and water are seemed constant as
1.225 kg/m3 and 1025 kg/m3, respectively. While the density of oil

is a variable in different case ranging from 780 kg/m3 to 960 kg/m3

with increment of 30 kg/m3, in order to analyze the effect of oil
density. The viscosities of air, oil and water are dened as
1.8  105 Pa$s, 1.003  103 Pa$s and 0.048 Pa$s, respectively.
The information of simulation cases is listed in Table 1, in which
oil density, oil leaking rate, diameter of leak and the maximum

Table 1
Simulation cases.
Case

Oil density
(kg/m3)

The maximum
water velocity
(m/s)

Oil leaking
rate (m/s)

Diameter of
leak (m)

Volume ux
of leaking oil
(m3/s)

Flux multiple
(comparing
with case 12)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

780
810
840
870
900
930
960
870
870
870
870
870
870
870
870
870
870

0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.04
0.07

2
2
2
2
2
2
2
1
3
4
5
2
2
2
2
2
2

0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.01
0.02
0.03
0.04
0.05
0.05

0.003925
0.003925
0.003925
0.003925
0.003925
0.003925
0.003925
0.0019625
0.0058875
0.00785
0.0098125
0.000157
0.000628
0.001413
0.002512
0.003925
0.003925

25
25
25
25
25
25
25
12.5
37.5
50
62.5
1
4
9
16
25
25

H. Zhu et al. / Energy 64 (2014) 887e899

891

Fig. 2. The process of oil spill from submarine pipeline to free surface at standard case.

water velocity are the four variables. For ow rate and leak size are
related to each other, we have set the minimum volume ux occurs
in case 12 (oil leaking rate 2 m/s and leak size 0.01 m). And
volume uxes in others cases are some times more than the minimum volume ux, as listed in Table 1. Therefore, a wide range of
volume ux of leaking oil has been considered in this paper.
3. Numerical results and discussion
3.1. Standard case
We adopt case 4 (as shown in Table 1) as the standard case. Fig. 2
presents the volume fraction of water at different times, corresponding to the process of oil spill from submarine pipeline to free
surface. It is indicated 39 s are required for oil droplets which have the
longest horizontal migrate distance when they reach the sea surface.
As Fig. 2 shows, continuous oil ow stream presents just
releasing from the leakage hole with the height of about 3 m.
However, it is tore apart at a certain depth by current. Under the
joint action of gravity, inertia force, buoyancy and shear stress, oil
spill appears in the form of droplets or droplet groups in most of the
computational region. With the increase in rising height, oil
droplets become more dispersed. At t 39 s, the horizontal span of
oil droplets is about 3.36 times longer than that at t 15 s.
Oil droplets move downstream under the action of water
owing from the left boundary. However, this phenomenon is not

obvious near the seabed for the small water velocity. With the
decrease in water depth, the horizontal migrate distance has a
signicant increase. The reason is that high-speed water has a
greater horizontal shear stress exerting on spilled oil and transfers
greater kinetic energy to oil ow, resulting in a longer horizontal
migrate. Therefore, the shear distribution current plays an important role in oil migration. We should consider the actual water
velocity distribution to conduct a relative accurate prediction.
From the Fig. 2 we can see that the rst oil droplet reaching
the free surface is not the one which has the longest horizontal
migrate distance. This main reason is that oil droplets have
different sizes and the sizes are variable with time due to the
dispersion and combining of oil droplets under the joint action of
gravity, inertia force, buoyancy and shear stress. In the process of
oil spill from the leak to free surface, the oil droplet which reaches the free surface rstly has larger buoyancy, and then the
furthest-migration droplet receives a greater shearing action of
current. The time required for the rst oil droplet reaching the
free surface is less than 35 s. While the droplet which has the
longest horizontal migrate distance reaches the surface at
t 39 s. And the longest horizontal migrate distance is 16.7 m
(18.5 m  1.8 m, the horizontal distance from the inlet to pipe
leak). At this time, the horizontal and vertical displacement ratio
of this droplet is 1.201.
Since the oil spill is subjected to the joint action of gravity,
inertia force, buoyancy and shear stress, in the later stage, we have

892

H. Zhu et al. / Energy 64 (2014) 887e899

Fig. 3. The process of oil spill from submarine pipeline to free surface at different oil densities: (a) ro 780 kg/m3; (b) ro 870 kg/m3; (c) ro 960 kg/m3.

H. Zhu et al. / Energy 64 (2014) 887e899

893

Fig. 4. The process of oil spill from submarine pipeline to free surface at different oil leaking rates: (a) vo 1 m/s; (b) vo 3 m/s; (c) vo 5 m/s.

changed the oil density, oil leaking rate, leaking size and water
velocity to explore their inuences.
3.2. Effect of oil density
In this section, to study the effect of oil density on the length of
time for oil to reach the sea-surface and the distance for oil moving
downstream, simulations are conducted by changing the oil density
while leaving other parameters same as those in the standard case.
Fig. 3 illustrates the process of oil spill from submarine pipeline to
free surface at three different oil densities. It can be seen that the
larger the oil density, the longer the time required for oil to reach
free surface. For ro 960 kg/m3, the required time for the
maximum horizontal migration is about 1.84 times as long as that

when oil density is 780 kg/m3. It is attributed to the increasing

gravity of oil droplets. In the vertical direction, an oil droplet is
mainly subject to the force of gravity and buoyancy. For two
droplets of the same size, the upward buoyant forces are the same,
while the droplet of larger density has a larger gravity. Therefore,
the nal vertical upward force is small for high density droplet,
resulting in a slow rising rate.
At the same time (t 15 s), the maximum horizontal migration
distance of light oil droplet (ro 780 kg/m3) is 7 m (8.8 m minus
1.8 m), about one time longer than that with density of 960 kg/m3.
Before the leaking oil reaching the sea-surface, the horizontal
migration of oil ow under the sea-surface can not be easily
observed without monitoring instruments. In addition, oil containment boom is laid oating on the sea-surface. So the maximum

894

H. Zhu et al. / Energy 64 (2014) 887e899

Fig. 5. The process of oil spill from submarine pipeline to free surface at different water velocities: (a) vwmax 0.04 m/s; (b) vwmax 0.07 m/s; (c) vwmax 0.1 m/s.

horizontal migration distance when the oil droplet reaches the free
surface is a very vital parameter. This horizontal migration distance
for ro 960 kg/m3 is 17.1 m (18.9 m  1.8 m), a little shorter than that
when oil density is 780 kg/m3 which is 18 m (19.8 m  1.8 m). The

main cause of this result is that low density oil droplets rise faster
and enter into high-speed water zone earlier, leading to shearing
action of current acting on oil earlier. However, the difference in the
maximum horizontal migration distance is little. Therefore, light oil

H. Zhu et al. / Energy 64 (2014) 887e899

Fig. 6. The process of oil spill from submarine pipeline to free surface at different oil leak sizes: (a) d 0.01 m; (b) d 0.03 m; (c) d 0.05 m.

895

896

H. Zhu et al. / Energy 64 (2014) 887e899

can reach surface quickly, requiring short response times, while the
location be laid with oil containment boom to control oil dispersal is
basically the same for different-density oil ow.
3.3. Effect of oil leaking rate
Oil leaking rate is one of the key factors which impacts the
diffusion of oil spill. Fig. 4 depicts the process of oil spill from
submarine pipeline to free surface at different oil leaking rates. At
small leaking rate (vo 1 m/s), 79 s is required for oil ow to reach
the maximum horizontal migrate distance when it reaches the free
surface, and the maximum horizontal migration distance is arrived
at 16.4 m (18.2 m minus 1.8 m). However, for higher leaking rate
(vo 5 m/s), just 15 s is needed for oil to reach the maximum
horizontal migrate distance when it reaches the free surface. It can
be explained that high-speed leaking oil has more ascending kinetic energy. From Fig. 4, we can also see more dispersed oil
droplets present in computational domain at high leaking rate. The
reason is that the total amount of released oil is larger as the mass
rate of oil is larger (For incompressible uid, mass ow rate increases as the increase in velocity). Thus, in order to reduce the
environmental consequences, a relatively fast response is required
for high-speed leaking oil, and an adequate response should be
considered to solve a large number of oil spills.
3.4. Effect of water velocity
Current as a carrier plays a crucial role in the migration of the oil
ow. Therefore, in this section, we have changed the water velocity
to nd its effect on oil spill. The processes of oil spill from submarine pipeline to free surface at different water velocities are shown
in Fig. 5. The larger the water velocity, the more obvious the trajectory of oil ow skewed to the downstream. The reason is that
high-speed water exerts more shear stress on oil droplets and
transfers more kinetic energy to oil droplets. The maximum horizontal migrate distance for vwmax 0.04 m/s is 14.2 m (16 m minus
1.8 m), 0.9 m less than that for vwmax 0.07 m/s and 2.5 m less than
that for vwmax 0.1 m/s.

The time required for oil ow to reach the maximum horizontal

migrate distance when it reaches the free surface is 36 s when the
maximum water velocity is 0.04 m/s, 2 s earlier than that for
vwmax 0.07 m/s and 3 s earlier than that for vwmax 0.1 m/s. The
route of oil droplet is the square root of the sum of the squares on
vertical depth and horizontal displacement. Therefore, the route of
oil droplet is relative small in vwmax 0.04 m/s due to the short
horizontal displacement, resulting in the short rising time. Therefore, short response time is required for slow current, while oil
containment boom should be laid a longer distance for fast current.
3.5. Effect of oil leak size
Fig. 6 shows the process of oil spill from submarine pipeline to
free surface at different oil leakage sizes. The results indicate that
the effect of the diameter of leakage hole plays a signicant role in
the spread of oil spill. With increasing leakage size, the time
required for oil to reach the maximum horizontal migrate distance
when it reaches the free surface is shortened. As leakage size reduces from 0.05 m to 0.01 m, the required time decreased by 23.53
percents. It can be explained that at the same leaking rate, the
bigger the diameter of leak, the larger the amount of released oil
and the greater the upward momentum. Due to the large mass ow
rate, oil droplets released from the leak with d 0.05 m are easier
to collision and have greater chance of gathering into large droplets,
as shown in Fig. 6. Though the water velocities are the same, large
active faces of big oil droplets lead to great shear stress. Under the
action of shear stress, the maximum horizontal migrate distance,
16.7 m (18.5 m minus 1.8 m), presents in the case of d 0.05 m. This
distance is about 1.5 times than the maximum horizontal migrate
distance for d 0.01 m. Therefore, big-hole leaks may lead to more
serious consequences.
3.6. Dimensionless analysis
As shown in Fig. 7, the dimensionless time required for oil
droplets which have the longest horizontal migrate distance when
they reach the sea surface is analyzed and the tting formulas are
obtained. We can clearly see that the larger the density of oil, the

Fig. 7. Dimensionless time required for oil droplets which have the longest horizontal migrate distance when they reach the sea surface (vot/H) versus ro/rw, 10vwmax/vo and 100 d/H.

H. Zhu et al. / Energy 64 (2014) 887e899

897

slower the oil leaking, or the smaller the leak size, the longer the
dimensionless oil spill time is. All the three parameters meet the
natural logarithm distribution well, and the adjust R-squares are all
larger than 0.98. The three tting formulas are as follows:

10vwmax/vo and 100 d/H, they meet the natural logarithm distribution well, and the adjust R-squares are all larger than 0.99. While
for ro/rw, polynomial is the most appropriate one. These three
tting formulas are as follows:



r
vo t
1
6:14146  106 exp o $
4:2567
rw 0:07163
H

(18)

 2
r
r
Lf
194:556 o 107:6361 o
118:7261
rw
rw
D

(22)



vo t
10vwmax
1
3:40596 exp 
5:68636
$
H
0:13146
vo

(19)



Lf
10vwmax
1
30:4548
19:738 exp 
$
0:25236
D
vo

(23)



Lf
100d
1
$
31:1406
21:4234 exp 
H
0:09199
D

(24)



vo t
100d
1
3:41601 exp 
$
5:43198
H
H
0:12289

(20)

where t is the time required for oil droplets which have the longest
horizontal migrate distance when they reach the sea surface.
Taking Eq. (18) for example, if the actual parameters are the
same as that used in this paper such as H 13.9 m and
rw 1025 kg/m3, this equation can be simplied as:

t 4:2683  105 exp0:014ro 29:584

(21)

Then we can select an arbitrary oil density substituted into the

formula to calculate the required time. For example, if the oil
density is set as 850 kg/m3, the time would be calculated as 35.86 s.
Using the same approach, we can calculate the required times for
different oil leaking rates, water velocities and leak sizes based on
the Eq. (19) and Eq. (20).
Fig. 8 shows the dimensionless longest horizontal distance the
droplets migrate when they reach the sea surface. The dimensionless longest horizontal distance increases with the increase of
leak size, while increases with the decrease of oil leaking rate. With
the increase in oil density, the dimensionless longest horizontal
distance decreases rstly and then increases. But the value does not
change signicantly. Adopting the same method as above, we have
obtained three tting formulas about the dimensionless longest
horizontal distance versus ro/rw, 10vwmax/vo and 100 d/H. For

where Lf represents the maximum horizontal migrate distance

when oil droplets reach the free surface.
Here we take Eq. (22) for example, if the actual parameters are
the same as that used in simulation, this equation can be simplied
as:

Lf 0:1169ro 6:48  105 r2o 71:236

(25)

Based on Eq. (25), if the oil density is set as 850 kg/m3, the
maximum horizontal migrate distance when oil reaches the free
surface can be calculated as 18.69 m. For different oil leaking rates,
water velocities or leak sizes, we can also predict the maximum
horizontal migrate distance by solving Eq. (23) and Eq. (24). The
above tting equations may provide some useful information for
government and oil business to adopt rapid response in case oil
spill occurs.
3.7. Prediction of the maximum horizontal migration distance
In most cases, the leak in submarine pipeline is far away from
platforms and shore. Thus, a few hours would be taken for rescue
ships to reach the accident location. When the rescue ships arrive,
how long the horizontal migration distance is the major concern. In

Fig. 8. The dimensionless longest horizontal distance the droplets migrate when they reach the sea surface (Lf/D) versus ro/rw, 10vwmax/vo and 100 d/H.

898

H. Zhu et al. / Energy 64 (2014) 887e899

order to prevent further oil spilling, the maximum horizontal

migration distance of oil at certain time should cover all oil droplets, then the location is a proper place to lay oil containment boom.
As shown in Fig. 9, after oil droplets reaching the free surface,
they will drift downward with current. So the process of oil spill
consists of rising process and drifting process. The rising process
has been analyzed in detail in the above, and we can predict when
and where the spilled oil droplets reach the maximum horizontal
migration distance when they reach the free surface. For drifting
process, the main motion of oil droplets is moving downstream
along the free surface. In this process, water is acting as a carrier.
Therefore, the horizontal displacement of oil at free surface is a
function of surface water velocity. Although the arrival of oil
droplets causes the slight undulation of sea surface, it has a little
effect on the horizontal displacement of oil droplets. Even though
its effect is considered, the undulation of sea surface just plays as
obstacle for oil migration. Therefore, the following equation can be
used to predict the maximum horizontal migration distance of oil
at certain time:

L Lf atr  tvwmax

(26)

where L represents the maximum horizontal migrate distance of oil

at certain time, tr is the certain time and a is a coefcient, which can
be taken as 1.0e1.2. A larger a corresponds to a longer migration
distance, which can cover all oil droplets. Since the undulation of
sea surface and collision of oil droplets exert resistance on oil
migrate, the actual value of a may less than 1. However, we take a as
1.0e1.2 to ensure the prediction more secure and reliable.
Lets take the standard case for example. If 2 h are needed for
rescue ships to reach the accident location, the maximum horizontal
migration distance of oil when rescue ships arrive is 876.02 m,
which is calculated based on Eq. (26) where a is taken as 1.2.
4. Conclusions
An approach, for predicting the process of oil spill under the
action of current with shear velocity distribution, by nite volume
simulation combined with VOF method is proposed. Effects of oil
density, oil leaking rate, leak size and water velocity are examined.
And tting formulas are obtained to predict when and where to see
oil reaching the sea surface. Finally, a forecasting model of the
maximum horizontal migration distance of oil at certain time is
proposed. These results provide useful guidance to place the oil
containment boom. Based on our numerical results, we draw the
following conclusions:
(1) Continuous oil ow stream presents just releasing from the
leakage hole with the height of about 3 m. Then it is tore apart
at a certain depth (for standard case it is 3 m) by current and

appears in the form of droplets. Under the action of water

shear stress, oil droplets become more dispersed with the
increase in rising height. The rst oil droplet reaching the free
surface is not the one which has the longest horizontal
migrate distance for the size of oil droplet varying with time.
For standard case, 39 s is required for oil reaching the longest
horizontal migrate distance when it reaches the surface, and
the longest horizontal migrate distance is 16.7 m.
(2) Vertical upward force is small for high density oil, leading to
slow rising rate of droplets. For ro 960 kg/m3, the required
time for the maximum horizontal migration is about 1.84
times as long as that when oil density is 780 kg/m3. However,
the difference in the maximum horizontal migration distance is little for different-density oil ow. Due to high-speed
leaking oil has more ascending kinetic energy, for vo 5 m/s,
just 15 s is needed for oil to reach the maximum horizontal
migrate distance, while it is 79 s for vo 1 m/s. Since the
shearing action times of water are different, the maximum
horizontal migrate distance for vo 1 m/s is 5.2 m longer
than that for vo 5 m/s. High-speed water exerts more shear
stress on oil droplets and transfers more kinetic energy to oil
droplets. Therefore, the longest horizontal migrate distance
for vwmax 0.04 m/s is 14.2 m, 2.5 m less than that for
vwmax 0.1 m/s. Because the route of oil droplet is the square
root of the sum of the squares on vertical depth and horizontal displacement, the route of oil droplet is relative small
in vwmax 0.04 m/s, resulting in the short rising time. The
bigger the diameter of leak, the larger the amount of released
oil and the greater the upward momentum is. As leakage size
reduces from 0.05 m to 0.01 m, the required time decreased
by 23.53 percents. The maximum horizontal migrate distance presenting in the case of d 0.05 m is about 1.5 times
than that for d 0.01 m.
(3) The larger the density of oil, the slower the oil leaking, or the
smaller the leak size, the longer the dimensionless oil spill
time is. The three tting formulas of dimensionless time
meet the natural logarithm distribution well. The dimensionless longest horizontal distance increases with the increase of leak size, while increases with the decrease of oil
leaking rate. With the increase in oil density, the dimensionless longest horizontal distance decreases rstly and
then increases. The two formulas for dimensionless rate and
dimensionless leak size meet the natural logarithm distribution, while the formula for dimensionless density meets
the polynomial. Using the formulas we can obtain when and
where to see oil reaching the sea surface, and conduct rapid
response. We have taken ro 850 kg/m3 for example to
show the using method of formulas. The calculated results
are 35.86 s and 18.69 m.

Fig. 9. The whole process of oil spill.

H. Zhu et al. / Energy 64 (2014) 887e899

(4) The process of oil spill consists of rising process and drifting
process. For drifting process, the main motion of oil droplets
is moving downstream along the free surface, and water is
acting as a carrier. A forecasting model, L Lf a (tr  t)
vwmax, is proposed to calculate the maximum horizontal
migration distance of oil at certain time. For standard case, its
value would be 876.02 m if 2 h are needed for rescue ships to
reach the accident location.
Acknowledgments

[9]
[10]

[11]
[12]
[13]
[14]
[15]

Research work was supported by the special fund of Chinas

Central Government for the development of local colleges and
universitiesdthe project of national rst-level discipline in Oil and
Gas Engineering (P019) and Key Project of Sichuan Provincial Education Department (No: 12ZA189). Without the support, this work
would not have been possible.
References
[1] Wang CY, Chen B, Zhang BY, He SJ, Zhao MM. Fingerprint and weathering
characteristics of crude oils after Dalian oil spill, China. Mar Pollut Bull
2013;71:64e8.
[2] Xu Q, Li XF, Wei YL, Tang ZY, Cheng YC, Pichel WG. Satellite observations and
modeling of oil spill trajectories in the Bohai Sea. Mar Pollut Bull 2013;71:
107e16.
[3] Biksey TM, Schultz AC, Bernhardt AM, Marion B, Peterson C. Ecological and
human health risk assessment. Water Environ Res 2010;83:2067e94.
[4] Berry A, Dabrowski T, Lyons K. The oil spill model OILTRANS and its application to the Celtic Sea. Mar Pollut Bull 2012;64:2489e501.
[5] Gonzlez M, Ferrer L, Uriarte A, Urtizberea A, Caballero A. Operational
oceanography system applied to the prestige oil-spillage event. J Marine Syst
2008;72:178e88.
[6] Lamine S, Xiong DQ. Guinean environmental impact potential risks assessment of oil spills simulation. Ocean Eng 2013;66:44e57.
[7] Li W, Pang YJ, Lin JG, Liang X. Computational modeling of submarine oil spill with
current and wave by FLUENT. Res J Appl Sci Eng Technol 2013;21:5077e82.
[8] Yadav AS, Bhagoria JL. A CFD (computational uid dynamics) based heat
transfer and uid ow analysis of a solar air heater provided with circular

[16]
[17]

[18]

[19]

[20]

[21]
[22]
[23]

[24]

[25]
[26]

899

transverse wire rib roughness on the absorber plate. Energy 2013;55:

1127e42.
Arpino F, Massarotti N. Numerical simulation of mass and energy transport
phenomena in solid oxide fuel cells. Energy 2009;34:2033e41.
Jalilinasrabady S, Palsson H, Saevarsdottir G, Itoi R, Valdimarsson P. Experimental and CFD simulation of heat efciency improvement in geothermal
spas. Energy 2013;56:124e34.
Alfonsi G. Reynolds-averaged Navier-Stokes equations for turbulence
modeling. Appl Mech Rev 2009;62. 040802e1-20.
Hinze JO. Turbulence. New York: McGraw Hill, USA; 1975.
Frisch U. Turbulence. London: Cambridge University Press, UK; 1995.
Hieu PD, Katsutoshi T, Ca VT. Numerical simulation of breaking waves using a
two-phase ow model. Appl Math Model 2004;28:983e1005.
Hirt CW, Nichols BD. Volume of uid (VOF) method for the dynamics of free
boundaries. J Comput Phys 1981;39:201e25.
Karim VM, Bart M. Application of two buoyancy-modied k- turbulence
models to different types of buoyant plumes. Fire Safety J 2006;41:122e38.
Chintala V, Subramanian KA. A CFD (computational uid dynamics) study for
optimization of gas injector orientation for performance improvement of a
dual-fuel diesel engine. Energy 2013;57:709e21.
Mohammed SL, Michael F, Peter G, Derek BI, Lin M, Mohamed P, et al. CFD
predictions of wake-stabilised jet ames in a cross-ow. Energy 2013;57:
259e69.
Rohdin P, Moshfegh B. Numerical predictions of indoor climate in large industrial premises: a comparison between different k- models supported by
eld measurements. Build Environ 2007;42:3872e82.
Ji MK, Utomo T, Woo JS, Lee YH, Jeong HM, Chung HS. CFD investigation
on the ow structure inside thermo vapor compressor. Energy 2010;35:
2694e702.
Patankar SV. Numerical heat transfer and uid ow. New York: McGraw-Hill,
USA; 1980.
Mikilkumar BG, Rupa V, Hari V, David F, Kalpit S. CFD based prediction of
erosion rate in large scale wall-red boiler. Appl Therm Eng 2012;42:90e100.
Zaim EH, Nassab SAG. Numerical investigation of laminar forced convection of
water upwards in a narrow annulus at supercritical pressure. Energy 2010;35:
4172e7.
Almohammadia KM, Inghama DB, Maa L, Pourkashana M. Computational uid
dynamics (CFD) mesh independency techniques for a straight blade vertical
axis wind turbine. Energy 2013;58:483e93.
Jeffrey WB, William DH. Upwind schemes for the wave equation in secondorder form. J Comput Phys 2012;231:5854e89.
Terashima HS, Koshi M. Approach for simulating gaseliquid-like ows under
supercritical pressures using a high-order central differencing scheme.
J Comput Phys 2012;231:6907e23.