You are on page 1of 6

2E.

Bubble Curtain for Blocking Spilled Oil


on Water Surface

Isamu FUJITA
Oil Spill Response Research Group,
National Institute of Maritime, Port and Aviation Technology,
Port and Airport Research Institute
3-1-1 Nagase, Yokosuka, JAPAN
fujita@pari.go.jp

Abstract— For protecting important port facilities such as


emergency evacuation bases from oil spill incident triggered by I. INTRODUCTION
natural disasters, some self-sufficient countermeasures are In Japan, port areas are highly concentrated with industrial
required for blocking or eliminating floating oil from the facilities. In some cases, oil refinery plants and a disaster
facilities. For this purpose, bubble curtain was studied. The evacuation base locate near to each other. In such a situation,
bubble curtain itself already has some preceding researches or
once an oil spill incident is triggered by a natural disaster such
applications for containment of floating oil, most of all, however,
as earthquake or tsunami, the spilt oil from the plant may cover
deals with the phenomena in open sea. This paper especially
focuses on the bubble curtain deployed near a vertical wall, the sea surface around the evacuation base and impede rescue
intended for application to protecting land facilities. vessel access to the evacuation base as shown in Figure 1. The
Experimental measurements using video processing were carried evacuation base should have some self-sufficient counter
out in a test tank to see how large oil-free area the bubble curtain measure to prevent the oil from accessing to the base or to
can generate. The experimental results were analyzed and exclude the oil from the quay or navigation channel for
arranged to a simple empirical equation providing the size of oil enabling the rescue vessel’s immediate approach.
free area. 2-D and 3-D Numerical simulations using OpenFOAM
For this purpose, we focused on a bubble curtain system.
was also conducted for verifying the experimental results and for
deeper understanding of air-water multiphase physics. This
The bubble curtain itself already has some preceding
paper also includes discussion about the horizontal velocity field researches or applications for containment of floating oil[1],
that the bubble promotes near the water surface and comparison but most of all deals with the phenomena in open sea and cross
with some previous studies. setting which means the bubble line is orthogonal to the water
current[2]. In this study, we focused on the bubble curtain
Keywords— oil spill; bubble curtain; OpenFOAM; tracking; deployed near a vertical wall, intended for application to an
turbulent diffusivity; autonomous equipment for blocking or eliminating drifting oil
patches from the quay like a typical configuration shown in

Fig. 1. Example of coastal area sensitive to


natural disaster-triggered oil spill Fig. 2. Concept of bubble curtain oil eliminator

978-1-5090-2445-2/16/$31.00 ©2016 IEEE 354 Techno-Ocean 2016


Figure 2. In this configuration, quay wall may suppress the picture camera mounted on the upper side was used for
normal direction flow and make the water current parallel to recording the movement of the oil patches drifting from the
the wall and the bubble line. upper stream as shown in Fig. 4. In this case, the air feed pipe
is placed along the tank wall (y=0) and the water flows from
In this study, we dealt with a parallel setting bubble curtain the left (from –x direction). We can see an oil-free area formed
primarily with two focuses. The first one is to propose a model
by the bubble curation. We defined an boundary line y  W x 
which provides the oil exclusion area by the bubble curtain in
case that the bubble curtain is aligned parallel to the water as the oil exclusion width as Fig. 5 shows. In this process, an
current. For this purpose, we conducted experiments in a sea image processing software of OpenCV® was used. Sequential
water tank using underwater air nozzles in line in various processes from RGB decomposition, luminance inversion,
conditions, and measured how large area the bubble curation binarization and averaging gives an oil cloud image as Fig. 5(a)
can exclude the floating oil using video image processing. shows and W x  can be defined as a y coordinate where the
luminance exceeds a certain threshold as Fig. 5(b) shows.
Secondly, for further understanding of air-water multiphase
physics, we measured the horizontal velocity field that an In addition to the observation of the oil-free area, surface
inline bubble promotes near the water surface and compared flow induced by the bubble curtain was measured with an
the results with several models proposed by preceding
researchers.
For analysis and discussion, in addition to tank test, we also
deployed OpenFOAM® to numerically simulate bubble curtain
both in 2D and 3D space. The numerical simulation results
were compared with the experimental measurement and the
proposed model.

II. EXPERIMENTAL AND NUMERICAL SIMULATION

A. Experimental
In this study, to know how large oil-free area can be
provided by a bubble curtain, we conducted experiments in a Fig. 4. Experimental setup
sea water circulation tank (30 x 6 x 2.5[m] : L x W x D) using
underwater air nozzles in line in various conditions. A
schematic of the experimental setup is given in Fig. 3. High
pressure air provided by an air compressor is fed to air feed
pipes submerged in the water after its pressure and flow
regulated. The air in the pipe comes out through drilled small-
diameter holes on the pipe wall and forms uprising plume. The
plume reaching the water surface changes its flow in an
horizontally outward direction. This horizontal flow blocks or
pushes away oil patches floating on the water surface. Motion-

(a) Oil cloud extraction (image processing)

Fig. 3. Experimental setup (b) Edge detection and line fitting


Fig.5. Example of oil exclusion experiment

978-1-5090-2445-2/16/$31.00 ©2016 IEEE 355 Techno-Ocean 2016


(a) 2-D model

Fig. 6. Cross section of bubble curtain and surface velocity measurement

electromagnetic current meter as shown in Fig. 6. This


measurement was conducted in a calm water for comparing
with numerical simulation as well as preceding researches.

B. Numerical Simulation
In addition to the experiments, we also conducted 2-D and (a) 3-D model
3-D numerical simulation using a CFD software of
OpenFOAM. In our study, InterFoam, a solver for 2 Fig. 7. Schematic of CFD model
incompressible fluids was used for the simulation. Strictly, air
compressibility should be considered, however, in case of
shallow water, this effect can be thought negligible. A
schematic of 2-D simulation is given in Fig 7(a). A semicircle
of the air feed pipe was place on the left symmetric wall.
In 3-D simulation, the experimental test tank was
numerically reproduced. The air feed pipe, which was
expressed as a tape-wise baffle plate having a fixed volumetric
inlet flow rate and VOF value, was placed in various depth,
position and direction. An example of 3-D model is illustrated
in Fig. 7(b). Floating oil was not modeled in the CFD
calculation, however, simulated in the post process assuming
that the oil movement follows the interfacial velocity field
between the air and water. Figure 8 shows an example of oil
Fig.8. Example of particle trace on the water surface
trajectory calculated in this way.

experiments, this area defined as W x  seems to be


approximated well with a parabolic function. To the contrary,
III. RESULS AND DISCUSSIONS the 3-D numerical calculation told that the oil exclusion width
converges to a constant value as x is getting large and its
A. Oil exclusion area modeling shape is, as Fig. 8 shows, approximated very well with a
So far, we can see a number of researches on bubble curtain. following exponential function,
However, many of them dealt horizontal velocity which the
uprising bubble induces. For practical application, more W x   x  x0 
 1  exp   
 (1)
straight forward information about the size of the oil-free area W  xt 
which we can get with a bubble curtain is desirable. Here we
would like to propose a model providing the shape of oil-free where, W is an convergence width, x0 is the position of the
area. In case that the bubble line is aligned parallel to the water apex and xt is a shape parameter which determines the growth
current, our experimental observation gave a wedge-shaped oil- of the oil-free area. We used this form to fit the experimental
oil free area as shown in Fig.5(b). Judging only from the results. In the Eq. 1, W is the most important and our focus.

978-1-5090-2445-2/16/$31.00 ©2016 IEEE 356 Techno-Ocean 2016


Fig.9. Correlation between model and CFD simulation Fig.10. Correlation between model and experiments

1
  U Us 

There are five major factors determining W [m] which are air U   1  erf  (3)
2 
 2 
2
supply rate Q [m3/s], current velocity U [m/s], air supply depth  
D [m], air pipe length L [m], density difference  [kg/m3]
and calculate an average velocity U s and a standard deviation
and gravity g [m/s2]. These factors include three basic units of
 . For instance, Figure 12 gives the case that the nozzle depth
m, s and kg. According to Buckingham’s  theorem, an was 1.93m and the air feed rate was 200 l/min.
equation including three non-dimensional numbers can
describe the bubble curtain system. In this study, we assumed
the following model,
p p
U  gD  D
 gL  L

W  c 


 P 
 (2)
Q  P0   0 
where, P0 is atmospheric pressure and P0 g is a
representative length scale. Power indices of p D and p L were
determined by a regression using the CFD simulation results.
In our research, 0.194 for p D and -0.394 for p L gives the best
fit and as Fig.9 shows, excellent correlation was found between Fig.11. Example of surface velocity measurement
the model and CFD simulation. Equation 2 is also applicable to
the experimental results. Figure 10 shows the case using the
same p D and p L to the CFD.

B. Horizontal flow velocity by air bubble


In this section, horizontal flow velocity near the water
surface induced by bubble curtain is discussed by being
compared with some preceding models that several researchers
have proposed so far. Figure 11 shows an example of the
horizontal velocity measured at a depth of 50mm and a
distance from the center of the air feed pipe of 0.5m. Since the
horizontal velocity has much fluctuation, we fitted the data
with a normal distribution of

Fig.12. Distribution of horizontal velocity

978-1-5090-2445-2/16/$31.00 ©2016 IEEE 357 Techno-Ocean 2016


Fig.13. Surface velocity comparison between model and CFD simulation
Fig.14. Surface velocity comparison among CFD,
experiment and models

On the other hand, 2-D numerical simulation, of which  gD 


md

model was shown Fig.7(a), also gives the information about the  D    
 (7)
surface velocity. In this study, we carried out the simulation  P0 
over a range of 0.9 to 2.3 m in pipe depth and 1 to 1000 and fitted the simulation result with it. We found that the power
l/min/m in air feed rate. There have been many preceding index md gives the best fit when its value is about 0.3 as
models by researchers so far such as Bulson[3], Kobus[4] as
well as Iwanaga[5]. They all share the same form Fig.13 shows.
Figure 14 gives comparison between the experimental data
U s  c D gq3
1
(4) and the CFD simulation. The CFD simulation gave about 15 %
higher surface velocity than that of the experiments. This
where q  Q L is an air flow rate per unit length and  D  is discrepancy seems to be caused by the following reasons. In
a function giving pipe depth effect which has varieties the experiment, a 2.3 m length pipe set in 6.5 m width test tank
depending on the researchers as follows: might affect two-dimensional quality and also surface velocity

1 measurement depth at y=-30mm might give smaller value than
 gD  3
the true surface value. In spite of these constraints on the
1   ………. Bulson
 P0  experiments, Eq. (7) also represents the experiments with the

same power index md of 0.3 to the CFD case. To the contrary,
1
 D  
  gD  2
(5) Kobus model gives 0.5 as the power index which seems to

P0
 ln1   Kobus describe our result well as the figure shows.
 g D  D0  
 
P0 

aP0
1-
gD
Iwanaga
IV. CONCLUDING REMARKS
Our concern is the oil exclusion in a shallow water, hence, We studied on bubble curtain for drift control of floating oil.
an asymptotic behavior where D is small or D  0 is We proposed a simple model which gives the oil exclusion
important. From this viewpoint, Bulson’s model gives area by a bubble curtain aligned parallel to the water current.
irrational estimation since it doesn’t converge to zero when The proposed model was compared with experiments as well
D  0 as well as Iwanaga’s model. To the contrary, Kobus’s as numerical simulation, and its good representation was
model has a asymptotic form like verified.
1
 D 2
 D    
 (6)
 D0  REFERENCES
, it satisfies the condition  D  0 D  0 . D0 is an [1] T. McClimans, I. Leifer, S. H. Gjosund, E. Grimaldo, P. Daling,
F.Leirvik: Pneumatic oil barriers: The promise of area bubble plumes, J.
depth parameter. Kobus called it “analytical origin” giving an Engineering for the Marintime Environment, Vol 227, No. 1, pp.22-38.
extension depth for the air feed pipe and . recommended 0.8 m 2012
for its value. [2] S. Hara, M. Ikai, S. Nmaie: Two-dimnsional Plume Induced by the Air
Bubbles in Water (Fundamental Study on an Air Bubble Type of Oil
In our research, we assumed the form of  D  as

978-1-5090-2445-2/16/$31.00 ©2016 IEEE 358 Techno-Ocean 2016


Boom), Technical report of NMRI, Vol. 22, No. 3, pp.261-285, in [4] H. E. Kobus: Analysis of the Flow Induced by Air-Bubble Systems,
Japanese, 1985. Proc. 11th Conference on Coastal Engineering, Vol. II, pp 1016-1031,
[3] P. S. Bulson, M. I. Struct: The theory and design of bubble breakwater, 1968.
Proc. Proc. 11th Conference on Coastal Engineering, Vol. II, pp. 995- [5] Y. Iwagaki, T. Asano: Study on Neumatic Break Water (4 th report), Proc.
1015, 1968. 24th Coastal Engineering Conference, p.295, in Japa

978-1-5090-2445-2/16/$31.00 ©2016 IEEE 359 Techno-Ocean 2016

You might also like