The 26th International Workshop on Water Waves and Floating Bodies, Copenhagen, Denmark, 2012

Study on the Effect of Density Ratio of Liquid and Gas in Sloshing Experiment
Yangjun Ahn*, Sang-Yeob Kim*, Kyong-Hwan Kim*, Yonghwan Kim*, Jong-Jin Park**
*Department of Naval Architecture and Ocean Engineering, Seoul National University,
Seoul, Korea
** Samsung Heavy Industry,
Goje, Korea

INTRODUCTION SCALING LAWS and DIMENSIONLES NUMBERS

Sloshing in LNG carriers can lead to large impacts on the containment Several parameters including the density ratio have been proposed for
system. It is important to assess these impact pressures and forces for sloshing problem, and they consider compressible phenomena.
adequate design of containers. Because of stochastic character of Nondimensional approach to the Navier-Stokes equation for
sloshing, experimental analysis is mainly recommended by ship homogeneous compressible flows is
classification societies (ABS, 2006; DNV, 2006). An experimental
 ' L
2
u 'L  1    G ' u 'G2  1  
system for sloshing has been settled down in Seoul National University   2 
z '   '   t '  2   Fr 2  z '
 Fr    
G
(SNU) to predict pressure impacts. t ' 2 
In the application of experiment analysis, how to scale the experimental
results to the actual design is one of the difficulties. Dimensionless 1  1 1 
numbers for this application have been studied, but none of them have    
drawn a complete conclusion. Global behavior of the fluids is governed We  k '1 k '2 
by the Froude number, so the ullage pressures should be Froude scaled
(Bass et al., 1980). The local behavior, however, needs another scaling  
where Fr  U / gD ,   G /  L  1/  , We   ' L U D /  ,
2

law. respectively (Yung et al., 2009).

To find out appropriate scaling law for local phenomenon of sloshing, Sloshing impact can be categorized by the magnitude of the fluid
analytic, numerical, and experimental studies have been conducted. velocity u (Dias et al., 2007). According to this category, a
Acoustic scaling including the density ratio has been theoretically homogeneous flow with mixture of liquid and gas hit the wall at the
identified, it is to be relevant as well as Froude scaling to the sloshing same speed, or otherwise separated flow give an impact on the wall
problem (Dias et al., 2007). During the impact, transfer of momentum with different velocities. These two types of flows need different
between liquid and gas is occurred, so the density ratio has an influence scaling law, and they are
on the impact pressure. Numerical studies have also shown this
influence of the density ratio (Dias et al., 2007).  p  S , SF
 1  2 p  M , SF
The consequences of experiments have substantiated the importance of
the density ratio (Maillard and Brosset, 2009), and a dimensionless  p  S , HF

3
 p  M , HF
number, which consists of the density ratio and polytropic index, has where
been proposed with experimental results (Yung et al., 2009). Sloshing
model tests at small scale which give the density ratio consideration is vapor cvapor 1/ 2  1  liquid g Fr L1/M2
1 
considered more representative. Model tests of previous studies,
 gas cgas  1   water g FrL1/M2
however, have been performed with 2D harmonic motions for reducing
uncertainties of an experiment. Influence of the density ratio on 2
liquid  gas  cliquid   cgas 
sloshing is still unclear, and another parameter of sloshing experiments 2     
can be dependent on the density ratio. The density ratio, for example,  water vapor  cwater   cvapor 

could affect the resonance frequency, which is usually shifted with the
experimental conditions from the theoretical approach. liquid  gas cvapor
3  
2
This paper presents a variation of the sloshing loads with respect to the  water vapor cgas
density ratio between fluids. 2D tank has been manufactured for finding
appropriate resonance frequencies. Moreover, 3D tank and actual The subscript S refers to the real scale, M refers to the model scale, HF
design sea conditions have been carried out, and the experimental to the homogeneous flow, and SF to the separated flow (Dias et 마,,
results with different filling levels are illustrated in this study. The 2007).
results of actual design conditions could be affected by secondary The governing equation and those dimensonless numbers indicate that
effects of the test conditions such as heading angles, sea conditions, 3D consideration of the density ratio is strongly required for prediction of
effects, and other parameters, the influence of the density ratio on the sloshing impact. Therefore, in this study, the density ratio is considered
statistical pressure is still observed even in the irregular motions of the as a parameter of sloshing experiment, and possible influences are
significant environment conditions. observed.
EXPERIMENTAL SETUP Measuring System

Motion Platform and Model Tank Fig. 4 shows a simple diagram of experimental setup. Tank model is
excited on the motion platform and sloshing impact pressures inside of
The experiments were carried out in Seoul National University (SNU). the tank are measured by pressure sensors. A coupler adjusts the bias
On a motion platform, two model tanks have been mounted: 1/50-scale voltage of pressure signal to zero and the DAQ system converts the
3D model tank, and 1/40-scale 2D model. The motion platform, which pressure signal to digital data with direct current coupling.
has five ton capacity, is hexapod type consisting of six actuators. The Sloshing pressures were measured with Integrated Circuit Piezoelectric
3D model tank is based on a membrane tank of 160K LNG carrier, and (ICP) sensor (model: 211B5) made by KISTLER Co. The sensing
it has been design by Samsung Heavy Industry. The geometry of 2D diameter of this sensor is 5.54mm, and the maximum pressure range is
model tank is given by Gaztransport & Techigaz SAs at the benchmark 7 bars. The sensors were placed in a various arrays. The sampling
on sloshing model test. frequency was fixed at 20 kHz which is suggested as in the reasonable
The inner side of the 2D tank has a length of 946mm, a width of range (Nitin et al., 2010).
118mm, and a height of 670mm on model scale. The test tanks were
made of acrylic to enable visualization of the fluid motion inside of the
tank and its thickness 35mm was used for preventing vibration of tank
walls. Surfaces of the tank are flat, so invar edges or corrugations of an
actual tank were not considered. Geometries of the tank are presented Fig. 4 Experimental setup
in Fig. 1 and the model tanks are shown in Fig. 2 and Fig. 3.
THE DENSITY RATIO VARIATION

Mixed gas properties

Water and air have been commonly used for sloshing model tests due
to practicality. The density ratio of water and air, however, is different
from that of natural gas (NG) and liquefied natural gas (LNG). The
condition of the model test using water and air at ambient is physically
different, so an alternative material is required instead of water or air at
ambient to have similar density ratio between liquid and gas.
A mixture of SF6 and N2 replacing air was used as an alternative. The
right proportion of these two constituents was used in the experiments
for varying the density ratio, and water was used as it is. Properties of
gas which were used are presented in Table 1.

Fig. 1 2D model tank geometry Table 1 Gas properties

Density Ratio (%) of
Products
(kg/m3) mixture
Sulfur hexafluoride (SF6) 6.1620 56.9
Nitrogen (N2) 1.1455 43.1
Mixture 3.9999 -

Procedure of gas injection

Tests were carried out with the completely sealed tank. To prevent
losing gas from the tank, the water type silicon was used for sealing,
and gas loss from the tank was checked with bubble at every test before
and after. Pressure inside of the tank could influence on the result of the
sloshing impacts, so a pressure gauge was manufactured for observing
Fig. 2 3D model tank inside pressure. To make the target density ratio of each experiment,
the oxygen level inside of the tank was checked instead of the level of
mixture gas because of two reasons: the oxygen level required a less
heavy tester than the sulfur hexafluoride level did; and oxygen has the
second most proportion of air. The procedure of the gas injection and
sealing tests are shown in Fig. 5.
Most gases including SF6 and N2, which were used in the experiments,
are soluble. After the injection of mixture, gas which was dissolved in
air inside of the tank had different density to that in water. This
difference could change the density ratio while the experiments were
conducting. Hence, water should be fully saturated with the gas in
advance; therefore, the fluids inside of the tank were shaken to be
mixed. One degree-of-freedom forced motion was performed with
several repetitions until the oxygen levels were converged. The oxygen
Fig. 3 2D model tank levels were also recorded before and after each test, and the density
ratio inside of the tank was checked with this record.
TEST CONDITIONS

2D regular conditions

The resonance frequency equation, which is generally used, is given by

1  h 
fn  g tanh  
2 l  l 
and this is based on 2D approach. Moreover, because standing wave is
the primary concern for this equation, breaking waves or hydraulic (a) Filling level = 95% (b) Filling level = 80%
flows are not included. Amplitude of forced motion is also excluded, so
the experimental results could varies with a lot of experimental
conditions.
Based on the resonance frequency equation, the frequency of forced
motion varies to find out the condition giving the largest sloshing
pressure. Two different amplitudes of 40 mm and 15 mm are
considered for different filling levels.

3D irregular conditions

All the irregular test conditions have drawn from screening experiments. (c) Filling level = 70% (d) Filling level = 60%
In the screening experiments, the significant wave height is considered Fig. 5 Measured sloshing pressure varying the excitation frequency
as 40 years in North Atlantic Ocean for head and quartering seas, and 1
year for a beam sea. Sloshing pressures are measured during 36
minutes corresponding to 5 hours in real scale. The screening
experiments, which covered various sea conditions, filling levels, and
heading angles, led to the experimental results of sloshing pressures,
and three conditions giving significant results were selected as the test
conditions.
Each selected condition was repeated with five times of five different
density ratios between liquid and gas inside of the model tank were
derived for each test. Temperature and pressure inside of the tank were
kept as a room condition, so the density ratio varies only with
controlling the composition of gas. Fig. 6 Biased impact location at the same experimental conditions

EXPERIMENTAL RESULTS

The results from the post-processing of data are summarized in this part.
Resonance tests have been being derived, and parts of them show the
results which are away from the prediction. In the irregular conditions,
sloshing impacts are occurred on the different tank faces according to
the filling levels; therefore, the experimental results are categorized
along the filling levels and tank faces.

2D regular conditions Fig. 7 Change of overall flow with small difference of frequency
Fig. 5 presents the 1/10 largest impact pressures of the 2D harmonic 3D irregular conditions
motions. The filling level is higher; distribution of the sloshing pressure
tends to have wider bands. In the same filling condition, the magnitude The 3 hours extreme statistical pressures, which have been obtained by
of the amplitude does not indicate the certain shift direction of the Weibull and Pareto schemes, are shown in Fig. 8. Tank top comes first
largest impact frequency. for high filling conditions. The low filling level tests are not presented
Complex phenomena have been observed in particular conditions. here, because no sloshing impacts have occurred on the tank top at the
Sloshing impacts only occur on the one side of the tank in the regular low filling tests. Fig. 8 indicates that the statistical pressures decrease
motion of 70%H filling level with amplitude 40 mm and 0.948 Hz in approximately 50% as the density ratios increase from 0.0012 to
model scale. The position which the impacts are observed switches its around 0.004.
side depending on the initial mass movement of the liquid. Fig. 6 shows The density ratio is predicted to influence on pressure rise times. Since
that biased impact locations appear although the experiments of (a) and high sloshing impacts have been primary concerns, the average rise
(b) of Fig. 6 have been carried out in the same conditions. times of 10 largest, 20 largest, and 30 largest impact pressures are
Small differences of frequency give significant change of sloshing considered with respect to the density ratio. In the high filling
pressures. The frequency of (a) of Fig. 7, which is 0.8959 with the experiments, global slowdowns of the pressure rise times have been
filling level 70%H and the amplitude 40 mm, is 0.0009 lower than that recorded, and these results are presented in (a) and (b) of Fig. 9.
of (b). The sloshing impact, however, only occurs in the condition of Consideration of more impact signals distracts the trend of rise time;
(a) of Fig. 7. Rise time of fewer largest impacts is more clearly increase.
 significant changes and the obvious trend through the experiments
while the low filling conditions do not show such clear results.
 Between large impacts of each test case, it is observed that rise time
increases as the density ratio does. The magnitude of its change is
more obvious in the experiments of high filling conditions than in
those of low filling conditions.
 While the sloshing impact pressures is extremely varied between a
density ratio of 0.0012 and 0.004, the rate of the change of pressure
decreases as the density ratio approaches to that of actual LNG cargo
tank.
 Location occurring sloshing impact on relates to the experimental
(a) Filling level = 95%H (b) Filling level = 70%H conditions; the density ratio might have an influence on the location
Fig. 8 Influence of the density ratio to the statistical pressures on the of the impact.
tank top  The most significant sloshing impacts would not occur at the
resonance frequencies of the model tanks, and their frequencies are to
be shifted. Sloshing impacts are sensitive to the frequency of tank
motion, and influence of the density ratio on the resonance frequency
is not yet identified.
In the experiments, it is hard to satisfy the conditions having the
exactly same density ratio as the conditions of actual LNG cargo tank.
Because small magnitude of density is dealt with, the density ratio of
liquid and gas is very sensitive to the environmental conditions such as
room temperature or pressures inside of the tank during the
experiments. However, if the sloshing impact pressure in the
experiments would be converged and become less sensitive in some
specific region of the density ratio around 0.004, the experiments will
(a) Filling level = 95%H (b) Filling level = 70%H be easy to be handled.
Fig. 9 The average of rise time on the tank top – n largest pressures
REFERENCES
Low filling level of irregular conditions
Advanced Marine Technology (2010) “Development of large scale
Every low filling condition was repeated twice to get more appropriate sloshing experiment facilities,” AMEC Project Report, Seoul
results. During the low filling tests, most of the sloshing impacts are National University, Seoul, Korea.
occurred on the side wall. These statistical sloshing impacts on the side American Bureau of Shipping (2006) Guidance notes on strength
wall are shown in Fig. 10. assessment of membrane-type LNG carriers, Guidance Note
Although the tests have been repeated, some discrepancies are observed. (updated in 2009), Houston, USA.
In the first low filling tests, the highest statistical pressures are recorded Braeunig, J. –P., Brosset, L., Dias, F., Ghidaglia, J. –M. (2009).
when the test are performed with the density ratio around 0.0033. It is “Phenomenological Study of Liquid Impacts through 2D
noticed that a weak trend of decrease in the second low filling test, but Compressible Two-fluid Numerical Simulations” Proc. 19th
the graph shows an unpredicted result around 0.0033 density ratio. International Offshore and Polar Engineering Conference, ISOPE,
Osaka, Japan.
Det Norske Veritas (2006) Sloshing Analysis of LNG Membrane
Tanks, Classification Notes, Oslo, Norway.
Dias, F., Ghidaglia, J. –M., Le Coq, G. (2007). “On the Fluid Dynamics
Models for Sloshing” Proc. 17th International Offshore and Polar
Engineering Conference, ISOPE, Lisbon, Portugal.
Kim, K.H., Kim, Y., Kim, S.Y. (2011). “Comparative Study on Model-
Scale Sloshing Tests,” Proc. 21th International Offshore and Polar
Engineering Conference, ISOPE, Maui, USA.
Kuo, J. F., Campbell, R. B., Ding, Z., Hoie, S. M., Rinehart, A. J.,
SandstrÖ m, R. E., Yung, T. W., Greer, M. N., Danaczko, M. A.
(2009). “LNG Tank Sloshing Assessment Methodology – The New
(a) Filling level = 10%H (case 1) (b) Filling level = 10%H (case 2) Generation,” Proc. 19th International Offshore and Polar
Fig. 10 Influence of the density ratio to the statistical pressures on the Engineering Conference, ISOPE, Osaka, Japan, Vol. 19, pp 241-253.
side wall Maillard, S., Brosset, L. (2009). “Influence of Density Ratio between
Liquid and Gas on Sloshing Model Test Results,” Proc. 19th
CONCLUSIONS International Offshore and Polar Engineering Conference, ISOPE,
Osaka, Japan.
The experiments draw the conclusions that the sloshing impact varies Yung, T-W., Sandström, R. E., He, H., Minta, M. K. (2010). “On the
as the density ratio is changed. Mixed gas of SF6 and N2 has been Physics of Vapor/Liquid Interaction During Impact on Solids,”
injected, and an experimental procedure with completely sealed 2D Journal of Ship Research, Vol 54, No 3, pp 174-183.
and 3D tanks has been proposed for varying the density ratio. Some Yung, T-W., Ding, J., He, H., Sandström, R. E. (2009). “LNG
conclusions are made as follows: Sloshing: Characteristics and Scaling Laws” Proc. 19th International
 In the irregular conditions, the high filling conditions show more Offshore and Polar Engineering Conference, ISOPE, Osaka, Japan.