Upload
ngokhanh
View
223
Download
3
Embed Size (px)
Citation preview
Experimental Study of Effective Water Spray Curtain Application in
Dispersing LNG Vapor Clouds
Morshed A. Rana, Benjamin Cormier, Jaffee Suardin, Yingchun Zhang, M. Sam
Mannan*.
Mary Kay O’ Connor Process Safety Center, Artie McFerrin Department of Chemical
Engineering, Texas A & M University, College Station, TX 77843.
Abstract
The installation of new LNG storage facilities in the US to meet the demand of natural
gas has brought increasing attention to LNG safety issues. Because of its highly
flammable nature, LNG poses several hazards to workers, properties, and the surrounding
communities. One of the major hazards is the formation of a flammable vapor cloud from
any accidental LNG release, which may result in a massive fire. The safety measures to
prevent and mitigate an accidental LNG release are essential and critical to protect the
employees and the public from injury or harm.
The water spray curtain is currently recognized as a promising technique to control and
mitigate many toxic and flammable vapors. Much theoretical and experimental work has
been carried out to determine the effectiveness of the water spray curtain in dispersing
heavier vapor. However, LNG vapor dispersion behaves differently from other dense
gases due to its low molecular weight and extremely low temperature. In this context, it is
essential to carry out research to understand the effects of water curtain on LNG vapor
clouds. There have been a very limited number of publications on the use of a water
spray curtain to mitigate LNG vapor clouds. These previous studies show that water
spray curtains can enhance LNG vapor dispersion from small spills. However, in order to
develop comprehensive and structured engineering guidelines for the design of an
effective water spray curtain for controlling LNG vapor many key questions still remain
to be answered.
An experimental methodology to study the LNG vapor dispersion behaviors with the
application of water spray curtain is presented in this paper. This field experiment
involves the fundamental study of forced dispersion, dilution due to air entrainment, and
heat exchange to determine the effectiveness of water spray in reducing the LNG vapor
“exclusion zone”. This paper discusses and outlines the experimental method and some
results based on gas concentration data analysis to emphasize the observed effectiveness
of water spray curtain on LNG vapor dispersion.
Keywords: LNG spill, vapor control, forced dispersion, spray nozzle.
1 INTRODUCTION
Liquefied Natural Gas (LNG) refers to natural gas converted into liquid state by super
cooling to 111K (-162.2°C). LNG commonly consists of 85%-98% methane with the
remainder as a combination of nitrogen, carbon dioxide, ethane, propane, and other
heavier hydrocarbon gases. LNG is usually stored at 111K and atmospheric pressure in
heavily insulated tanks. As forecasted by the U.S. Department of Energy, demand for
natural gas in US is increasing by 20% over the next 25 years because of its clean-
burning characteristics [1]. Due to this increasing demand for natural gas, more LNG
receiving terminals are proposed to be constructed in the US. The massive volume of
LNG storage tanks poses fire and explosion hazards due to its highly flammable feature.
Barriers such as dikes or impounding walls are usually built around the tanks to hold
accidentally spilled LNG to protect adjacent properties.
The vapor cloud formed from a LNG liquid spill disperses in the prevailing wind
direction and is diluted. When the gas concentrations are between the lower and upper
flammability limits (LFL and UFL), 5-15% by volume, the vapor cloud becomes
flammable. Safety standards require LNG facilities to have a “dispersion exclusion zone”
so that flammable vapor cloud from an accidental release will not propagate beyond the
plant boundaries [2]. This exclusion zone is defined from the LNG spill source to the
predicted distance at which the average vapor concentration is one-half of the LFL, i.e.
2.5% volume. This distance is estimated with various dispersion modeling techniques
incorporating appropriate control and mitigation measures.
Water spray curtains are recognized as an effective engineering measure to mitigate
various types of hazards in the industries because of its availability, simplicity of use,
efficiency and adaptability for different hazards such as gas dispersion, absorption, and
fire inhibition [3-4]. Generally, water curtains can be classified as full cone, full square
cone, hollow cone and flat-fan characterized by water droplet sizes and flow patterns [4].
Water spray nozzles can be directed downward and upward, vertically or inclined. The
types and application methods of water sprays have significant effects on the efficiency
and effectiveness in mitigating different hazards.
Water curtain has also been suggested as one of the most economic and promising LNG
vapor mitigation techniques, which can reduce the size of the “exclusion zone”
effectively, by forced dispersion. The effectiveness of water curtain in controlling LNG
vapor depends on a number of parameters. These parameters can be divided into two
groups:
i. Parameters of its own characteristics: water droplet distribution, nozzle type and
size, direction, width and height, water pressure etc.; and
ii. External parameters: vapor cloud features, LNG properties, wind speed,
atmospheric stability etc
The paper presents the experimental methodology and some results of gas concentration
data of the first phase of a research program to study the effectiveness of water spray
curtain to disperse LNG vapor clouds.
2 BACKGROUND AND PREVIOUS WORK
The physical processes involved in using a water curtain to reduce the gas concentration
of a vapor cloud include forced dispersion, dilution, and heat transfer. A water spray
directed vertically through a gas cloud will have four effects: (i) mechanical effects of
creating a barrier to the passage a gas cloud, imparting momentum to the gas cloud to
disperse it, moving gas cloud upwards, (ii) dilution of the gas cloud by air entrainment;
(iii) thermal effects between the gas cloud, water, and entrained air; and (iv) physico-
chemical effects of absorbing gases with or without chemical reaction [5]. The actions of
a water spray in mitigating a vapor cloud may consist of all or any combination of these
mechanisms, if properly designed. Water curtains enhance LNG vapor cloud dispersion
mainly through two mechanisms: mechanical effects (forced dispersion and dilution) and
thermal effects. As the solubility of natural gas in water is minimal, thus the physico-
chemical effects can be ignored.
The interaction among vapor cloud (gas/air mixture), air and water droplet of a water
curtain is a sophisticated phenomenon to model and predict. Several models, based on
macroscopic or microscopic approaches, can be used to predict gas concentration
reduction with the application of a water curtain. The macroscopic models are semi-
empirical and can estimate the concentrations downstream of a water curtain and the
dilution factor of the water spray [5-7]. The microscopic models, based on fluid
dynamics, require accurate information on hydrodynamics and can only predict the
mitigation ability of water curtain without considering the subsequent dispersion of the
gas cloud. An easy-to-use model with a minimum requirement of inputs is essential to
estimate the downstream concentration reasonably well [3-5, 8-9].
Apart from theoretical modeling, much experimental work has been carried out to assess
the effectiveness of the water curtain in gas dispersion. Most work was conducted to
determine the optimum configuration of water sprays to mitigate dense gas clouds like
NH3, Cl2, HF, N2O4, CO2, etc [5, 10-15]. It has been proved that water curtains can be
effective in facilitating the dispersion of toxic and flammable vapor clouds evolved from
accidental spills.
There have been two published studies on the use of water curtain to control LNG vapor
clouds. Experiments on using water curtain for small LNG spills by University Engineers
for US Coast Guard in the late 70’s [16-17], and wind tunnel and small scale spill tests by
Factory Mutual Research Corporation for Gas Research Institute in the early 80’s [18-19]
are the major experimental works so far. The former one simulated spills on LNG ships
and used fan-shaped sprays. The experimental results showed reduction of methane
concentration by the water spray. Two mechanisms, mechanical turbulence and heating
by the sprays, were suggested as the cause of concentration reduction [16-17]. However,
as the temperature of the vapor cloud was not measured, the effect of heating on vapor
dispersion was not clearly identifiable from the experiments.
The later one simulated LNG spill contained in a dike and performed small scale tests
using both downward and upward solid cone sprays, to determine the effectiveness of
water spray curtain. Results showed that water sprays dilute the concentrations to about
1-2%. Air entrainment and heating of the cloud by the spray were considered to enhance
the vapor dispersion. The analysis also suggested that the vapor plume was lofted above
the diked area by the sprays, producing a low ground level concentration downwind with
some vapor potentially escaping between adjacent full cone sprays [18-19].
Because of low molecular weight and extremely low temperature, dispersion of LNG
vapor cloud behaves differently from other dense gases. Though previous work has
confirmed that water curtain represents one promising technique to mitigate the hazard of
accidental LNG release, it was not clear from such experimental work how to design an
effective water curtain. Indeed, it is probably true that there is no simple set of optimum
parameters, rather there will be a specific set of design parameters for each particular
type of circumstances [20]. Other key issues, like (i): the extent of LNG vapor
concentration reduction by water curtain; (ii) the effectiveness as a function of water
curtain configuration; (iii) the optimum distance between the water curtain and potential
spill site; (iv) the effective height and width of the curtain; and (v) the highest wind speed
for which water curtain remains effective; are yet remained undetermined [18].
.
3 EXPERIMENTAL
In the LNG industry, upward-pointing water curtains are mainly used to facilitate LNG
dispersion and prevent water from entering the LNG pool by inclining them from the spill
location. The mixing process for an upward spray is by the upward momentum imparted
to the cloud by the water spray. This upward momentum enhances air entrainment
through the lower surface of the cloud and thus better dilution is achieved. The
momentum also pushes the diluted cloud upward and facilitates dispersion [10].
It was the intent of this research to understand the underlying physical phenomena on the
interaction of water spray with LNG vapor as well as observe the effectiveness of the
water curtain to mitigate and control LNG release through multiple sets of experiments.
This paper is based on the first set of the experiment and the detailed experimental set-up
and procedure are described including different parts involved in the tests. The
experiment consists of the LNG source, the water spray nozzle and water curtain,
placements of the measurement points and meteorological measurements. The procedure
was to perform natural and forced dispersion of LNG to evaluate the effectiveness of the
water spray curtain in terms of concentration and temperature reduction.
3.1 Test Facility This set of outdoor spill experiments was conducted on November 16, 2007, at the LNG
Emergency Response Training Facility of the Brayton Fire Training Field (Emergency
Service Training Institute (ESTI)) in College Station, Texas, USA. The Texas
Engineering Extension Service (TEEX) of the Texas A & M University System provides
comprehensive firefighter training and emergency services at the Brayton Fire Training
Field. Hands-on spill control and fire suppression training are also provided to personnel
involved in LNG production, transportation, storage, and response activities at the LNG
training facility. The LNG facility is made of concrete and consists of three pits, as
shown in Figure 1 including:
i. An L-shaped pit to simulate the trenches used to divert any LNG spills into
containment pits (A).
ii. Two 1.22 meter (4 feet) deep pits with areas of 3.05m X 3.05m (10ft X 10ft) (B)
and 10.06m X 6.40m (33ft X 21ft) (C), respectively.
iii. One 2.44 meter (8 feet) deep pit with area 6.4m X 6.4m (21ft X 21ft) (D). This
pit includes a simulation of a high dike wall, typical of the containment facilities
used during LNG offloading.
The Brayton Field LNG facility is equipped with a 20.3cm (8") water pipe loop.
Sufficient water supply was available during the test, from the city water supply, with the
aid of a pump powered by a pumper truck. The pumper truck could deliver up to 3.8
m3/min (1000 USGPM) water at 130psi across an isolating block valve to supply the
water curtain units. The facility is installed with fire water monitors and removable water
spray heads.
D
C
B
A
52.43m long, 76.2 mm (3") OD, insulated LNG Discharge Pipe (Al)
LNG Truck Position
Water Curtain Test
Area
Pumper Truck& Pump LocationWater Drainage
20.3 cm (8") Water Line
Water Monitor/Hydrant
50.8 mm (2") Water Discharges, Valves & Water Spray Heads
N
safe area outside the boundary
LNG Discharge Line
Figure 1 Test facility and spill location
3.2 Spill Area LNG spill tests were conducted on a flat concrete pad of the LNG training facility. The
concrete spill area was enclosed with 1.52m × 1.52m (5ft × 5ft) inner and 1.83m × 1.83m
(6ft × 6ft) outer wooden frames. To minimize the LNG flow through the gap between the
wooden frame and the concrete ground, the space between the frames were filled with
wet sand during the tests. The intension of using wet sand was that when LNG contacted
wet sand it would freeze immediately and stop liquid LNG flow out of the spill area
through the gaps. Figure 2 shows the photograph of the enclosed spill area.
Figure 2 Enclosed spill area
3.3 Source and Flow Measurements LNG was supplied by a tanker truck of 41.64 m3 (11,000 gal) capacity and discharged at
the spill location with 52.4 m (172 ft) long, 7.6 cm (3 inch) OD, 3.2 mm (1/8 inch) thick
insulated, fixed Aluminium pipe line and around 15.2 m (50 ft) long, 7.6 cm (3 inch) OD,
flexible discharge hose. The flow rate of the LNG from the tanker was manipulated by
manually controlling the opening of the outlet valve from the tanker to achieve a steady
flow rate. The flow rate (in GPM) was continuously measured with a cryogenic flow
meter connected between the pipe line and the discharge hose during the tests. The
cryogenic flow meter used was a 3-inch FTB-911 turbine meter with male NPT end
fittings.
3.4 Water Curtain and Flow Measurements Two types of portable spray curtains were employed and both were upward directed. One
was constructed of seven 1" TF 48 NN nozzles, connected to a 6.1 m (20 ft) long 5.1 cm
(2 inch) OD carbon pipe. These nozzles are manufactured by BETE Fog Nozzles and
produce a full cone, spiral, 60º spray pattern. According to the manufacturer, each nozzle
can throw water upward as high as 4.6 m (15 ft) at 100 psi and 0 wind speed. Nozzles can
create smaller droplets of sauter mean diameter 580 µm. Prior tests showed that the water
curtain made with the nozzles created 7 m (23 ft) wide, 3.7 m (12 ft) high coverage.
Figure 3 illustrates the nozzle and water spray curtain.
Figure 3 Photographs of 1" TF 48 NN nozzle and water spray curtain
The other spray instrument was the Hydro-Shield spray device available through the
Brayton Fire School. With this spray head water flows out through a pipe, which has half
circular end, and hits a flat plate at the exit. The obstacle created by the flat plate
produces 180º flat fan shaped spray pattern. Initially it creates a sheet of water at the pipe
exit and then breaks up as larger water droplets. This device is capable of projecting a
water curtain covering about 15.2 m (50 ft) wide and 7.6 m (25 ft) high according to
visual observation. Figure 4 shows the hydro-shield water spray.
Figure 4 Hydro-shield water spray curtain
During the tests, water was supplied from the water line to the water spray curtain with a
6.4 cm (2½") OD fire hose. A flow meter and a pressure gauge were connected at the exit
of the fire hose to measure the water flow rate and water pressure at the inlet of the
curtain during the tests. Water flow was turned on and off with the ball valve already
connected to the water supply line.
3.5 Concentration and Temperature Measurements 34 IR point gas detectors and 32 type K thermocouples were used and installed on 14
tripods at different heights to measure the methane gas concentration at 0.5m, 1.2m and
2.1m above ground and temperature of air-methane mixture at 0.5m, 1.0m and 2.0m
above ground. The heights were chosen based on the previous research. The tripods were
placed at different downwind distances from the LNG spill area to measure the
concentration and temperature both upstream and downstream of the water curtain. The
instruments were placed according to the predicted wind direction of the experiment day.
Table 1 shows the position of tripod poles and gas detectors and temperature sensors
heights on the poles.
Table 1 Location of 14 tripod poles and 36 gas detectors and 32 temperature sensors
positions on the poles
Pole no.
y m
x m
Gas Detector, GD Temperature Sensor, TS
z =0.5 m z = 1.2 m z = 2.1 m z = 0.5 m z = 1.0 m z = 2.0 m
1 1.2 1
22ª 21 --- 3¹ 4 --- 2 -1.2 24 23 --- 1 2 --- 3 3.7
3.3
25 26 36 5 6 7 4 1.2 27 28 35 8 9 10 5 -1.2 30 29 39 11 12 13 6 -3.7 38 37 40 14 15 16 7 1.2 5.5 --- --- --- 29 30 --- 8 -1.2 --- --- --- 31 32 --- 9 3.7
11.3
13 12 11 17 18 19 10 1.2 01 15 07 20 21 22 11 -1.2 19 18 17 23 24 25 12 -3.7 20 02 10 26 27 28 13 1.2 13.7 06 34 04 --- --- --- 14 -1.2 08 32 31 --- --- ---
x = downwind distance from the edge of the enclosed spill location z = height above the ground ª = gas detector identification number ¹ = temperature sensor identification number
Figure 5 and 6 shows the experimental set up and the sensors, detectors and water curtain
placements.
Figure 5 Experimental set-up
Figure 6 Photograph of equipment placement
The type-K thermocouples, manufactured by Omega Engineering, were capable of
measuring temperature as low as 173K (-100ºC) up to 373K (100ºC) up to. The gas
detectors used were designed to detect methane and calibrated to measure the
concentration in % v/v unit. These electric powered IR point gas detectors were
manufactured by Honeywell Analytics. The detectors were not physically installed on the
tripods, but were installed in a wooden vacuum chamber. 30.5m (100 ft) long Teflon
sample tubes were used for the purpose of flowing methane from the field to the
detectors. One end of each tube was connected to a gas detector and the other end was
placed on the tripod pole. A vacuum pump was used to pull LNG vapors at 0.3 L/min to
the detector through the tubes and then released to the environment. Figure 7 and 8 shows
the gas detector connection and photograph of detector chamber.
Figure 7 IR Gas Detector-Teflon tube connection
Figure 8 Gas detector chamber
3.6 Other Equipment The meteorological conditions were measured during the experiments with three weather
stations. Two stations were placed in two separate positions of the test facility, 2.1m (7 ft)
above ground-level and the third was installed around 15.2m (50 ft) east of the test
ground, 9.1m (30 ft) above. Purpose of these stations was to measure the wind velocity,
direction, temperature, humidity and heat index.
Two hydrocarbon (H/C), one IR and two regular digital cameras were used to capture the
whole experiment. The hydrocarbon and IR cameras were used to observe the actual
dispersion of invisible flammable hydrocarbons (mainly methane). The H/C cameras
were placed away at a safer location from the visible cloud region and at 90º to each
other, so that they can capture all the moments.
Gas detector, temperature sensor, flow meter and pressure gauge data were acquired at an
interval of 1 second with a data acquisition system and recorded simultaneously in two
computers. Data receiving and acquisition systems were installed carefully to receive
signals, transform signals to data and record data successfully. The weather stations have
built-in data record system, which could save data for three days. The weather data were
transferred to computers after the experiments. All of the detectors, sensors and gauges
were calibrated and checked several times prior to the test day.
3.7 Experimental Procedure Four similar spill tests were conducted to evaluate the effectiveness of two types of water
spray curtains in controlling an LNG vapor cloud. Each curtain was tested twice with a
separate spill. The first two tests were conducted with a full cone spray curtain. The water
spray curtain system was then replaced with the Hydro-Shield flat fan spray and two
more similar tests were conducted.
During the tests, water curtains were directed vertically upward and positioned
perpendicularly to the prevailing wind direction, at a fixed downwind distance from the
spill location. All of the tests consist of two measurements, the natural and forced
dispersion. During the natural dispersion no action was taken to control the LNG vapor
flow. Forced dispersion of the LNG vapor occurred in the presence of water spray curtain
application.
Each test started with continuous release of liquid LNG onto the enclosed spill area.
Initially only LNG vapor started to emerge from the hose and within a couple of minutes
liquid stared to flow from the hose. As soon as the liquid touched the concrete ground it
vaporized and dispersed naturally towards the prevailing wind direction. Several minutes
after the start of the LNG release, the water spray curtain was turned on to disperse the
cloud forcefully. The flow of LNG was also continued for a few minutes during forced
dispersion. When the LNG pool reached about 5 cm to 8 cm (2" to 3") height, the LNG
flow was turned off while continuing the water curtain operation. The curtain was kept on
until the visible white cloud disappeared.
4 RESULTS AND DISCUSSION
A total of four tests were completed in a 1½ hour period. During the tests, the wind speed
averaged 2 meters per second (4.5 miles per hour) and blew mainly from the south-west
to south-east region. The average ambient temperature during the tests was 295.4K and
humidity was 25.5%.
Table 2 includes the average flow rates and total flow time of LNG and water and
average pressure of water at the curtain inlet for each test.
Table 2 Average flow rate and pressure
Test
Water Curtain LNG Flow Water Flow
Spray Flow
Pattern
Avg.
Rate
m3/s
(GPM)
Total
Duration
min
Avg.
Rate
m3/s
(GPM)
Avg.
Pressure
psi
Total
Duration
min
1 BETE
Fog
Full Cone
Spiral 2.5×10-3
(39.3) 9.0
10.1×10-3
(160.8) 40 9.8
2 BETE
Fog
Full Cone
Spiral 3.8×10-3
(60.7) 8.2
15.5×10-3
(245.8) 45 5.8
3 Hydro-
Shield Flat Fan 3.5×10-3
(55.9) 5.9
11.3×10-3
(178.6) 79.5 4.3
4 Hydro-
Shield Flat Fan 3.4×10-3
(54.8) 3.5
15.1×10-3
(240.1) 99.7 3.4
All tests were video taped using digital, hydrocarbon and Infrared (IR) cameras. The
hydrocarbon and IR cameras were used to differentiate between the dispersion of
flammable hydrocarbon and the visible condensate cloud. The following, Figures 9–11,
show clips from the videos of the tests captured with the regular and hydrocarbon
cameras, almost at the same time.
(a) (b)
Figure 9 Spill and natural dispersion: (a) regular image (b) hydrocarbon camera
image
(a) (b)
Figure 10 Forced dispersion with full cone spray curtain: (a) regular image (b)
hydrocarbon camera image
(a) (b)
Figure 11 Forced dispersion with flat fan curtain: (a) regular image (b)
hydrocarbon camera image
In Figure 9 the regular image shows a smaller visible condensate cloud but hydrocarbon
camera shows actual size of the flammable cloud. In Figure 10 and Figure 11, the regular
camera image shows that white condensate cloud becomes invisible after it crosses the
spray area. The black areas of the hydrocarbon camera image show the presence of colder
hydrocarbon in the gas-air-water vapor cloud. The light darker area after the spray
indicates that still there are some colder hydrocarbons (which are not visible) and the
concentration/temperature is lower.
The gas detectors were calibrated with methane gas (50-50% CH4-Air Mixture) to
measure methane concentration in % volume/volume unit. However, during the tests one
gas detector (GD 18) malfunctioned and did not show any reliable readings the entire
time. The other six detectors (GD 7, 8, 11, 34, 2 and 17), provided reliable data for tests 1
and 2, for a portion of test 3, but failed during test 4. A potential cause for this failure
might have been caused by water going in the Teflon tubes of the detectors; thus,
blocking the path of methane from flowing to the detectors. Figure 12 shows some gas
detector readings at different locations.
Concentration Reading 0.5m Above Ground and Different Downwind Distances
Test 1
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
27.5
30.0
0 60 120 180 240 300 360 420 480 540 600 660 720 780 840 900 960time, sec
CH
4 co
ncen
trat
ion,
% v
/v
GD24GD30GD19
Figure 12 Gas detector reading at different downwind locations [Position of gas
detector 24 ,GD 24 ≡ (x, y): (1m, 1.2m); GD 30 ≡ (x, y): (3.3m, 1.2m); GD 19 ≡ (x,
y): (11.3m, 1.2m)]
The fluctuations in the methane concentration readings from the gas detector were caused
by the wind gustiness and crosswind turbulence. More fluctuation was observed close to
the spill location due to the turbulence created by the momentum of the spill. The water
curtain was positioned 7m (23 ft) downwind from the spill location. The wind’s frequent
change of direction sometimes allowed the LNG vapor to disperse without passing
through the water spray region and caused fluctuations in the methane concentration
measurements. Therefore, to analyze the gas concentration data, the average of
concentration readings of all the detectors placed at certain downwind distance and
heights are considered. Figure 13 illustrates the method for averaging gas concentration
data.
Figure 13 Average concentration at different heights and downwind positions
As there was time lag between the LNG flow from the pipe and gas detector reading, all
the data were edited to synchronize between the videos, temperature sensor and gas
detector readings. The following charts summarize the average concentration reading:
3.3m (11 ft) and 11.3m (37 ft) downwind from the spill; before and after the water curtain
action; at different heights (0.5m, 1.2m and 2.1m) for two tests with two types of water
curtain.
Average Concentration vs TimeHeight (z=) 0.5m and Distance from the Release (x=) 3.3m & 11.3m
Test 1: Spray 7m from the release
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
0 60 120 180 240 300 360 420 480 540 600 660 720 780 840 900 960
time, sec
CH
4 co
ncen
trat
ion,
%v/
v Before the spray location
After the spray locationat x = 11.3 m
LNG release discontinued
Water turned on
Figure 14 Average gas concentration 3.3m and 11.3m from the spill, at 0.5m above
ground for test 1; (full cone water curtain position: 7m downwind from the spill)
Average Concentration vs Time
Height (z=) 1.2m and Distance from the Release (x=) 3.3m & 11.3mTest 1: Spray 7m from the release
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
0 60 120 180 240 300 360 420 480 540 600 660 720 780 840 900 960
time, sec
CH
4 co
ncen
trat
ion,
%v/
v
LNG release discontinued
Water turned on
Before the spray location at x = 3.3 m
After the spray locationat x = 11.3 m
Figure 15 Average gas concentration 3.3m and 11.3m from the spill, at 1.2m above
ground for test 1; (full cone water curtain position: 7m downwind from the spill)
Average Concentration vs TimeHeight (z=) 2.1m and Distance from the Release (x=) 3.3m & 11.3m
Test 1: Spray 7m from the release
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
0 60 120 180 240 300 360 420 480 540 600 660 720 780 840 900 960
time, sec
CH
4 co
ncen
trat
ion,
%v/
v
LNG release discontinued
Water turned on
Before the spray location at x = 3.3 m
After the spray locationat x = 11.3 m
Figure 16 Average gas concentration 3.3m and 11.3m from the spill, at 2.1m above
ground for test 1; (full cone water curtain position: 7m downwind from the spill)
Average Concentration vs TimeHeight (z=) 0.5m and Distance from the Release (x=) 3.3m & 11.3m
Test 3: Spray 7m from the release
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
0 60 120 180 240 300 360 420 480 540 600
time, sec
CH
4 co
ncen
trat
ion,
%v/
v
LNG release discontinued
Water turned on
Before the spray location at x = 3.3 m
After the spray locationat x = 11.3 m
Figure 17 Average gas concentration 3.3m and 11.3m from the spill, at 0.5m above
ground for test 3; (flat fan water curtain position: 7m downwind from the spill)
Average Concentration vs TimeHeight (z=) 1.2m and Distance from the Release (x=) 3.3m & 11.3m
Test 3: Spray 7m from the release
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
0 60 120 180 240 300 360 420 480 540 600
time, sec
CH
4 co
ncen
trat
ion,
%v/
v
LNG release discontinued
Water turned on
Before the spray location at x = 3.3 m
After the spray locationat x = 11.3 m
Figure 18 Average gas concentration 3.3m and 11.3m from the spill, at 1.2m above
ground for test 3; (flat fan water curtain position: 7m downwind from the spill)
Average Concentration vs TimeHeight (z=) 2.1m and Distance from the Release (x=) 3.3m & 11.3m
Test 3: Spray 7m from the release
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
0 60 120 180 240 300 360 420 480 540 600
time, sec
CH
4 co
ncen
trat
ion,
%v/
v
LNG release discontinued
Water turned on
Before the spray location at x = 3.3 m
After the spray locationat x = 11.3 m
Figure 19 Average gas concentration 3.3m and 11.3m from the spill, at 2.1m above
ground for test 3; (flat fan water curtain position: 7m downwind from the spill)
During natural dispersion, before the water curtain action, concentrations at 3.3m
downwind of the spill and 0.5m above the ground are higher than concentrations at 1.2m
and 2.1m height for both the tests. Horizontal and vertical momentum of LNG flow from
the pipe, flashing and evaporation from the LNG pool caused some turbulence close to
the spill area. Because of these data still shows fluctuation at this position. At 11.3m
downwind, concentration near the ground (0.5m) becomes less than the higher level
(1.2m and 2.1m) concentrations after a period of time. This clearly indicates that LNG
vapor naturally dilutes and disperses upward. This dispersion process illustrates that the
negative buoyancy of an LNG vapor cloud changes to neutral and then to positive
buoyancy as it travels downwind.
The forced dispersion process occurs when the water curtain is turned on. As the LNG
release continued for some time after turning the water spray on, the average methane
concentrations at 3.3m expectedly continued to increase. However, the average
concentration readings at 11.3m downwind distance behaved differently, for both the
tests, during the water curtain action. For test 1, the full cone test, concentrations at 1.2m
and 2.1m heights started decreasing behind the spray position (11.3m) immediately after
the spray was turned on. The concentration reading at 0.5m height of this location
indicates a reduction in the rate of concentration increase. For the flat fan test,
concentration rate increase was reduced at 11.3m downwind distance during the water
curtain action when the LNG was still flowing from the pipe.
In both tests, the gas concentration continued to decrease at both distances during the
water curtain application after discontinuing the LNG flow. The differences between
3.3m and 11.3m downwind curves show the change in methane concentration behavior
due to the water curtain application for both tests. However, in all of the tests the methane
concentration at 11.3m downwind never reached below one half of the LFL, 2.5% (v/v),
by forced dispersion..
Tests with the full-cone curtain showed the average gas concentration at 11.3m
downwind of the release is less than the concentration at 3.3m downwind before and after
the curtain action for 0.5m and 1.2m heights. But, concentration 2.1m above ground at
11.3m downwind is higher than 3.3m downwind before the water action and decreases
after the curtain was turned on. During the tests, visually, one was able to observe that the
constructed full-cone-spiral-nozzle curtain created more turbulence near the spray region
than the flat-fan-spray curtain. However, the full cone spray curtain could not provide
enough crosswind and above ground coverage to interact completely with the incoming
vapor cloud. The crosswind coverage of full cone water spray was smaller than the flat
fan spray which allowed the LNG to pass around the curtain and disperse. For this test,
the methane concentration after the curtain region (11.3m) did not decrease significantly.
Again, the difference in LNG concentrations before (3.3m) and after (11.3m) the full
cone spray region at 2.1m high was not very significant as the vapor cloud could crossed
over the top of the water curtain without complete interaction. This result indicates that
this constructed water curtain provides more mixing effect than momentum effect.
Average gas concentration values at 11.3m downwind distance were observed to be
lower for the fan type spray than the full cone spray. Concentration profiles at 0.5m
above ground at 11.3m and 3.3m downwind show significant differences in the flat fan
spray test. At 1.2m above ground, average concentrations at 3.3m downwind are higher
than 11.3m before and during water curtain application. Concentrations at 2.1m above
ground and at 11.3m downwind are higher than at 3.3m, both before and during the entire
period after the water curtain application. This indicates that the flat fan water curtain
was dispersing the LNG vapor upward through momentum effect more than mixing and
diluting it with air. Figure 20 describes the visual coverage and interpreted interactions of
two types of water curtains with vapor cloud during the tests.
Figure 20 Spray-cloud interaction and water curtain coverage
5 CONCLUSION
This paper analyzed the experimental concentration data to determine the effect of water
curtains in changing LNG vapor concentration. The tests conducted in this research
indicate that water curtain is able to control LNG vapor cloud by changing the ground
level methane concentration. However, the tests could not show a significant change in
concentration at higher level than the ground level. The constructed full cone spray shows
more effectiveness in creating turbulence closer to the spray region than the flat fan
spray. The flat fan is more effective in reducing the ground level concentration by
creating a barrier in the path of the cloud and pushing the cloud upward. From the
analysis, it is clear that the action of water sprays reduced the methane concentration by
dilution and mixing with entrained air as well as pushed the vapor upward. More analysis
on the heat transfer effect of water spray on cold LNG cloud, which is ongoing, will help
reach more decisive quantitative results. Each test had to stop within a short time because
amount of the large amount of water used would cause problems to some of the
instruments located near the test area. This also indicates that to control a large spill the
required system has to provide sufficient water. In addition, draining the large amounts of
water can present problems. However, to establish a definitive engineering guideline on
water curtain design for LNG cloud dispersion, controlled experiments for longer times
and with different wind conditions are needed.
Acknowledgement
The authors are grateful to BP Global Gas SPU for their financial support of this study.
They would also like to thank the Brayton Fire Training School and their LNG
Emergency Response training team for their help and contributions to this research.
LITERATURE CITED
1. Center for Liquefied Natural Gas (CLNG), http://www.lngfacts.org/LNG-
Today/Need-More.asp (Accessed December 2007).
2. Federal Energy Regulatory Commission (FERC), http://ferc.gov/for-citizens/citizen-
guides/citz-guide-lng.pdf (Accessed August 2008).
3. D.T. Uzanski and J.M. Buchlin Mitigation of Industrial Hazards by Water Spray
Curtains, The 5th Conference of the International Emergency Management Society
(TIEMS), Washington, DC, (May 1998), 19-22.
4. K. Hald, J. M. Buchlin, A. Dandrieux and G. Dusserre, Heavy gas dispersion by
water spray curtains: A research methodology, J. Loss Prevention in the Process
Industries, 18 (2005), 506-511.
5. G. Dusserre, A. Dandrieux and O. Thomas, The DVS Model: a New Concept for
Heavy Gas Dispersion by Water Curtain, Environmental Modeling and Software, 18
(2003), 253-259.
6. P. A. C. Moore and W. D. Rees, Forced Dispersion of Gases by Water and Steam,
North Western Branch Papers, IChemE, (1981), 4.1-4.14.
7. J. M. Buchlin, Mitigation of Problem Clouds, J. Loss Prevention in the Process
Industries, 7 (1994), 167-174.
8. V. M. Fthenakis, HGSPRAY: A Complete Model of Spraying Unconfined Gaseous
Release, J. Loss Prevention in the Process Industries, 6 (1993), 327-331.
9. V. M. Fthenakis and D. N. Blewitt, Recent Development in Modeling Mitigation of
Accidental Releases of Hazardous Gases, J. Loss Prevention in the Process
Industries, 8 (1995), 71-77.
10. J. McQuaid, and R. D. Fitzpatrick, Air Entrainment by Water Sprays: Strategies for
Application to the Dispersion of Gas Plumes, J. Occupational Accidents, 5 (1983),
121-133.
11. K. Moodie, Experimental Assessment of Full Scale Water Spray Barriers for
Dispersing Dense Gases, Institution of Chemical Engineers Symposium on
Containment and Dispersion of Flammable and Toxic Gases with Water sprays,
Manchester, England, (November 1981).
12. K. Moodie, The Use of Water Spray Barriers to Disperse Spill in the Event of Heavy
Gases – The Performance Characteristics of Full-Scale Water Spray Barriers when
Dispersing Spills of Heavy Gases, Plant Operation Progress, 4 (1985), 234-241.
13. J. McQuaid, and R. D. Fitzpatrick, The Uses and Limitations of Water-Spray
Barriers, North Western Branch Papers, IChemE, (1981), 1.1-1.13.
14. D. N. Blewitt, R. L. Petersen, M. A. Ratcliff and G. Heskestad, Evaluation of Water
Spray Mitigation System for an Industrial Facility, International Conference and
Workshop, Modeling and Mitigating the Consequence of Accidental Releases of
Hazardous Materials, (1991), 483-510.
15. K. Hald, J. M. Buchlin, A. Dandrieux and G. Dusserre, A Methodology to Investigate
Heavy Gas Dispersion by Water-Curtains, Proceedings of European Safety and
Reliability (ESREL) Conference, Maastricht, Netherlands, (June 2003), 741-746.
16. L. E. Brown, Small Scale Tests on Control Methods for Some Liquefied Natural Gas
Hazards, Coast Guard Report No CG-D-95-76. May, 1976.
17. W. E. Martinsen and S. P. Muhenkamp, Disperse Vapors with Water, Hydrocarbon
Processing, (1977), 260-267.
18. R. G. Zalosh, R. L. Alpert and G. Heskestad, Dispersal of LNG Vapor Cloud with
Water Spray Curtains, North Western Branch Papers, IChemE, (1981), 3.1-3.17.
19. G. Heskestad, R. N. Meroney, and K. M. Kothari, Effectiveness of Water Spray
Curtains in Dispersing LNG Vapor Clouds, Proceedings of the American Gas
Association, Transmission Conference Paper, 83-T-69, Seattle, Washington, 2-4 May
1983, T169-T183.
20. N. C. Harris, The Design of Effective Water Sprays – What We Need To Know,
North Western Branch Papers, IChemE, (1981), 7.1-7.5.