Modeling Software for Safety Analysis of LNG Facilities
1. Modeling Software for EHS Professionals
PERFORMING SAFETY MODELING ANALYSIS TO
COMPLY WITH LNG FACILITY SITING REQUIREMENTS
Prepared By:
Weiping Dai
Rob J. Liles
Erwin T. Prater
BREEZE SOFTWARE
12700 Park Central Drive,
Suite 2100
Dallas, TX 75251
+1 (972) 661-8881
breeze-software.com
2. PERFORMING SAFETY MODELING ANALYSIS TO COMPLY WITH LNG
FACILITY SITING REQUIREMENTS
Weiping Dai
Rob J. Liles
Erwin T. Prater
Trinity Consultants
12770 Merit Drive, Suite 900
Dallas, Texas 75251
ABSTRACT
Natural gas (NG) is an important industrial and residential energy source. Growing domestic
demand is expected to increase NG imports in the future. Natural gas liquefies to become Liquefied
Natural Gas (LNG) at –260 degrees Fahrenheit (°F) at ambient atmospheric pressure. Liquefaction
reduces the volume of the gas phase by a factor of 600 and storage at ambient pressure makes LNG
feasible for transport. However, re-gasification, storage, and handling facilities are needed in the
supply chain to convert LNG back to gas. These facilities must comply with many regulatory
requirements to ensure safe operation. Safety modeling analysis is required for LNG facilities with
recommended models and techniques. Part 193 of Title 49 in the U.S. Code of Federal Regulations
(CFR) requires analyses of both thermal radiation protection and flammable vapor-gas dispersion
protection as the siting requirements for applicable LNG facilities. The National Fire Protection
Association (NFPA) 59A also specifies the standard for the production, storage, and handling of LNG.
Performing safety modeling analyses for LNG facilities requires consideration of many factors,
including site configuration, operating conditions, model selection, and appropriate modeling
techniques. The unique chemical and physical properties of LNG must be considered in analyses.
LNG releases could occur on land or over water, and under a variety of conditions that may
complicate the technical analyses. This paper describes appropriate models and technical details for
developing modeling analyses that meet LNG facility siting requirements. The widely-used LNG-
specific source-term model (Source5), DEGADIS air-dispersion model, and LNGFIRE III thermal
radiation model are discussed in detail. Findings from this study demonstrate that evaluating the
effects of various conditions (e.g., wind speed, ambient temperature, relative humidity, atmospheric
stability, and/or surface roughness) will be necessary in determining the maximum exclusion distances
for both the thermal radiation and the vapor dispersion. The analyst has to be particularly attentive to
roughness length, wind speed, and atmospheric stability when performing analyses that support LNG
facilities.
3. PERFORMING SAFETY MODELING ANALYSIS TO COMPLY WITH LNG
FACILITY SITING REQUIREMENTS
INTRODUCTION
Natural gas (primarily methane) is a growing industrial and residential energy source. It
liquefies at –260 degrees Fahrenheit (°F) under ambient atmospheric pressure. Since the volume of
liquefied natural gas (LNG) is only about 1/600 of its gas volume, LNG is the most economical way to
transport large volumes of natural gas over long distances. However, re-gasification, storage, and
handling facilities are needed in the supply chain to convert LNG back to a gas phase.
LNG facilities are regulated by Part 193 of Title 49 in the U.S. Code of Federal Regulations (49
CFR Part 193) that prescribes the safety standards for LNG facilities used in the transportation of gas
by pipeline. The rule applies to the siting, design, installation, or construction of LNG facilities.
Furthermore, the National Fire Protection Association (NFPA) 59A Standard also specifies the
standard and requirements for the production, storage, and handling of LNG. In particular, each LNG
facility that is designed, built, replaced, re-located, or significantly altered after March 31, 2000 must
be provided with siting requirements in accordance with the requirements of both 49 CFR 193 and the
NFPA 59A Standard in order to ensure safe operation. For example, 49 CFR Part 193 requires
analyses of both thermal radiation protection and flammable vapor-gas dispersion protection as the
siting requirements for applicable LNG facilities. Similar analyses are also applicable in developing a
fire prevention plan for LNG facilities. Specifically, a thermal exclusion zone and a vapor exclusion
zone must be determined for each LNG container and transfer system pursuant to the design spills in
accordance with the NFPA 59A Standard. Moreover, each LNG facility operator must provide and
maintain the fire protection at LNG facilities pursuant to the NFPA 59A Standard. In siting the
impounding areas, it is necessary to perform technical analyses to determine the thermal radiation
distances and the flammable vapor-gas dispersion distances. [1]
Safety modeling analysis for LNG facilities can be performed using recommended computer
models and techniques. Performing safety modeling analysis for LNG requires consideration of many
factors including site configuration, operating conditions, model selection, and modeling technique and
options. Moreover, the unique chemical and physical properties of LNG must be considered in the
analysis. Potential release scenarios of LNG could occur on land, or over water, and under a variety of
conditions. All these factors complicate the technical analysis of LNG releases for safety concerns.
This paper discusses the regulatory requirements and standards related to the LNG facility safety
analyses and explores the preferred models/techniques for performing such analyses. Technical
details for utilizing selected models with appropriate techniques and parameter values are also
discussed in order to comply with the LNG facility siting requirements specified in the regulations. In
particular, the recommended computer programs such as the LNG-specific source model (Source5),
the DEnse GAs DISpersion model (DEGADIS) air dispersion model, and LNGFIRE III thermal
radiation model are discussed in detail.
DETERMINATION OF THERMAL RADIATION DISTANCES
Thermal radiation distances (also called thermal exclusion distances) are defined as the
distances from an LNG fire to the locations beyond which thermal radiation fluxes do not exceed
4. specific limits. These limits are specified in §5.2.3.2 of NFPA 59A. The following thermal radiation
flux limits are defined in the NFPA 59A Standard for atmospheric conditions of calm winds (zero wind
speed), 70 °F (21 °C) temperature, and 50% relative humidity:
• 1,600 Btu/hr-ft2
(5,000 W/m2
) flux for exposure at a property line that can be built upon for
ignition of a design spill or at the nearest point located outside the owner’s property line that, at
the time of plant siting, is used for outdoor assembly by groups of 50 or more persons for a fire
in an impounding area.
• 3,000 Btu/hr-ft2
(9,000 W/m2
) flux for exposure at the nearest point of the building or structure
outside the owner’s property line that is in existence at the time of plant siting and used for
assembly, educational, health care, detention and correction, or residential occupancies for a
fire in an impounding area.
• 10,000 Btu/hr-ft2
(30,000 W/m2
) flux for exposure at a property line that can be built upon for a
fire over an impounding area.
Pursuant to the NFPA 59A Standard, the following three methods are allowed to determine the
thermal radiation distances:
• Use LNGFIRE III computer program developed by the Gas Research Institute (GRI) based on
the GRI Report No. 89/0176. (Note: GRI has recently changed its name to the Gas
Technology Institute - GTI).
• Use a validated model incorporating various technical considerations and accepted by
regulatory agency for use.
• Use the formula given in the standard for an impoundment whose ratio of the major and minor
dimensions does not exceed 2.
In calculating the thermal radiation exclusion distances, the atmospheric ambient conditions
(wind speed, ambient temperature, and relative humidity) should be determined to produce the
maximum thermal radiation distances (or exclusion distances) except for the conditions that occur less
than 5% of the time based on recorded data for the area.
OVERVIEW OF LNGFIRE III COMPUTER PROGRAM
The LNGFIRE III program implements GRI Report No. 89/0176, which is referenced in the
NFPA 59A Standard. LNGFIRE III has been validated with field experiments. The mathematical
approach used in LNGFIRE III assumes that a flame behaves as a gray-body cylinder emitting thermal
radiation from its surface. LNGFIRE III calculates the thermal radiation flux level from an LNG fire in a
tank or dike on various vertically- and horizontally-oriented targets that may be above, or below, ground
level. The program also calculates the maximum thermal flux at user-specified points. When the
horizontal and vertical targets are in full view of the flame, the maximum flux is calculated as the vector
sum of the fluxes to the vertical and horizontal targets. If either or both targets see a fraction of the flame,
the maximum flux is found by starting at the smallest angle where the target sees the whole flame and by
tilting the target at 1° interval until the maximum flux is found. The program allows for the attenuation of
thermal radiation due to the presence of water vapor in the atmosphere. A relative humidity of 0% may be
entered to remove the attenuation of thermal radiation by water vapor and produce conservative results.
5. LNGFIRE III also considers flame tilt from the effects of the wind and the effects of wind drag on
flames when the flame base is within 3 ft (1 m) of the ground. The program can determine radiation
fluxes for both circular and rectangular pool fires. In the case of rectangular pools, the wind is assumed to
be perpendicular to the side of interest. Because the maximum emissive power and the tilt of the flame
depend on the direction from which the wind is blowing, LNGFIRE III calculates two results (views): one
wherein the wind blows perpendicular to the long side (front view); and the other wherein the wind blows
perpendicular to the narrow side of the rectangle (side view). LNGFIRE III calculates the exclusion zones
to various radiation flux levels. The angle of tilt of the flame is dependent on the user’s input of prevailing
wind speeds at the site of interest. To produce conservative estimates of the thermal exclusion zone, the
user must select various ambient conditions (e.g., wind speed, ambient temperature, and relative humidity)
and select the condition that gives the largest exclusion zone for the radiation flux of interest.[2]
EFFECTS OF AMBIENT CONDITIONS ON THERMAL RADIATION:
PARAMETRIC ANALYSIS
As discussed above, the analyst varies the ambient wind speed, ambient temperature, and
relative humidity parameters to determine the longest thermal radiation distances. This often raises the
question as to what parameters drive the calculated thermal radiation distances. This study used the
LNGFIRE III program to explore this question by examining modeled results from three hypothetical
pool fires. Table 1 lists the pool sizes (i.e., small, medium, and large) considered.
Table 1 - Pool Fire Parameters
Fire Size Pool Diameter (ft) Flame Base Height (ft)
Large 300 10
Medium 100 5
Small 10 1
Figure 1 shows the variation of the thermal radiation distance with the wind speed for the large
pool described in Table 1. Figure 1 shows that the thermal exclusion distances for the thermal flux
levels in NPFA 59A Standard are insensitive to low wind speeds (speeds less than 5 mph). The figure
also shows that the exclusion distances increase sharply in the 5-15 mph range. At higher wind speeds,
the exclusion distances decrease slightly for the two lower exposure levels (e.g., 1,600 Btu/hr-ft2
and
3,000 Btu/hr-ft2
) and level off gradually for the highest flux level (e.g., 10,000 Btu/hr-ft2
).
Figure 2 shows the predicted thermal radiation flux as a function of distance for a large pool
fire as a function of wind speed. The figure shows that the thermal radiation levels vary inversely (and
strongly) with distance at all wind speeds. The spatial rate of decrease is larger at higher wind speeds.
Similar behavior is shown for medium and small pool fires (results not shown for brevity).
Figures 3 and 4 demonstrate the sensitivity of the thermal exclusion distances to ambient
temperature and relative humidity levels. Results show that the thermal exclusion distance is virtually
insensitive to variations of both parameters.
Figures 1-4 suggest that the thermal exclusion distances are most sensitive to variations in the
wind speed, suggesting that the analyst should pay particular attention to the treatment of wind speed
in parametric analyses that support LNG-related regulatory modeling. Similar results are observed for
6. all fire sizes with respect to the effects of the wind speed, ambient temperature, and relative humidity
on the thermal radiation.
0
200
400
600
800
1000
1200
0 10 20 30 40 50
Wind Speed (mph)
ThermalRadiationDistance(feet)_
10,000 Btu/hr-ft2
3,000 Btu/hr-ft2
1,600 Btu/hr-ft2
Figure 1 - Variation of Thermal Radiation Distance with Wind Speed (Large Pool)
10
100
1,000
10,000
0 1000 2000 3000 4000 5000 6000
Distance from Pool Center (feet)
ThermalRadiation(Btu/hr-ft2)_
WS = 5 mph
WS = 15 mph
WS =30 mph
Figure 2 - Variation of Thermal Radiation Level with Distance from Fire (Large Pool)
8. DETERMINATION OF FLAMMABLE VAPOR-GAS DISPERSION DISTANCES
As LNG spills into a dike or impounding area, it evaporates as a flammable vapor and disperses
into the ambient air. The flammable vapor-gas dispersion distances (vapor exclusion distances) should
be established in such a way that, in the event of an LNG spill, an average methane concentration in air
will not exceed 50% of the lower flammability limit (LFL), i.e., 2.5% methane, at the property line.
NFPA 59A allows the analysis to be conducted with the model described in the GRI Report No.
89/0242.[3]
Alternatively, in order to account for additional cloud dilution that may be caused by the
complex flow patterns induced by tank and dike structure, dispersion distances may be calculated in
accordance with the model described in the GRI Report No. 96/0396.5 entitled “Evaluation of
Mitigation Methods for Accidental LNG Releases. Volume 5: Using FEM3A for LNG Accident
Consequence Analyses”. The use of alternate models that take into account the same physical factors
and have been validated by experimental test data could be permitted subject to the regulatory
agency’s approval.
The analysis should include calculations based on the combination of wind speed and
atmospheric stability that can occur simultaneously and result in the longest predictable downwind
dispersion distance that is exceeded less than 10% of the time. Alternatively. “F” atmospheric stability
and 4.5 miles per hour (2 m/s) wind speed can be used. In addition, the analysis should be based on
the actual liquid characteristics and the maximum vapor outflow rate from the vapor containment
volume (i.e., the vapor generation rate plus the displacement due to liquid flow).
In summary, the following dispersion parameters must be used in computing vapor exclusion
distances:
• A average gas concentration of 2.5% in air;
• Atmospheric conditions as a combination of those which result in longer predicted downwind
dispersion distances than other weather conditions at the site at least 90% of the time, based on
figures maintained by National Weather Service (NWS), or as an alternative where the model
used gives longer distances at lower wind speeds, “F” atmospheric stability, 4.5 miles per hour
(mph, 2 m/s) wind speed at reference height of 10 meters, 50% relative humidity, and average
atmospheric temperature in the region;
• A receptor elevation of 0.5 m above the ground;
• A surface roughness factor of 0.03 m (or 3 cm, approximately 1 inch). Higher values for the
roughness factor may be used if it can be shown that the terrain both upwind and downwind of
the vapor cloud has dense vegetation and that the vapor cloud height is more than ten times the
height of the obstacles encountered by the vapor cloud.
OVERVIEW OF SOURCE5 COMPUTER PROGRAM
In 1993 GRI conducted a study that developed the necessary equations for calculating the transient
vapor generation rate from LNG releases. The findings are detailed in GRI Report No. 92/0534. GRI
implemented the results from this study in a program called Source5. Source5 produces the transient
source strength table required to run the vapor dispersion program DEGADIS. DEGADIS is a dense-gas
dispersion model preferred by the NFPA 59A Standard. Source5 can predict the vaporization rate and
radius of the LNG pool as a function of time for the following LNG five LNG release scenarios:
9. • Confined, instantaneous land spill - All of the LNG is assumed to spill into the dike
instantaneously due to a catastrophic failure of the tank.
• Confined, continuous land spill - The model used for this scenario was developed by Arthur D.
Little, Inc. The model assumes a point discharge of liquid at the dike center with the LNG
spreading uniformly in a radial direction.
• Unconfined, instantaneous land spill - The model derived by the Netherlands Organization for
Applied Scientific Research (TNO) is used for describing the vapor release rate from an
unconfined spill. The LNG spill occurs on a horizontal surface so that the pool takes on the shape
of a disk. The pool is assumed to spread to a maximum radius and then shrink due to vaporization
until it eventually disappears.
• Instantaneous water spill – The spill scenario considers with or without ice formation.
• Continuous water spill – The model developed by Raj and O’Farrel is used to model continuous
LNG spills on water. It was postulated that the hazardous material was released as a jet at a low
velocity. This type of release causes the material to strike the surface of water, sink to a specified
depth and then float back to the surface to begin spreading in a radial direction. The maximum
radius of the spreading pool is a function of the rate at which the liquid was spilled from the
storage tank and its evaporation from the pool. The pool no longer spreads when the spill rate
equals the evaporation rate. This model does not consider the formation of an ice layer when LNG
contacts water.
Furthermore, Source5 incorporates four types of dike floor materials (i.e., concrete, dry soil, IPC
epoxy formulation, and Foamglas HLB125) and allows user-defined properties of other dike floor
materials. The program also implements five types of dike configuration (i.e., vapor overflow from
rectangular dike with sloping embankment and vertical fence; sloping embankment without vertical fence,
vertical walls, or vertical thick wall and fence and vapor overflow from circular dike). A sump can also be
considered within the dike. The program generates the input file for the dense gas vapor dispersion
program DEGADIS.
OVERVIEW OF DEGADIS PROGRAM
DEGADIS was developed specifically to model heavier-than-air gaseous releases from a single
source over flat terrain. A single meteorological condition is specified for the duration of the release. It
incorporates the Ooms vertical jet plume model, which is useful for chemical processes storing
pressurized substances that, if released, produce high-velocity emissions. The Ooms jet plume model
predicts the trajectory and dilution of vertically oriented gas or aerosol jets. It also accounts for ground
reflection when the plume’s lower boundary reaches the ground. DEGADIS is suitable for modeling
the following types of releases:
• Continuous release: a steady-state release of dense gas at a constant rate into the atmosphere
over a long period of time. The output from modeling a steady-state release is concentration
estimates at various downwind distances determined by the model.
• Finite-duration release: a steady-state release of dense gas at a constant rate into the
atmosphere over a short period of time. Finite-duration model output is organized either by
time or distance, depending on which parameter is of greater interest.
• Transient release: varying release rates over time. An example is a liquid pool boiling off — as
the size of the pool shrinks, the emission rate changes. Another example is the near-
10. instantaneous release when a pressurized container ruptures. Transient modeling output is
organized either by time or distance, depending on which parameter is of most interest.
• Jet release: a vertical release of a dense gas, or aerosol. The Ooms jet plume model requires
that the jet be vertical, with a definable exit velocity. If the jet release is such that the plume
centerline does not reach the ground before dispersing, the jet plume model is run alone. If this
is unclear, or if the plume centerline does reach the ground, the jet plume model is run in
conjunction with the regular DEGADIS model as either a continuous or finite-duration release.
• Liquid spill: a release of a chemical in its liquid state. The liquid is assumed to form a pool at
ground level, with the evaporation rate calculated using one of three different evaporation
models incorporated into DEGADIS. The evaporation model is run as either a continuous or
finite-duration release. Note that the liquid spill option will only be available if the chemical’s
normal boiling point is greater than the ambient temperature.
The DEGADIS model is based on several assumptions that impose limitations on its
applicability. Its use is restricted to dense-gas releases and liquid spills that evaporate to a dense gas.
DEGADIS (as well as the Ooms jet plume models) assumes flat terrain with uniform and constant
wind speed and direction. DEGADIS use is limited to conditions at which the depth of the dispersing
gas layer is much greater than the surface roughness of the surrounding area. In practice this limitation
alone does not limit DEGADIS’ use to support LNG analyses. The Ooms jet model is strictly for
vertical releases; it incorporates no horizontal jet release velocity. If the jet release is not perpendicular
to the ground, the modeling results will not be accurate.[4],[5],[6]
EFFECTS OF AMBIENT CONDITIONS ON FLAMMABLE VAPOR-GAS
DISPERSION: PARAMETRIC ANALYSIS
A case study was conducted to evaluate the sensitivity of the vapor exclusion distance with
respect to the change of ambient conditions (e.g., atmospheric stability, wind speed, and surface
roughness). LNG vapor dispersion was evaluated for a spill of LNG from a 6,000,000 gallons tank
(22,710 m3
) in a dike with vertical walls. The release source term was determined with Source5 and
the vapor dispersion was evaluated with DEGADIS. Figure 5 shows the dependence of the vapor
exclusion distance (at the 2.5% LNG concentration level) with the atmospheric stability (“A” – very
unstable; “B” – moderately unstable; “C” – slightly unstable; “D” – neutral; “E” – slightly stable; and
“F” – stable atmosphere). Under the same wind speed, the vapor exclusion distance increases as the
atmosphere becomes more stable. This behavior is commonly observed in the field where quiescent,
stable conditions with low wind speeds allow contaminants to travel relatively long distances before
dispersing below a concentration of interest. Figure 6 illustrates the variation of the vapor exclusion
distance with the wind speed and appropriate atmospheric stability. It is observed that the vapor
exclusion distance decreases with the increase of wind speed when the wind speed is low (e.g., less
than 5 mph). At higher wind speeds, the vapor exclusion distance generally increases with the increase
of wind speed. Figure 7 shows the variation of vapor exclusion distance with the surface roughness.
As shown in the figure, vapor exclusion distance is relatively sensitive to the value of surface
roughness. Higher surface roughness tends to decrease the vapor exclusion distance. Figures 6 and 7
suggest that the analyst has to be especially attentive to selection of the appropriate roughness length,
wind speed and atmospheric stability.
11. 0
500
1000
1500
2000
2500
3000
A B C D E F
Atmospheric Stability Class
VaporExclusionDistance(ft)_
Figure 5 - Variation of Vapor Exclusion Distance with Atmospheric Stability
(4.5 mph Wind Speed and 0.03 m Surface Roughness)
Figure 6 - Variation of Vapor Exclusion Distance with Wind Speed
(0.03 m Surface Roughness)
0
500
1000
1500
2000
2500
3000
3500
2 4 6 8 10 12 14 16
Wind Speed (mph)
VaporExclusionDistance(ft)
A - Very Unstable
B - Moderately Unstable
C - Slightly Unstable
D - Neutral
E - Slightly Stable
F - Stable
12. 0
500
1000
1500
2000
2500
3000
0 0.2 0.4 0.6 0.8 1
Surface Roughness (m)
VaporExclusionDistance(ft)_
Figure 7 - Variation of Vapor Exclusion Distance with Surface Roughness
(F Stability Class and 4.5 mph Wind Speed)
SUMMARY
Performing safety modeling analysis for LNG facility siting and fire protection requires the
understanding of the applicable regulatory requirements and the proper utilization of recommended
modeling programs and techniques. Findings from this study demonstrate that evaluating the effects
of various conditions (e.g., wind speed, ambient temperature, relative humidity, atmospheric stability,
and/or surface roughness) will be necessary in determining the maximum exclusion distances for both
the thermal radiation and the vapor dispersion. The analyst has to be particularly attentive to
roughness length, wind speed, and atmospheric stability when performing analyses that support LNG
facilities.
REFERENCES CITED
1. National Fire Protection Associate (NFPA), NFPA 59A – Standard for the Production, Storage, and
Handling of Liquefied Natural Gas (LNG), 2006 Edition.
2. Gas Research Institute (GRI), Report 89/0176 – LNGFIRE: A Thermal Radiation Model for LNG
Fires, 1989.
3. Gas Research Institute (GRI), Report 89/0242 – LNG Vapor Dispersion Prediction with the
DEGADIS Dense Gas Dispersion Model, 1989.
4. Spicer, T. and Havens, J., EPA’s User’s Guide for the DEGADIS 2.1 Dense Gas Dispersion Model,
EPA-450/4-89-019.
13. 5. Spicer, T., Havens, J., Tebeau, P. Key, L., DEGADIS: Heavier-Than-Air Gas Atmospheric
Dispersion Model, Paper Presented at the 79th Meeting of the Air Pollution Control Association,
Minneapolis, MN. June 22-27, 1986.
6. Ooms, G.A., Mahieu A.P., and Zelis, F., The Plume Path of Vented Gases Heavier than Air, First
Symposium on Loss Prevention and Safety Promotion in the Process Industries. Elsevier Press, 1974.