This is the accessible text file for GAO report number GAO-07-316 
entitled 'Maritime Security: Public Safety Consequences of a Terrorist 
Attack on a Tanker Carrying Liquefied Natural Gas Need Clarification' 
which was released on March 14, 2007. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as part 
of a longer term project to improve GAO products' accessibility. Every 
attempt has been made to maintain the structural and data integrity of 
the original printed product. Accessibility features, such as text 
descriptions of tables, consecutively numbered footnotes placed at the 
end of the file, and the text of agency comment letters, are provided 
but may not exactly duplicate the presentation or format of the printed 
version. The portable document format (PDF) file is an exact electronic 
replica of the printed version. We welcome your feedback. Please E-mail 
your comments regarding the contents or accessibility features of this 
document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

Report to Congressional Requesters: 

United States Government Accountability Office: 

GAO: 

February 2007: 

Maritime Security: 

Public Safety Consequences of a Terrorist Attack on a Tanker Carrying 
Liquefied Natural Gas Need Clarification: 

GAO-07-316: 

GAO Highlights: 

Highlights of GAO-07-316, a report to congressional requesters 

Why GAO Did This Study: 

The United States imports natural gas by pipeline from Canada and by 
tanker as liquefied natural gas (LNG) from overseas. LNG—a supercooled 
form of natural gas—currently accounts for about 3 percent of total 
U.S. natural gas supply, with an expected increase to about 17 percent 
by 2030, according to the Department of Energy (DOE). With this 
projected increase, many more LNG import terminals have been proposed. 
However, concerns have been raised about whether LNG tankers could 
become terrorist targets, causing the LNG cargo to spill and catch on 
fire, and potentially explode. DOE has recently funded a study to 
consider these effects; 
completion is expected in 2008. 

GAO was asked to (1) describe the results of recent studies on the 
consequences of an LNG spill and (2) identify the areas of agreement 
and disagreement among experts concerning the consequences of a 
terrorist attack on an LNG tanker. To address these objectives, GAO, 
among other things, convened an expert panel to discuss the 
consequences of an attack on an LNG tanker. 

What GAO Found: 

The six unclassified completed studies GAO reviewed examined the effect 
of a fire resulting from an LNG spill but produced varying results; 
some studies also examined other potential hazards of a large LNG 
spill. The studies’ conclusions about the distance at which 30 seconds 
of exposure to the heat (heat hazard) could burn people ranged from 
less than 1/3 of a mile to about 1-1/4 miles. Sandia National 
Laboratories (Sandia) conducted one of the studies and concluded, based 
on its analysis of multiple attack scenarios, that a good estimate of 
the heat hazard distance would be about 1 mile. Federal agencies use 
this conclusion to assess proposals for new LNG import terminals. The 
variations among the studies occurred because researchers had to make 
modeling assumptions since there are no data for large LNG spills, 
either from accidental spills or spill experiments. These assumptions 
involved the size of the hole in the tanker; the volume of the LNG 
spilled; and environmental conditions, such as wind and waves. The 
three studies that considered LNG explosions concluded explosions were 
unlikely unless the LNG vapors were in a confined space. Only the 
Sandia study examined the potential for sequential failure of LNG cargo 
tanks (cascading failure) and concluded that up to three of the ship’s 
five tanks could be involved in such an event and that this number of 
tanks would increase the duration of the LNG fire. 

GAO’s expert panel generally agreed on the public safety impact of an 
LNG spill, but believed further study was needed to clarify the extent 
of these effects, and suggested priorities for this additional 
research. Experts agreed that the most likely public safety impact of 
an LNG spill is the heat hazard of a fire and that explosions are not 
likely to occur in the wake of an LNG spill. However, experts disagreed 
on the specific heat hazard and cascading failure conclusions reached 
by the Sandia study. DOE’s recently funded study involving large-scale 
LNG fire experiments addresses some, but not all, of the research 
priorities identified by the expert panel. The leading unaddressed 
priority the panel cited was the potential for cascading failure of LNG 
tanks. 

Figure: LNG Tanker Passing Downtown Boston on Its Way to Port: 

[See PDF for Image] 

Source: GAO. 

[End of figure] 

What GAO Recommends: 

GAO recommends that the Secretary of Energy ensure that DOE 
incorporates into its LNG study the key issues identified by the expert 
panel. 

In reviewing our draft report, DOE agreed with our recommendation. 

[Hyperlink, http://www.gao.gov/cgi-bin/getrpt?GAO-07-316]. 

To view the full product, including the scope and methodology, click on 
the link above. For more information, contact Jim Wells at (202) 512-
3841 or wellsj@gao.gov. 

[End of section] 

Contents: 

Letter: 

Results in Brief: 

Background: 

Studies Identified Different Distances for the Heat Effects of an LNG 
Fire: 

Experts Generally Agreed That the Most Likely Public Safety Impact of 
an LNG Spill Is Fire's Heat Effect, but That Further Study Is Needed to 
Clarify the Extent of This Effect: 

Conclusions: 

Recommendation for Executive Action: 

Agency Comments and Our Evaluation: 

Appendix I: Scope and Methodology: 

Appendix II: Names and Affiliations of Members of GAO's Expert Panel on 
LNG Hazards: 

Appendix III: Summary of Expert Panel Results: 

Appendix IV: GAO Contact and Staff Acknowledgments: 

Tables: 

Table 1: Key Assumptions and Results of the LNG Spill Consequence 
Studies: 

Table 2: Expert Panel's Ranking of Need for Research on LNG: 

Figures: 

Figure 1: Existing, Approved, and Proposed LNG Terminals in the United 
States, as of October 2006: 

Figure 2: LNG Membrane Tanker: 

Abbreviations: 

BLEVE: boiling liquid expanding vapor explosion: 

DOE: Department of Energy: 

DOT: Department of Transportation: 

FERC: Federal Energy Regulatory Commission: 

kW/m(squared): kilowatts per square meter: 

LNG: liquefied natural gas: 

LPG: liquefied petroleum gas: 

m2: square meters: 

m3: cubic meters: 

m/s: meters per second: 

RPT: rapid phase transition: 

WSA: Waterway Suitability Assessment: 

United States Government Accountability Office: 
Washington, DC 20548: 

February 22, 2007: 

The Honorable John D. Dingell: 
Chairman: 
The Honorable Joe Barton: 
Ranking Member: 
Committee on Energy and Commerce: 
House of Representatives: 

The Honorable Bennie G. Thompson: 
Chairman: 
The Honorable Peter King: 
Ranking Member: 
Committee on Homeland Security: 
House of Representatives: 

The Honorable Edward J. Markey: 
House of Representatives: 

Worldwide, over 40,000 tanker cargos of liquefied natural gas (LNG) 
have been shipped since 1959, and imports of LNG are projected to 
increase over the next 10 years. LNG is a supercooled liquid form of 
natural gas--a crucial source of energy for the United States. Natural 
gas is used in homes for cooking and heating and as fuel for generating 
electricity, and it accounts for about one-fourth of all energy 
consumed in the United States each year. Prices for natural gas in the 
United States have risen over the past 5 years as demand for natural 
gas has increased faster than domestic production. To make up for the 
domestic shortfall, the United States imports some natural gas in 
pipelines from Canada. However, most reserves of natural gas are 
overseas and cannot be transported through pipelines. Natural gas from 
these reserves has to be transported to the United States as LNG in 
tankers. Because of the projected increase in LNG tankers arriving at 
U.S. ports, concerns have been raised about whether the tankers could 
become terrorist targets. 

LNG--primarily composed of methane--is odorless and nontoxic. It is 
produced by supercooling natural gas to minus 260 degrees Fahrenheit at 
atmospheric pressure, thus reducing its volume by more than 600 times. 
This process makes transport by tankers feasible. The tankers are 
double-hulled, with each tanker containing between four and six 
adjacent tanks heavily insulated to maintain the LNG's supercool 
temperature. Generally, these ships can carry enough LNG to supply the 
daily energy needs of over 10 million homes. Importing LNG requires 
specialized facilities--called regasification terminals--at ports of 
entry. At these terminals, the liquid is reconverted into natural gas 
and then injected into the pipeline system for consumers. Currently, 
the United States has a total of five LNG import terminals--four are 
considered onshore terminals, that is, they are located within 3 miles 
of the shore; one is an offshore terminal located 116 miles off the 
Louisiana coast in the Gulf of Mexico.[Footnote 1] 

The United States imports about 3 percent of its total natural gas 
supply as LNG in recent years, but by 2030, LNG imports are projected 
to account for about 17 percent of the U.S. natural gas supply, 
according to the Department of Energy's (DOE) Energy Information 
Administration. To meet this increased demand, energy companies have 
submitted 32 applications to build new terminals for importing LNG in 
10 states and five offshore areas. Figure 1 shows the locations of LNG 
terminals that are operational, approved, and proposed. 

Figure 1: Existing, Approved, and Proposed LNG Terminals in the United 
States, as of October 2006: 

[See PDF for image] 

Sources: FERC and GAO. 

[End of figure] 

As of October 2006, the Federal Energy Regulatory Commission 
(FERC)[Footnote 2]--responsible for approving onshore LNG terminal 
siting applications--and the U.S. Coast Guard[Footnote 3]--responsible 
for approving offshore LNG terminal siting applications--had together 
approved 13 of these applications. In addition, the Coast Guard 
contributes to FERC's review of onshore LNG facilities by reviewing and 
validating an applicant's Waterway Suitability Assessment (WSA) and 
reaching a preliminary conclusion as to whether the waterway is 
suitable for LNG operations with regard to navigational safety and 
security considerations. The WSA includes a security risk assessment to 
evaluate the public safety risk of an LNG spill from a tanker following 
an attack. The security risk assessment analyzes potential types of 
attacks, their probability, and the potential consequences. The WSA 
also identifies appropriate strategies that can be used to reduce the 
risk posed by a terrorist attack on an LNG tanker, either by reducing 
the probability of an attack, or by reducing its consequences. If the 
WSA deems the waterway suitable for LNG tanker traffic, the Coast Guard 
provides FERC with a "Letter of Recommendation," which describes the 
overall risk reduction strategies that will be used on LNG tankers 
traveling to the proposed terminal. The Coast Guard is the lead federal 
agency for ensuring the security of active LNG import terminals and 
tankers traveling within U.S. ports. 

As figure 1 shows, six new facilities have been proposed for the 
northeastern United States, a region that faces gas supply challenges. 
The Northeast has limited indigenous supplies of natural gas, and 
receives most of its natural gas either through pipelines from the U.S. 
Gulf Coast or Canada, or from overseas via tanker as LNG. The pipelines 
into the Northeast currently run at or near capacity for much of the 
winter, and demand is projected to significantly increase over the next 
5 years, exceeding available supply by 2010. To meet the increasing 
demand, new supplies of natural gas must reach the Northeast by 
expanding existing pipeline capacity, constructing new pipelines, or 
constructing new LNG terminals--all of which have risk associated with 
them. Difficulties siting LNG facilities in the Northeast could lead to 
higher natural gas prices unless additional supply can be brought into 
the region via new, or expansion of old, pipelines. 

Scientists and the public have raised concerns about the potential 
hazards that an LNG spill could pose. When LNG is spilled from a 
tanker, it forms a pool of liquid on the water. Individuals who come 
into contact with LNG could experience freeze burns. As the liquid 
warms and changes into natural gas, it forms a visible, foglike vapor 
cloud close to the water. The cloud mixes with ambient air as it 
continues to warm up and eventually the natural gas disperses into the 
atmosphere. Under certain atmospheric conditions, however, this cloud 
could drift into populated areas before completely dispersing. Because 
an LNG vapor cloud displaces the oxygen in the air, it could 
potentially asphyxiate people who come into contact with it. 
Furthermore, like all natural gas, LNG vapors can be flammable, 
depending on conditions.[Footnote 4] If the LNG vapor cloud ignites, 
the resulting fire will burn back through the vapor cloud toward the 
initial spill. It will continue to burn above the LNG that has pooled 
on the surface--this is known as a pool fire. Experiments to date have 
shown that LNG fires burn hotter than oil fires of the same size. Both 
the cold temperatures of spilled LNG and the high temperatures of an 
LNG fire have the potential to significantly damage the tanker, causing 
multiple tanks on the ship to fail in sequence--called a cascading 
failure. Such a failure could increase the severity of the incident. 
Finally, concerns have been raised about whether an explosion could 
result from an LNG spill. 

Although LNG tankers have carried over 40,000 shipments worldwide since 
1959, there have been no LNG spills resulting from a cargo tank 
rupture. Some safety incidents, such as groundings or collisions, have 
resulted in small LNG spills that did not affect public safety. In the 
1970s and 1980s, experiments to determine the consequences of a spill 
examined small LNG spills of up to 35 meters in diameter. Following the 
terrorist attacks of September 11, 2001, however, many experts 
recognized that an attack on an LNG tanker could result in a large 
spill--a volume of LNG up to 100 times greater than studied in past 
experiments. Since then, a number of studies have reevaluated safety 
hazards of LNG tankers in light of a potential terrorist threat. 
Because a major LNG spill has never occurred, studies examining LNG 
hazards rely on computer models to predict the effects of hypothetical 
accidents, often focusing on the properties of LNG vapor fires. The 
Coast Guard uses one of these studies, conducted in 2004 by Sandia 
National Laboratories,[Footnote 5] as a basis for conducting the 
security risk assessment required in the WSA for proposed onshore LNG 
facilities.[Footnote 6] Access to accurate information about the 
consequences of LNG spills is crucial for developing accurate risk 
assessments for LNG siting decisions. While an underestimation of the 
consequences could expose the public to additional risk in the event of 
an LNG spill, an overestimation of consequences could result in the use 
of inappropriate and costly risk mitigation strategies. DOE recently 
funded a new study--to be completed by Sandia National Laboratories in 
2008--that will conduct small-and large-scale LNG fire experiments to 
refine and validate existing models (such as the one used by Sandia 
National Laboratories in their 2004 study) that calculate the heat 
hazards of large LNG fires. 

In this context, you asked us to (1) describe the results of recent 
unclassified studies on the consequences of an LNG spill and (2) 
identify the areas of agreement and disagreement among experts 
concerning the consequences of a terrorist attack on an LNG tanker. 

To address the first objective, we identified eight unclassified, 
completed studies of LNG hazards and reviewed the six studies that 
included new, original research (either experimental or modeling) and 
clearly described the methodology used. While we have not verified the 
scientific modeling or results of these studies, the methods used seem 
appropriate for the work conducted. We also interviewed agencies 
responsible for LNG regulations and visited all four onshore LNG import 
facilities and one export facility. To address the second objective, we 
identified 19 recognized experts in LNG hazard analysis and convened a 
Web-based expert panel to obtain their views on LNG hazards and to get 
agreement on as many issues as possible. In selecting experts for the 
panel, we sought individuals who are widely recognized as having 
experience with one or more key aspects of LNG hazard analysis. We 
sought to achieve balance in representation from government, academia, 
consulting, research organizations, and industry. Additionally, we 
ensured that our expert panel included at least one author from each of 
the six unclassified studies of LNG hazards. Because some of the 
studies conducted are classified, this public version of our findings 
supplements a more comprehensive classified report produced under 
separate cover. A more detailed description of our scope and 
methodology is presented in appendix I. We conducted our work from 
January 2006 through January 2007 in accordance with generally accepted 
government auditing standards. 

Results in Brief: 

The six unclassified studies we reviewed all examined the heat impact 
of an LNG pool fire but produced varying results; some studies also 
examined other potential hazards of a large LNG spill and reached 
consistent conclusions on explosions. Specifically, the studies' 
conclusions about the distance at which 30 seconds of exposure to the 
heat could burn people ranged from about 500 meters (less than 1/3 of a 
mile) to more than 2,000 meters (about 1-1/4 miles). The Sandia 
National Laboratories' study concluded that the most likely distance 
for a burn is about 1,600 meters (1 mile). These variations occurred 
because researchers had to make numerous modeling assumptions to scale- 
up the existing experimental data for large LNG spills since there are 
no large spill data from actual events. These assumptions involved the 
size of the hole in the tanker, the number of tanks that fail, the 
volume of LNG spilled, key LNG fire properties, and environmental 
conditions, such as wind and waves. Three of the studies also examined 
other potential hazards of an LNG spill, including LNG vapor 
explosions, asphyxiation, and cascading failure. All three studies 
considered LNG vapor explosions unlikely unless the LNG vapors were in 
a confined space. Only the Sandia National Laboratories' study examined 
asphyxiation, and it concluded that asphyxiation did not pose a hazard 
to the general public. Finally, only the Sandia National Laboratories' 
study examined the potential for cascading failure of LNG tanks and 
concluded that only three of the five tanks would be involved in such 
an event and that this number of tanks would increase the duration of 
the LNG fire. 

Our panel of 19 experts generally agreed on the public safety impact of 
an LNG spill, disagreed with a few conclusions reached by the Sandia 
National Laboratories' study, and suggested priorities for research to 
clarify the impact of heat and cascading tank failures. Experts agreed 
that (1) the most likely public safety impact of an LNG spill is the 
heat impact of a fire; (2) explosions are not likely to occur in the 
wake of an LNG spill, unless the LNG vapors are in confined spaces; and 
(3) some hazards, such as freeze burns and asphyxiation, do not pose a 
hazard to the public. Experts disagreed with the heat impact and 
cascading tank failure conclusions reached by the Sandia National 
Laboratories' study, which the Coast Guard uses to prepare WSAs. 
Specifically, all experts did not agree with the heat impact distance 
of 1,600 meters. Seven of 15 experts thought Sandia's distance was 
"about right," and the remaining eight experts were evenly split as to 
whether the distance was "too conservative" or "not conservative 
enough" (the other 4 experts did not answer this question). Experts 
also did not agree with the Sandia National Laboratories' conclusion 
that only three of the five LNG tanks on a tanker would be involved in 
a cascading failure. Finally, experts suggested priorities to guide 
future research aimed at clarifying uncertainties about heat impact 
distances and cascading failure, including large-scale fire 
experiments, large-scale LNG spill experiments on water, the potential 
for cascading failure of multiple LNG tanks, and improved modeling 
techniques. DOE's recently funded study involving large-scale LNG fire 
experiments addresses some, but not all, of the research priorities 
identified by the expert panel. 

We are recommending that DOE incorporate into its current LNG study the 
key issues identified by the expert panel. We particularly recommend 
that DOE examine the potential for cascading failure of LNG tanks. 

Background: 

Natural gas is primarily composed of methane, with small percentages of 
other hydrocarbons, including propane and butane. When natural gas is 
cooled to minus 260 degrees Fahrenheit at atmospheric pressure, the gas 
becomes a liquid, known as LNG, and it occupies only about 1/600th of 
the volume of its gaseous state. Since LNG is maintained in an 
extremely cooled state--reducing its volume--there is no need to store 
it under pressure. This liquefaction process allows natural gas to be 
transported by trucks or tanker vessels. LNG is not explosive or 
flammable in its liquid state. When LNG is warmed, either at a 
regasification terminal or from exposure to air as a result of a spill, 
it becomes a gas. As this gas mixes with the surrounding air, a 
visible, low-lying vapor cloud results. This vapor cloud can be ignited 
and burned only within a minimum and maximum concentration of air and 
vapor (percentage by volume). For methane, the dominant component of 
this vapor cloud, this flammability range is between 5 percent and 15 
percent by volume. When fuel concentrations exceed the cloud's upper 
flammability limit, the cloud cannot burn because too little oxygen is 
present. When fuel concentrations are below the lower flammability 
limit, the cloud cannot burn because too little methane is present. As 
the cloud vapors continue to warm, above minus 160 degrees Fahrenheit, 
they become lighter than air and will rise and disperse rather than 
collect near the ground. 

If the cloud vapors ignite, the resulting fire will burn back through 
the vapor cloud toward the initial spill and will continue to burn 
above the LNG that has pooled on the surface. This fire burns at an 
extremely high temperature--hotter than oil fires of the same size. LNG 
fires burn hotter because the flame burns very cleanly and with little 
smoke. In oil fires, the smoke emitted by the fire absorbs some of the 
heat from the fire and reduces the amount of heat emitted. Scientists 
measure the amount of heat given off by a fire by looking at the amount 
of heat energy emitted per unit area as a function of time. This is 
called the surface emissive power of a fire and is measured in 
kilowatts per square meter (kW/m2). Generally, the heat given off by an 
LNG fire is reported to be more than 200 kW/m2. By comparison, the 
surface emissive power of a very smoky oil fire can be as little as 20 
kW/m2. The heat from fire can be felt far away from the fire itself. 
Scientists use heat flux--also measured in kW/m2--to quantify the 
amount of heat felt at a distance from a fire. For instance, a heat 
flux of 5 kW/m2 can cause second degree burns after about 30 seconds of 
exposure to bare skin. This heat flux can be compared with the heat 
from a candle--if a hand is held about 8 to 9 inches above the candle, 
second degree burns could result in about 30 seconds. A heat flux of 
about 12.5 kW/m2, over an exposure time of 10 minutes, will ignite 
wood, and a heat flux of about 37.5 kW/m2 can damage steel structures. 

Four types of explosions could potentially occur after an LNG spill: 
rapid phase transitions (RPT), deflagrations, detonations, and boiling- 
liquid-expanding-vapor-explosions (BLEVE).[Footnote 7] More 
specifically: 

* An RPT occurs when LNG is warmed and changes into natural gas nearly 
instantaneously. An RPT generates a pressure wave that can range from 
very small to large enough to damage lightweight structures. RPTs 
strong enough to damage test equipment have occurred in past LNG spill 
experiments on water, although their effects have been localized at the 
site of the RPT. 

* Deflagrations and detonations are explosions that involve combustion 
(fire). They differ on the basis of the speed and strength of the 
pressure wave generated: deflagrations move at subsonic velocities and 
can result in pressures (overpressures) up to 8 times higher than the 
original pressure; detonations travel faster--at supersonic velocities--
and can result in larger overpressures--up to 20 times the original 
pressure. Methane does not detonate as readily as other hydrocarbons; 
it requires a larger explosion to initiate a detonation in a methane 
cloud. 

* A BLEVE occurs when a liquefied gas is heated to above its boiling 
point while contained within a tank. For instance, if a hot fire 
outside an LNG tanker sufficiently heated the liquid inside, a 
percentage of the LNG within the tank could "flash" into a vapor state 
virtually instantaneously, causing the pressure within the tank to 
increase. LNG tanks do have pressure relief valves, but if these were 
inadequate or failed, the pressure inside the tank could rupture the 
tank. The escaping gas would be ignited by the fire burning outside the 
tank, and a fireball would ensue. The rupture of the tank could create 
an explosion and flying debris (portions of the tank). 

World natural gas reserves are abundant, estimated at about 6,300 
trillion cubic feet, or 65 times the volume of natural gas used in 
2005. Much of this gas is considered "stranded" because it is located 
in regions far from consuming markets. Russia, Iran, and Qatar combined 
hold natural gas reserves that represent more than half of the world 
total. Many countries have imported LNG for years. In 2005, 13 
countries shipped natural gas to 14 LNG-importing countries. LNG 
imports, as a percentage of a country's total gas supply, for each of 
the importing countries ranged from 3 percent in the United States to 
nearly 95 percent in Japan. In 2005, LNG imports to the United States 
originated primarily in Trinidad and Tobago (70 percent), Algeria (15 
percent), and Egypt (11 percent). The remaining 4 percent of U.S. LNG 
imports came from Oman, Malaysia, Nigeria, and Qatar. 

LNG tankers primarily have two basic designs, called membrane or Moss 
(see fig. 2). Both designs consist of an outer hull, inner hull, and 
cargo containment system. In membrane tank designs, the cargo is 
contained by an Invar, or stainless steel double-walled liner, that is 
structurally supported by the vessel's inner hull. The Moss tank design 
uses structurally independent spherical or prismatic shaped tanks. 
These tanks, usually five located one behind the other, are constructed 
of either stainless steel or an aluminum alloy. LNG tankers ships are 
required to meet international maritime construction and operating 
standards, as well as U.S. Coast Guard safety and security regulations. 

Figure 2: LNG Membrane Tanker: 

[See PDF for image] 

Source: GAO. 

[End of figure] 

Studies Identified Different Distances for the Heat Effects of an LNG 
Fire: 

The six studies we examined identified various distances at which the 
heat effects of an LNG fire could be hazardous to people. The studies' 
variations in heat effects result from the assumptions made in the 
studies' models. Some studies also examined other potential hazards 
such as LNG vapor explosions, other types of explosions, and 
asphyxiation, and identified their potential impacts on public safety. 

Studies Identified Various Distances That the Heat Effects of an LNG 
Fire Could Be Hazardous to People because of Assumptions Made: 

The studies' conclusions about the distance at which 30 seconds of 
exposure to the heat could burn people ranged from about 500 meters 
(less than 1/3 mile) to more than 2,000 meters (about 1-1/4 miles). The 
results--size of the LNG pool, the duration of the fire, and the heat 
hazard distance for skin burn--varied in part because the studies made 
different assumptions about key parameters of LNG spills and also 
because they were designed and conducted for different purposes. Key 
assumptions made included the following: 

* Hole size and cascading failure. Hole size is an important parameter 
for modeling LNG spills because of its relationship to the duration of 
the event--larger holes allow LNG to spill from the tanker more 
quickly, resulting in larger LNG pools and shorter duration fires. 
Conversely, small holes could create longer-duration fires. Cascading 
failure is important because it increases the overall spill volume and 
the duration of the spill. 

* Waves and wind. These conditions can affect the size of both the LNG 
pool and the heat hazard zone. One study indicated that waves can 
inhibit the spread of an LNG pool, keeping the pool size much smaller 
than it would be on a smooth surface, and thereby reducing the size of 
the LNG pool fire. Wind will tend to tilt the fire downwind (like a 
candle flame blowing in the wind), increasing the heat hazard zone in 
that direction (and decreasing it upwind). 

* Volume of LNG spilled. The amount of LNG spilled is one of the 
factors that can affect the size of the pool. 

* Surface emissive power of the fire. While the amount of heat given 
off by a large LNG fire is unknown, assumptions about it directly 
affect the results for the heat hazard zone. It is expected that the 
surface emissive power of LNG fires will be lower for large fires 
because oxygen will not circulate efficiently within a very large fire. 
Lack of oxygen in the middle of a large fire would lead to more smoke 
production, which would block some of the heat from the fire. 

The LNG spill consequence studies' key assumptions and results are 
shown in table 1. 

Table 1: Key Assumptions and Results of the LNG Spill Consequence 
Studies: 

Quest Consultants Inc. (Quest)[A]; 
Key assumptions: Hole size (m[2] ): 19.6; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): 1.5; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): 1.5; 
Key assumptions: Spill volume (m[3] ): 12,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): [B]; 
Key results: Pool diameter (meters): 156; 
Key results: Distance to the 5kw/m[2] heat level (meters): 497; 
Key results: Duration (minutes): 14.3. 

Quest Consultants Inc. (Quest)[A]; 
Key assumptions: Hole size (m[2] ): 19.6; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): 5.0; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): 5.0; 
Key assumptions: Spill volume (m[3] ): 12,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): [B]; 
Key results: Pool diameter (meters): 146; 
Key results: Distance to the 5kw/m[2] heat level (meters): 531; 
Key results: Duration (minutes): 16.6. 

Quest Consultants Inc. (Quest)[A]; 
Key assumptions: Hole size (m[2] ): 19.6; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): 9.0; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): 9.0; 
Key assumptions: Spill volume (m[3] ): 12,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): [B]; 
Key results: Pool diameter (meters): 110; 
Key results: Distance to the 5kw/m[2] heat level (meters): 493; 
Key results: Duration (minutes): 28.6. 

Sandia National Laboratories (Sandia); 
Key assumptions: Hole size (m[2] ): 2; 
Key assumptions: Number of tanks that rupture (cascading failure: 3; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): [C]; 
Key assumptions: Spill volume (m[3] ): 37,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): 220; 
Key results: Pool diameter (meters): 209; 
Key results: Distance to the 5kw/m[2] heat level (meters): 784; 
Key results: Duration (minutes): 20. 

Sandia National Laboratories (Sandia); 
Key assumptions: Hole size (m[2] ): 5; 
Key assumptions: Number of tanks that rupture (cascading failure: 3; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): [C]; 
Key assumptions: Spill volume (m[3] ): 37,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): 220; 
Key results: Pool diameter (meters): 572; 
Key results: Distance to the 5kw/m[2] heat level (meters): 2,118; 
Key results: Duration (minutes): 8.1. 

Sandia National Laboratories (Sandia); 
Key assumptions: Hole size (m[2] ): 5; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): [C]; 
Key assumptions: Spill volume (m[3] ): 12,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): 350; 
Key results: Pool diameter (meters): 330; 
Key results: Distance to the 5kw/m[2] heat level (meters): 1,652; 
Key results: Duration (minutes): 8.1. 

Sandia National Laboratories (Sandia); 
Key assumptions: Hole size (m[2] ): 5[D]; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): [C]; 
Key assumptions: Spill volume (m[3] ): 12,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): 220; 
Key results: Pool diameter (meters): 330-405; 
Key results: Distance to the 5kw/m[2] heat level (meters): 1,305-1,579; 
Key results: Duration (minutes): 5.4-8.1. 

Sandia National Laboratories (Sandia); 
Key assumptions: Hole size (m[2] ): 12; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): [C]; 
Key assumptions: Spill volume (m[3] ): 12,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): 220; 
Key results: Pool diameter (meters): 512; 
Key results: Distance to the 5kw/m[2] heat level (meters): 1,920; 
Key results: Duration (minutes): 3.4. 

Pitblado, et al. (Pitblado)[E]; 
Key assumptions: Hole size (m[2] ): 1.77; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): 3.0; 
Key assumptions: Spill volume (m[3] ): 17,250; 
Key assumptions: Fire surface emissive power (kW/m[2] ): [B]; 
Key results: Pool diameter (meters): 171; 
Key results: Distance to the 5kw/m[2] heat level (meters): 750; 
Key results: Duration (minutes): 32. 

ABS Consulting (ABSC)[F]; 
Key assumptions: Hole size (m[2] ): 0.79; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): 8.9; 
Key assumptions: Spill volume (m[3] ): 12,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): 265; 
Key results: Pool diameter (meters): 200[G]; 
Key results: Distance to the 5kw/m[2] heat level (meters): 650; 
Key results: Duration (minutes): 51. 

ABS Consulting (ABSC)[F]; 
Key assumptions: Hole size (m[2] ): 19.6; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): 8.9; 
Key assumptions: Spill volume (m[3] ): 12,500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): 265; 
Key results: Pool diameter (meters): 620[G]; 
Key results: Distance to the 5kw/m[2] heat level (meters): 1,500; 
Key results: Duration (minutes): 4.2. 

Fay (Fay)[H]; 
Key assumptions: Hole size (m[2] ): 20; 
Key assumptions: Number of tanks that rupture (cascading failure: 1; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): [C]; 
Key assumptions: Spill volume (m[3] ): 14,300; 
Key assumptions: Fire surface emissive power (kW/m[2] ): [B]; 
Key results: Pool diameter (meters): [B]; 
Key results: Distance to the 5kw/m[2] heat level (meters): 1,900; 
Key results: Duration (minutes): 3.3. 

Lehr and Simecek-Beatty (Lehr)[I]; 
Key assumptions: Hole size (m[2] ): [B]; 
Key assumptions: Number of tanks that rupture (cascading failure: [B]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on waves (m/s): [C]; 
Key assumptions: Environmental conditions modeled: Wind speed and its 
effect on fire (m/s): [C]; 
Key assumptions: Spill volume (m[3] ): 500; 
Key assumptions: Fire surface emissive power (kW/m[2] ): 200; 
Key results: Pool diameter (meters): [B]; 
Key results: Distance to the 5kw/m[2] heat level (meters): 500; 
Key results: Duration (minutes): 2-3. 

Source: GAO analysis of spill consequence studies. 

[A] "Modeling LNG Spills in Boston Harbor." Copyright© 2003 Quest 
Consultants, Inc., Norman, OK 73609; 
Letter from Quest Consultants to DOE (October 2, 2001); 
Letter from Quest Consultants to DOE (October 3, 2001). 

[B] Information not available. 

[C] Not included in the model. 

[D] The study examined multiple scenarios of 5m2. The ranges listed 
summarize the highest and lowest values for those scenarios. 

[E] R. M. Pitblado, J. Baik, G. J. Hughes, C. Ferro, and S. J. Shaw. 
"Consequences of Liquefied Natural Gas Marine Incidents." Process 
Safety Progress 24 no. 2 (June 2005). 

[F] ABS Consulting Inc. Consequence Assessment Methods for Incidents 
Involving Releases from Liquefied Natural Gas Carriers. May 13, 2004. 
FERC "Staff's Responses to Comments on the Consequence Assessment 
Methods for Incidents Involving Releases from Liquefied Natural Gas 
Carriers," June 18, 2004. 

[G] ABS Consulting modeled pool size as a semicircle and reported the 
radius of that semicircle in the study. The reported radii were used to 
calculate the diameter of the semicircle so the study results could be 
compared with the other studies. 

[H] J.A. Fay. "Model of Spills and Fires from LNG and Oil tankers." 
Journal of Hazardous Materials B96 (2003): 171-188. 

[I] William Lehr and Debra Simecek-Beatty. "Comparison of Hypothetical 
LNG and Fuel Oil Fires on Water." Journal of Hazardous Materials 107 
(2004): 3-9. 

[End of table] 

In terms of the studies' results, we identified the following three key 
results: 

* Pool size describes the extent of the burning pool--and can help 
people understand how large the LNG fire itself will be. 

* Heat hazard distance describes the distance at which 30 seconds of 
exposure could cause second degree burns. 

* Fire duration of the incident describes how long people and 
infrastructure will be exposed to the heat from the fire. The longer 
the fire, the greater potential for damage to the tanker and for 
cascading failure. 

Although all the studies considered the consequences of an LNG spill, 
they were conducted for different purposes. Three of the six studies-- 
Quest, Sandia, and Pitblado--specifically addressed the consequences of 
LNG spills caused by terrorist attacks. Two of these three studies-- 
Quest and Sandia--were commissioned by DOE. The Quest study, begun in 
response to the September 11, 2001, attacks, was designed to quantify 
the heat hazard zones for LNG tanker spills in Boston Harbor. Only the 
Quest study examined how wind and waves would affect the spreading of 
the LNG on the water and the size of the resulting LNG pool. The Quest 
study based its wind and wave assumptions on weather data from buoys 
near Boston Harbor. The Quest study found that, while the waves would 
help reduce the size of the LNG pool, the winds that created the waves 
would tend to increase the heat hazard distance downwind. To simplify 
the modeling of the waves, the Quest study considered "standing" waves 
(rather than moving waves) of various heights and, therefore, did not 
consider the impact of wave movement on LNG pool spreading. The ABSC 
study expressed concern that Quest's standing wave assumption resulted 
in pool sizes that were too small because wave movement might help 
spread the LNG. 

The 2004 Sandia study was intended to develop guidance on a risk- based 
analysis approach to assess potential threats to an LNG tanker, 
determine the potential consequences of a large spill, and review 
techniques that could be used to mitigate the consequences of an LNG 
spill. The assumptions and results in table 1 for the Sandia study 
refer to the scenarios Sandia examined that had terrorist causes. 
According to Sandia, the study used available intelligence and 
historical data to develop credible and possible scenarios for the 
kinds of attacks that could breach an LNG tanker. Sandia then modeled 
how large a hole each of the weapon scenarios could create in an LNG 
tanker.[Footnote 8] Two of these intentional breach scenarios included 
cascading failure of three tanks on an LNG tanker. In these cases, the 
LNG spill from one tank, as well as the subsequent fire, causes the 
neighboring two tanks to fail on the LNG tanker, resulting in LNG 
spills from three of the five tanks on the tanker. After considering 
all of its scenarios, Sandia concluded that, as a rule-of-thumb, 1,600 
meters is a good approximation of the heat hazard distance for 
terrorist-induced spills. However, as the table shows, one of Sandia's 
scenarios--for a large spill with cascading failure of three LNG tanks-
-found that the distance could exceed more than 2,000 meters and that 
the cascading failure would increase the duration of the incident. 

Finally, the stated purpose of industry's Pitblado study was to develop 
credible threat scenarios for attacks on LNG tankers and predict hazard 
zones for LNG spills from those types of attacks. The study identified 
a hole size smaller than the other studies that specifically considered 
terrorist attacks. 

The other studies we reviewed examined LNG spills regardless of cause. 
FERC commissioned the ABS Consulting study to develop appropriate 
methods for estimating heat hazard zones from LNG spills. FERC uses 
these methods, in conjunction with the Sandia study, to examine the 
public safety consequences of tankers traveling to proposed onshore LNG 
facilities before granting siting approval. The two scenarios in the 
ABSC study illustrate how small holes could result in longer fires, 
which have a higher potential to damage the tanker itself. One scenario 
used a hole size of 0.79 square meters and the other a hole size of 
about 20 square meters. The difference in duration is striking--it 
takes 51 minutes and 4.2 minutes, respectively, for the fire to consume 
all the spilled LNG. 

Finally, the Lehr and Fay studies compared the fire consequences of LNG 
spills with known information about oil spills and fires. Although most 
studies made similar assumptions about the volume of LNG spilled from 
any single LNG tank, Lehr examined a much smaller spill volume-- just 
500 cubic meters of LNG, compared with a range of 12,500 to 17,250 
cubic meters. 

Some Studies Examined Other Potential Hazards and Identified Their 
Impact on Public Safety: 

Three of the studies also examined other potential hazards of an LNG 
spill, including LNG vapor explosions, other types of explosions, and 
asphyxiation. 

LNG vapor explosions. Three studies--Sandia, ABSC, and Pitblado-- 
examined LNG vapor explosions, and all agreed that it is unlikely that 
LNG vapors could explode and create a pressure wave if the vapors are 
in an unconfined space. Although the three studies agreed that LNG 
vapors could explode only in confined areas, they did not conduct 
modeling or describe the likelihood of such confinement after an LNG 
spill from a tanker. The Sandia study stated that fire will generally 
progress through the vapor cloud slowly and without producing an 
explosion with damaging pressure waves. The study did suggest, however, 
that if the LNG vapor cloud is confined (e.g., between the inner and 
outer hull of an LNG carrier), it could explode but would only affect 
the immediate surrounding area. The ABSC study and the Pitblado study 
agreed that a confined LNG vapor cloud could result in an explosion. 

Other types of explosions. Three studies--Sandia, ABSC, and Pitblado-
-examined the potential for RPTs. The Sandia study concluded that, 
while RPTs have generated energy releases equivalent to several pounds 
of explosives, RPT impacts will be localized near the spill. Sandia 
also noted that RPTs are not likely to cause structural damage to the 
vessel. The ABSC study noted that their literature search suggested 
that damage from RPT overpressures would be limited to the immediate 
vicinity, though it noted that the literature did not include large 
spills like those that could be caused by a terrorist attack. Only one 
study, Pitblado, discussed the possibility of a BLEVE. According to our 
discussions with Dr. Pitblado, an LNG ship with membrane tanks could 
not result in a BLEVE, but he said that Moss spherical tanks could 
potentially result in a BLEVE. A BLEVE could result because it is 
possible for pressure to build up in a Moss tanker. A 2002 LNG tanker 
truck incident in Spain resulted in an explosion that some scientists 
have characterized as a BLEVE of an LNG truck. Portions of the tanker 
truck were found 250 meters from the accident itself, propelled by the 
strength of the blast. 

Asphyxiation. Only the Sandia study examined the potential for 
asphyxiation following an LNG spill if the vapors displace the oxygen 
in the air. It concluded that fire hazards would be the greatest 
problem in most locations, but that asphyxiation could threaten the 
ship's crew, pilot boat crews, and emergency response personnel. 

Experts Generally Agreed That the Most Likely Public Safety Impact of 
an LNG Spill Is Fire's Heat Effect, but That Further Study Is Needed to 
Clarify the Extent of This Effect: 

Our panel of 19 experts generally agreed on the public safety impact of 
an LNG spill and disagreed with a few of the conclusions of the Sandia 
study.[Footnote 9] The experts also suggested priorities for future 
research--some of which are not fully addressed in DOE's ongoing LNG 
research--to clarify uncertainties about heat impact distances and 
cascading failure. These priorities include large-scale fire 
experiments, large-scale LNG spill experiments on water, the potential 
for cascading failure of multiple LNG tanks, and improved modeling 
techniques. 

Experts Agreed That the Heat from an LNG Fire Was Most Likely to Affect 
Public Safety, but That Explosions from an LNG Spill Are Unlikely: 

Experts discussed two types of fires: vapor cloud fires and pool fires. 
Eighteen of 19 experts agreed that the ignition of a vapor cloud over a 
populated area could burn people and property in the immediate vicinity 
of the fire. While the initial vapor cloud fire would be of short 
duration as the flames burned back toward the LNG carrier, any 
flammable object enveloped by the vapor cloud fire could ignite nearby 
objects, creating secondary fires that present hazards to the public. 
Three experts emphasized in their comments that the vapor cloud is 
unlikely to penetrate very far into a populated area before igniting. 
Expanding on this point, one expert noted that any injuries from a 
vapor cloud fire would occur only at the edges of a populated area, for 
example, along beaches. One expert disagreed, arguing that a vapor 
cloud fire is unlikely to cause significant secondary fires because it 
would not last long enough to ignite other materials. 

Experts agreed that the main hazard to the public from a pool fire is 
the heat from the fire but emphasized that the exact hazard distance 
depends on site-specific and scenario-specific factors. Furthermore, a 
large, unconfined pool fire is very difficult to extinguish; 
generally almost all the LNG must be consumed before the fire goes out. 
Experts agreed that three of the main factors that affect the amount of 
heat from an LNG fire are the following: 

* Site-specific weather conditions. Weather conditions, such as wind 
and humidity, can influence the heat hazard distances. For example, 
more humid conditions allow heat to be absorbed by the moisture in the 
air, reducing heat hazard distances. 

* Composition of the LNG. The composition of the LNG can also affect 
the distance at which heat from the fire is felt by the public. In 
small fires, methane, which comprises between 84 percent and 97 percent 
of LNG, burns cleanly, with little smoke. Other LNG components--propane 
and butane--produce more smoke when burned, absorbing some of the 
fire's heat and reducing the hazard distance. As the fire grows larger, 
the influence of the composition of LNG is hypothesized to be less 
pronounced because large fires do not burn efficiently. 

* Size of the fire. The size of the fire has a major impact on its 
surface emissive power; the heat hazard distance increases with pool 
size up to a point but is expected to decrease for very large pools, 
like those caused by a terrorist attack. 

Experts also discussed the following hazards related to an LNG spill: 

* RPTs. Experts agreed that RPTs could occur after an LNG spill but 
that the overpressures generated would be unlikely to directly affect 
the public. 

* Detonations and deflagrations. Experts made a key distinction between 
these types of explosions in confined spaces as opposed to unconfined 
spaces. For confined spaces, they agreed that it is possible, under 
controlled experimental conditions, to induce both types of explosions 
of LNG vapors; however, a detonation of confined LNG vapors is unlikely 
following an LNG spill caused by a terrorist attack. Experts were split 
on the likelihood of a confined deflagration occurring after a 
terrorist attack: eight thought it was unlikely, four thought it 
likely, and five thought neither likely nor unlikely.[Footnote 10] For 
unconfined spaces, experts were split on whether it is possible to 
induce such explosions; however, even experts who thought such 
explosions were possible agreed that deflagrations and detonations in 
unconfined spaces are unlikely to occur following an LNG spill caused 
by a terrorist attack. 

* BLEVE. Experts were split on whether a BLEVE is theoretically 
possible in an LNG tanker. Of the ten experts who agreed it was 
theoretically possible, six thought that a BLEVE is unlikely to occur 
following an LNG spill caused by a terrorist attack on a 
tanker.[Footnote 11] 

* Freeze burns and asphyxiation. Experts agreed that freeze burns do 
not present a hazard to the public because only people in close 
proximity to LNG spill, such as personnel on the tanker or nearby 
vessels, might come into contact with LNG or very cold LNG vapor. For 
asphyxiation, experts agreed that it is unlikely that an LNG vapor 
cloud could reach a populated area while still sufficiently 
concentrated to pose an asphyxiation threat. 

Experts Disagreed with a Few Key Conclusions of the Sandia National 
Laboratories Study: 

Experts disagreed with heat hazard and cascading failure conclusions of 
the Sandia study. Specifically, 7 of 15 experts thought Sandia's heat 
hazard distance was "about right," and the remaining 8 experts were 
evenly split as to whether the distance was "too conservative" (i.e., 
larger than needed to protect the public) or "not conservative enough" 
(i.e., too small to protect the public). Experts who thought the 
distance was too conservative generally listed one of two reasons. 
First, the assumptions about the surface emissive power of large fires 
were incorrect because the surface emissive power of large fires would 
be lower than Sandia assumed. Second, Sandia's hazard distances are 
based on the maximum size of a pool fire. However, these experts 
pointed out that once a pool fire ignites, its diameter will begin to 
shrink, which will also reduce the heat hazard distance. Experts who 
thought Sandia's heat hazard distance was not conservative enough 
listed a number of concerns. For example, Sandia's distances do not 
take into consideration the effects of cascading failure. One expert 
suggested that a 1-meter hole in the center tank of an LNG tanker that 
resulted in a pool fire could cause the near simultaneous failure of 
the other four tanks, leading to a larger heat hazard zone. 

Officials at Sandia National Laboratories and our panel of experts 
cautioned that the hazard distances presented cannot be applied to all 
sites. According to the Sandia study authors, their goal was to provide 
guidance to federal agencies on the order of magnitude of the hazards 
of an LNG spill on water. As they pointed out in interviews and in 
their original study, further analysis for specific sites is needed to 
understand hazards in a particular location. Six experts on our panel 
also emphasized the importance of site-specific and scenario-specific 
factors. For instance, one expert explained that the 5kW/m2 hazard 
distance depends on the size of the tanker and the spill scenario, 
including factors such as wind speed, timing of ignition, and the 
location of the hole. Other experts suggested that key factors are 
spill volume and the impact of waves. Additionally, two experts 
explained that there is no "bright line" for hazards--that is, 1,599 
meters is not necessarily "dangerous," and 1,601 meters is not 
necessarily "safe." 

Only 9 of 15 experts agreed with Sandia's conclusion that only three of 
the five LNG tanks on a tanker would be involved in cascading failure. 
Five experts noted that the Sandia study did not explain how it 
concluded that only three tanks would be involved in cascading failure. 
Three experts said that an LNG spill and subsequent fire could 
potentially result in the loss of all tanks on board the tanker. 

Twelve of 16 experts agreed, however, with Sandia's conclusion that 
cascading failure events are not likely to greatly increase (by more 
than 20 to 30 percent) the overall fire size or heat hazard ranges. The 
four experts who disagreed with Sandia's conclusion about the public 
safety impact of cascading failure cited two main reasons: (1) Sandia 
did not clearly explain how it reached that conclusion and (2) the 
impact of cascading failure will partly depend on how the incident 
unfolds. For instance, one expert suggested that cascading failure 
could include a tank rupture, fireball, or BLEVE, any of which could 
have direct impacts on the public (from the explosive force) and which 
would change the heat hazard zones that Sandia identified. 

Finally, experts agreed with Sandia's conclusion that consequence 
studies should be used to support comprehensive, risk-based management 
and planning approaches for identifying, preventing, and mitigating 
hazards from potential LNG spills. 

Experts Suggest Future Research Priorities to Determine the Public 
Safety Impact of an LNG Spill: 

In the second iteration of the Web-based panel, we asked the experts to 
identify the five areas related to the consequences of LNG spills that 
need further research. Then, in the final iteration of the Web-based 
panel, we provided the experts with a list of 19 areas--generated by 
their suggestions and comments from the second iteration--and asked 
them to rank these in order of importance. Table 2 presents the results 
of that ranking for the top 10 areas identified and indicates those 
areas that are funded in the DOE study discussed earlier. 

Table 2: Expert Panel's Ranking of Need for Research on LNG: 

Rank: 1; 
Research area: Large fire phenomena; 
Funded in: DOE's study: ÷. 

Rank: 2; 
Research area: Cascading failure; 
Funded in: DOE's study: [Empty]. 

Rank: 3; 
Research area: Large-scale spill testing on water; 
Funded in: DOE's study: ÷. 

Rank: 4; 
Research area: Large-scale fire testing; 
Funded in: DOE's study: ÷. 

Rank: 5; 
Research area: Comprehensive modeling: interaction of physical 
processes; 
Funded in: DOE's study: [Empty]. 

Rank: 6; 
Research area: Risk tolerability assessments; 
Funded in: DOE's study: [Empty]. 

Rank: 7; 
Research area: Vulnerability of containment systems (hole size); 
Funded in: DOE's study: [Empty]. 

Rank: 8; 
Research area: Mitigation techniques; 
Funded in: DOE's study: [Empty]. 

Rank: 9; 
Research area: Effect of sea water coming in as LNG flows out; 
Funded in: DOE's study: [Empty]. 

Rank: 10; 
Research area: Impact of wind, weather, and waves; 
Funded in: DOE's study: [Empty]. 

Source: GAO. 

Note: A rank of 1 is the highest rank, and a rank of 10 is the lowest. 
Panel members ranked 19 areas of research from 1 to 19; a score was 
calculated for each area based on this ranking. Only the 10 areas with 
the highest scores are presented in this table. 

[End of table] 

As the table shows, two of the top five research areas identified are 
related to large LNG fires--large fire phenomena and large-scale fire 
testing. Experts believe this research is needed to establish the 
relationship between large pool fires and the surface emissive power of 
the fire. Experts recommended new LNG tests for fires between 15 meters 
and 1,000 meters. The median suggested test size was 100 meters. Some 
experts also raised the issue of whether large LNG fires will stop 
behaving like one single flame but instead break up into several 
smaller, shorter flames. Sandia noted in its study that this behavior 
could reduce heat hazard distances by a factor of two to three. 

Experts also ranked research into cascading failure of LNG tanks second 
in the list of priorities. Concerning cascading failure, one expert 
noted that, although the consequences could be serious, there are 
virtually no data looking at the hull damage caused by exposure to 
extreme cold or heat. 

As table 2 shows, DOE's recently funded study involving large-scale LNG 
fire experiments addresses some, but not all, of the research 
priorities identified by the expert panel. For DOE, Sandia National 
Laboratories plans to conduct large-scale LNG pool fire tests beginning 
with a pool size of 35 meters--the same size as the largest test 
conducted to date. Sandia will validate the existing 35-meter data and 
then conduct similar tests for pool sizes up to 100 meters. The goal of 
this fire testing is to document the impact of smoke on large LNG pool 
fires. Sandia suggests that these tests will create a higher degree of 
knowledge of large-scale pool fire behavior and significantly lower the 
current uncertainty in predicting heat hazard distances. 

According to researchers at Sandia National Laboratories, some of the 
research our panel of experts suggested may not be appropriate. Sandia 
indicated that comprehensive modeling, which allows various complex 
processes to interact, would be very difficult to do because of the 
uncertainty surrounding each individual process of the model. One 
expert on our panel agreed, noting that while comprehensive modeling of 
all LNG phenomena is important, combining those phenomena into one 
model should wait for experiments that lead to better understanding of 
each individual phenomenon. 

Conclusions: 

It is likely that the United States will increasingly depend on the 
importation of LNG to meet the nation's demand for natural gas. 
Understanding and resolving the uncertainties surrounding LNG spills is 
critical, especially in deciding on where to locate LNG facilities. 
Because there have been no large-scale LNG spills or spill experiments, 
past studies have developed modeling assumptions based on small-scale 
spill data. While there is general agreement on the types of effects 
from an LNG spill, the results of these models have created what 
appears to be conflicting assessments of the specific consequences of 
an LNG spill, creating uncertainty for regulators and the public. 
Additional research to resolve some key areas of uncertainty could 
benefit federal agencies responsible for making informed decisions when 
approving LNG terminals and protecting existing terminals and tankers, 
as well as providing reliable information to citizens concerned about 
public safety. Although DOE has recently funded a study that will 
address large-scale LNG fires, this study will address only 3 of the 
top 10 issues--and not the second-highest ranked issue--that our panel 
of experts identified as potentially affecting public safety. 

Recommendation for Executive Action: 

To provide the most comprehensive and accurate information for 
assessing the public safety risks posed by tankers transiting to 
proposed LNG facilities, we recommend that the Secretary of Energy 
ensure that DOE incorporates the key issues identified by the expert 
panel into its current LNG study. We particularly recommend that DOE 
examine the potential for cascading failure of LNG tanks in order to 
understand the damage to the hull that could be caused by exposure to 
extreme cold or heat. 

Agency Comments and Our Evaluation: 

We requested comments on a draft of this report from the Secretary of 
Energy (DOE). DOE agreed with our findings and recommendation. In 
addition, DOE included technical and clarifying comments, which we 
included in our report as appropriate. 

As agreed with your offices, unless you publicly announce the contents 
of this report earlier, we plan no further distribution until 30 days 
from the report date. At that time, we will send copies to interested 
congressional committees, the Secretary of Energy, and other interested 
parties. We also will make copies available to others upon request. In 
addition, the report will be available at no charge on the GAO Web site 
at http://www.gao.gov. 

If you or your staff have any questions regarding this report, please 
contact me at (202) 512-3841 or wellsj@gao.gov. Contact points for our 
Offices of Congressional Relations and Public Affairs may be found on 
the last page of this report. Key contributors to this report are 
listed in appendix IV. 

Signed by: 

Jim Wells: 
Director, Natural Resources and Environment: 

[End of section] 

Appendix I: Scope and Methodology: 

To address the first objective, we identified eight unclassified, 
completed studies of liquefied natural gas (LNG) hazards and reviewed 
the six studies that included new, original research (either 
experimental or modeling) and clearly described the methodology used. 
While we have not verified the scientific modeling or results of these 
studies, the methods used seem appropriate for the work conducted based 
on conversations with experts in the field and our assessment. We also 
discussed these studies with their authors and visited all four onshore 
LNG import facilities and one export facility. We attended a 
presentation on LNG safety and received specific training on LNG 
properties and safety. We also conducted interviews with officials from 
Sandia National Laboratories, Federal Energy Regulatory Commission, 
Department of Transportation, Department of Energy, and the U. S. Coast 
Guard. During our interviews, we asked officials to provide information 
on past LNG studies and plans for future LNG spill consequences work. 

To obtain information on experts' opinions of the public safety 
consequences of an LNG spill from a tanker, we conducted a three-phase, 
Web-based survey of 19 experts on LNG spill consequences. We identified 
these experts from a list of 51 individuals who had expertise in one or 
more key aspects of LNG spill consequence analysis. In compiling this 
initial list, we sought to achieve balance in terms of area of 
expertise (i.e., LNG experiments, modeling LNG dispersion, LNG 
vaporization, fire modeling, and explosion modeling). In addition, we 
included at least one author of each of the six major LNG studies we 
reviewed, that is, studies by Sandia National Laboratories; ABS 
Consulting; Quest Consultants Inc; Pitblado, et al; James Fay (MIT); 
and William Lehr (National Oceanic and Atmospheric Administration). We 
gathered resumes, publication lists, and major LNG-related publications 
from the experts identified on the initial list. 

We selected 19 individuals for the panel. One or more of the following 
selection criteria were used: (1) has broad experience in all facets of 
LNG spill consequence modeling (LNG spill from hole, LNG dispersion, 
vaporization and pool formation, vapor cloud modeling, fire modeling, 
and explosion modeling); (2) has conducted physical LNG experiments; or 
(3) has specific experience with areas of particular importance, such 
as LNG explosion research. In addition, we included: (1) at least one 
author from each of the major LNG studies and (2) representatives from 
private industry, consulting, academia, and government. All 19 experts 
selected for the panel agreed to participate. The names and 
affiliations of panel members are included in appendix II. 

To obtain consensus concerning public safety issues, we used an 
iterative Web-based process. We used this method, in part, to eliminate 
the potential bias associated with group discussions. These biasing 
effects include the potential dominance of individuals and group 
pressure for conformity. Moreover, by creating a virtual panel, we were 
able to include more experts than possible with a live panel. 

For each phase in the process, we posted a questionnaire on GAO's 
survey Web site. Panel members were notified of the availability of the 
questionnaire with an e-mail message. The e-mail message contained a 
unique user name and password that allowed each respondent to log on 
and fill out a questionnaire but did not allow respondents access to 
the questionnaires of others. 

In the questionnaires, we asked the experts to agree or disagree with a 
set of statements about LNG hazards derived from GAO's synthesis of 
major LNG spill consequence studies. Prior to the first iteration, we 
had an LNG spill consequence expert who was not a part of the panel 
review each statement and provide comments about technical accuracy and 
tone. Experts were asked to indicate agreement on a 3-point scale 
(completely agree, generally agree, do not agree) and to provide 
comments about how the statements could be changed to better reflect 
their understanding of the consequences of LNG spills. 

If most experts agreed with a statement during the first iteration, we 
did not include it in the second iteration. If there was not agreement, 
we used the experts' comments to revise the statements for the second 
iteration. The second iteration was posted on the Web site, using the 
same protocol as used for the first. Again, panel members were asked to 
agree or disagree and provide narrative comments. We revised the 
statements where there was disagreement and posted them on the Web site 
again for the third iteration. At the end of the third iteration, at 
least 75 percent of the experts agreed or generally agreed with most of 
the ideas presented. 

Because some of the studies conducted are classified, this public 
version of our findings supplements a more comprehensive classified 
report produced under separate cover. We conducted our work from 
January 2006 through January 2007 in accordance with generally accepted 
government auditing standards. 

[End of section] 

Appendix II: Names and Affiliations of Members of GAO's Expert Panel on 
LNG Hazards: 

[End of section] 

Myron Casada	ABS Consulting  
	
T.Y. Chu	Sandia National Laboratories
	
Philip Cleaver	Advantica
	
Bob Corbin	U.S. Department of Energy 
	
John Cornwell	Quest Consultants, Inc.
	
James Fay	Massachusetts Institute of Technology 
	
Louis Gritzo	FM Global 
	
Jerry Havens	University of Arkansas 
	
Benedict Ho	BP 
	
Greg Jackson	University of Maryland 
	
Ron Koopman	Hazard Analysis Consulting
	
Bill Lehr	National Oceanic and Atmospheric Administration 
	
Georges Melhem	ioMosaic Corporation
	
Gordon Milne	Lloyd’s Register 
	
Robin Pitblado	Det Norske Veritas
	
Phani Raj	Technology and Management Systems, Inc.
	
Velisa Vesovic	Imperial College 
	
Harry West	Texas A&M University 
	
John Woodward	Baker Engineering and Risk Consultants, Inc.

[End of section] 

Appendix III: Summary of Expert Panel Results: 

For each question below, we show only those responses that were 
selected by at least one expert. The number of responses adds up to 19-
-the total number of experts on the panel. Percentages may not add to 
100% due to rounding. 

Introduction: 

Large LNG spills from a vessel could be caused by an accident, such as 
collision or grounding, or by an intentional attack. While large 
accidental LNG spills are highly unlikely given current LNG carrier 
designs and operational safety policies and practices, these spills do 
pose a hazard to the public if they occur in or near a populated area. 
What is your level of agreement with this paragraph? (Finalized in the 
second iteration.) 

Count: 8; 
Percentage: 42.11%; 
Label: Completely agree. 

Count: 11; 
Percentage: 57.89%; 
Label: Generally agree. 

[End of table] 

LNG Hazards: 

Overall Hazards: 

LNG is a cryogenic liquid composed primarily of methane with low 
concentrations of heavier hydrocarbons, such as ethane, propane, and 
butane. LNG is colorless, odorless, and nontoxic. When LNG is spilled, 
it boils and forms LNG vapor (natural gas). The LNG vapor is initially 
denser than ambient air and visible; 
LNG vapor will stay close to the surface as it mixes with air and 
disperses. LNG and LNG vapor pose four possible hazards: freeze burns, 
asphyxiation, fire hazard, and explosions. What is your level of 
agreement with this paragraph? (Finalized in the second iteration.) 

Count: 5; 
Percentage: 26.32%; 
Label: Completely agree. 

Count: 12; 
Percentage: 63.16%; 
Label: Generally agree. 

Count: 2; 
Percentage: 10.53%; 
Label: Do not agree. 

[End of table] 

LNG Hazards-Freeze Burns: 

LNG poses a threat of freeze burns to people who come into contact with 
the liquid or with very cold LNG vapor. Since LNG boils immediately and 
vaporizes after it leaves an LNG tank and LNG vapor warms as it mixes 
with air, only people in close proximity to the release, such as 
personnel on the tanker or nearby escort vessels, might come into 
contact with LNG or LNG vapor while it is still cold enough to result 
in freeze burns. Freeze burns do not present a direct hazard to the 
public. What is your level of agreement with this paragraph? (Finalized 
in the second iteration.) 

Count: 14; 
Percentage: 73.68%; 
Label: Completely agree. 

Count: 5; 
Percentage: 26.32%; 
Label: Generally agree. 

[End of table] 

LNG Hazards-Asphyxiation: 

After an LNG spill, LNG vapor forms a dense, visible vapor cloud that 
is initially heavier than air and remains close to the surface. The 
cloud warms as it mixes with air and as portions of the cloud reach 
ambient air temperatures, they begin to rise and disperse. Asphyxiation 
occurs when LNG vapor displaces oxygen in the air. Asphyxiation is a 
threat primarily to personnel on the LNG tanker or to people aboard 
vessels escorting the tanker at close range. An LNG vapor cloud could 
move away from the tanker as it mixes with air and begins to disperse. 
However, it is unlikely that the vapor cloud could reach a populated 
area while still sufficiently concentrated to pose an asphyxiation 
threat to the public. What is your level of agreement with this 
paragraph? (Finalized in the second iteration.) 

Count: 8; 
Percentage: 42.11%; 
Label: Completely agree. 

Count: 10; 
Percentage: 52.63%; 
Label: Generally agree. 

Count: 1; 
Percentage: 5.26%; 
Label: Do not agree. 

[End of table] 

LNG Hazards-Vapor Cloud: Wind Effect: 

The effect of wind on an LNG vapor cloud varies with wind speed. The 
most hazardous wind conditions, however, are low winds, which can push 
a vapor cloud downwind without accelerating the LNG vapor dispersion 
into the atmosphere. Low wind conditions have the highest potential of 
allowing an LNG vapor cloud to move a significant distance downwind. 
What is your level of agreement with this paragraph? (Finalized in the 
third iteration.) 

Count: 8; 
Percentage: 42.11%; 
Label: Completely agree. 

Count: 10; 
Percentage: 52.63%; 
Label: Generally agree. 

Count: 1; 
Percentage: 5.26%; 
Label: Do not agree . 

[End of table] 

LNG Hazards-Fire Hazard: 

Because LNG vapor in an approximately 5 to 15 percent mixture with air 
is flammable, LNG vapor within this flammability range is likely to 
ignite if it encounters a sufficiently strong ignition source such as a 
cigarette lighter or strong static charge. What is your level of 
agreement with this paragraph? (Finalized in the third iteration.) 

Count: 13; 
Percentage: 68.42%; 
Label: Completely agree. 

Count: 6; 
Percentage: 31.58%; 
Label: Generally agree. 

[End of table] 

LNG Hazards-Fire Hazard: Thermal Hazard End Point: 

The main hazard to the public from a pool fire is the thermal 
radiation, or heat, that is generated by the fire rather than the 
flames themselves. Often this heat is felt at considerable distance 
from the fire. Scientific papers have used two different thresholds as 
end points to describe the impact of thermal radiation on the public: 5 
kilowatts per square meter and 1.6 kilowatts per square meter. 
Which level do you think is the appropriate end point to use to define 
thermal hazard zones in order to protect the public? (Please indicate 
your response, then provide an explanation in the textbox below your 
answer.) 

Count: 8; 
Percentage: 42.11%; 
Label: 5 kilowatts per square meter. 

Count: 2; 
Percentage: 10.53%; 
Label: 1.6 kilowatts per square meter. 

Count: 6; 
Percentage: 31.58%; 
Label: Other. 

Count: 3; 
Percentage: 15.79%; 
Label: I do not have the expertise necessary to respond to this 
question. 

[End of table] 

Of the six experts who answered "other," two experts indicated that 
5kW/m2 is a useful or appropriate level for measuring the impact on 
people. One expert suggested that dosage (a measure that combines 
thermal radiation and duration of exposure) is most appropriate. 
Another expert suggested that both thresholds are appropriate, 
depending on the circumstances of the analysis. (Finalized in the first 
iteration.) 

LNG Hazards-Fire Hazard: Pool Fire: 

A pool fire could form in the wake of a vapor cloud fire burning back 
to the source or just after an LNG spill, if there is immediate 
ignition of the LNG vapor. A pool fire burns the vapor above a liquid 
LNG pool as the liquid boils from the pool. A large, unconfined pool 
fire is very difficult to extinguish; generally almost all the LNG must 
be consumed before the fire goes out. What is your level of agreement 
with this paragraph? (Finalized in the second iteration.) 

Count: 13; 
Percentage: 68.42%; 
Label: Completely agree. 

Count: 5; 
Percentage: 26.32%; 
Label: Generally agree. 

Count: 1; 
Percentage: 5.26%; 
Label: Do not agree. 

[End of table] 

The main hazard to the public from a pool fire is the thermal 
radiation, or heat, from the fire. This heat can be felt at a 
considerable distance from the flames themselves. Numerous factors can 
impact the amount of thermal radiation that could affect the public: 
site-specific weather conditions, including humidity and wind speed and 
direction, the composition of the LNG, and the size of the fire. What 
is your level of agreement with this paragraph? (Finalized in the 
second iteration.) 

Count: 13; 
Percentage: 68.42%; 
Label: Completely agree. 

Count: 6; 
Percentage: 31.58%; 
Label: Generally agree. 

[End of table] 

The wind speed and direction also affect the distance at which thermal 
radiation from the fire is felt by the public. In high winds, the 
flames will tilt downwind, increasing the amount of heat felt downwind 
of the fire and decreasing the amount of heat felt upwind. More humid 
conditions allow heat to be absorbed by the moisture in the air 
reducing the heat felt by the public. What is your level of agreement 
with the above paragraph? (Finalized in the second iteration.) 

Count: 6; 
Percentage: 31.58%; 
Label: Completely agree. 

Count: 11; 
Percentage: 57.89%; 
Label: Generally agree but suggest the following clarification. 

Count: 2; 
Percentage: 10.53%; 
Label: I do not have the expertise necessary to respond to this 
section. 

[End of table] 

The composition of the LNG can also affect the distance at which 
thermal radiation from the fire is felt by the public. In small fires, 
methane, which comprises between 84 percent and 97 percent of LNG, 
burns cleanly, with little smoke. Cleaner-burning LNG fires, 
particularly those burning LNG with higher methane content, result in 
higher levels of thermal radiation than oil or gasoline fires of the 
same size because the smoke generated by oil and gasoline fires acts as 
a shield, reducing the amount of thermal radiation emitted by the fire. 
While LNG composition can have a large impact on the thermal radiation 
from small LNG fires, as LNG fires get larger, these effects are 
hypothesized to be less pronounced. What is your level of agreement 
with this paragraph? (Finalized in the third iteration.) 

Count: 5; 
Percentage: 26.32%; 
Label: Completely agree. 

Count: 10; 
Percentage: 52.63%; 
Label: Generally agree. 

Count: 3; 
Percentage: 15.79%; 
Label: Do not agree. 

Count: 1; 
Percentage: 5.26%; 
Label: I do not have the expertise necessary to respond to this 
section. 

[End of table] 

The size of the fire has a major impact on the thermal radiation from 
an LNG pool fire. Thermal radiation increases with pool size up to a 
point but is expected to decrease for very large pools, like those 
caused by a terrorist attack. What is your level of agreement with this 
paragraph? (Finalized in the second iteration.) 

Count: 4; 
Percentage: 21.05%; 
Label: Completely agree. 

Count: 10; 
Percentage: 52.63%; 
Label: Generally agree. 

Count: 4; 
Percentage: 21.05%; 
Label: Do not agree. 

Count: 1; 
Percentage: 5.26%; 
Label: I do not have the expertise necessary to respond to this 
section. 

[End of table] 

LNG Hazards-Vapor Cloud Fire: 

If an LNG vapor cloud formed in the wake of an LNG spill and drifted 
away from the tanker as it warmed and dispersed, the vapor cloud could 
enter a populated area while areas of the cloud had LNG vapor/air 
mixtures within the flammability range. Since populated areas have 
numerous ignition sources, those portions of the cloud would likely 
ignite. The fire would then burn back through the cloud toward the 
tanker and continue to burn as a pool fire near the ship, assuming that 
liquid LNG still remains in the spill area. Ignition of a vapor cloud 
over a populated area could burn people and property in the immediate 
vicinity of the fire. While the initial fire would be of short duration 
as the flames burned back toward the LNG carrier, secondary fires could 
continue to present a hazard to the public. What is your level of 
agreement with the above paragraph? (Finalized in the second 
iteration.) 

Count: 7; 
Percentage: 36.84%; 
Label: Completely agree. 

Count: 11; 
Percentage: 57.89%; 
Label: Generally agree but suggest the following clarification. 

Count: 1; 
Percentage: 5.26%; 
Label: Do not agree. 

[End of table] 

LNG Hazards-Vapor Cloud Fire: Burn Back Speed: 

After ignition of a vapor cloud that drifted away from an LNG tanker 
spill, how fast could the flame front travel back toward the spill site 
if it was unconfined or confined? (Finalized in the second iteration.) 

Count: 15; 
Percentage: 78.95%; 
Label: Not checked. 

Count: 2; 
Percentage: 10.53%; 
Label: I do not have the expertise necessary to respond to this 
section. 

Count: 2; 
Percentage: 10.53%; 
Label: No answer. 

[End of table] 

Experts did not agree on the speed of a flame front traveling through 
an LNG vapor cloud in either a confined or unconfined state. Responses 
varied from less than 5 meters per second up to 50 meters per second in 
unconfined settings and from 0 meters per second to 2,000 meters per 
second in confined settings. 

Explosions-RPT: 

A rapid phase transition (RPT) can occur when LNG spilled onto water 
changes from liquid to gas virtually instantaneously due to the rapid 
absorption of ambient environmental heat. While the rapid expansion 
from a liquid to vapor state can cause locally large overpressures, an 
RPT does not involve combustion. RPTs have been observed during LNG 
test spills onto water. In some cases, the overpressures generated were 
strong enough to damage test equipment in the immediate vicinity. 
Overpressures generated from RPTs would be very unlikely to have a 
direct affect on the public. What is your level of agreement with this 
paragraph? (Finalized in the second iteration.) 

Count: 15; 
Percentage: 78.95%; 
Label: Completely agree. 

Count: 4; 
Percentage: 21.05%; 
Label: Generally agree. 

[End of table] 

Explosions-Deflagrations and Detonations: 

Deflagrations and detonations are rapid combustion processes that move 
through an unburned fuel-air mixture. Deflagrations move at subsonic 
velocities and can result in overpressures up to eight times the 
original pressure, particularly in congested/confined areas. 
Detonations move at supersonic velocities and can result in 
overpressures up to 20 times the original pressure. What is your level 
of agreement with this paragraph? (Finalized in the third iteration.) 

Count: 1; 
Percentage: 5.26%; 
Label: Not checked. 

Count: 7; 
Percentage: 36.84%; 
Label: Completely agree. 

Count: 10; 
Percentage: 52.63%; 
Label: Generally agree. 

Count: 1; 
Percentage: 5.26%; 
Label: Do not agree. 

[End of table] 

Explosions--Deflagrations, Detonations, and BLEVEs: 

Please choose the response that best describes your opinion about each 
type of explosion of LNG vapors in each setting described. (Finalized 
in the third iteration.) 

Answer: Under controlled experimental conditions, it is possible to 
induce this type of explosion in this type of setting; 
Deflagration with overpressure in an unconfined setting: 7; 
Deflagration with overpressure in a confined setting: 18; 
Detonation in an unconfined setting: 4; 
Detonation in a confined setting: 15; 
Boiling-liquid- expanding-vapor-explosion (BLEVE): 11. 

Answer: This type of setting cannot support this type of explosion; 
Deflagration with overpressure in an unconfined setting: 8; 
Deflagration with overpressure in a confined setting: 0; 
Detonation in an unconfined setting: 11; 
Detonation in a confined setting: 2; 
Boiling-liquid-expanding-vapor-explosion (BLEVE): 7. 

Answer: More research is necessary to answer this question; 
Deflagration with overpressure in an unconfined setting: 3; 
Deflagration with overpressure in a confined setting: 0; 
Detonation in an unconfined setting: 3; 
Detonation in a confined setting: 0; 
Boiling- liquid-expanding-vapor-explosion (BLEVE): 0. 

Answer: I don't have the expertise necessary to answer this question; 
Deflagration with overpressure in an unconfined setting: 0; 
Deflagration with overpressure in a confined setting: 0; 
Detonation in an unconfined setting: 0; 
Detonation in a confined setting: 1; 
Boiling- liquid-expanding-vapor-explosion (BLEVE): 0. 

Answer: No answer/not checked; 
Deflagration with overpressure in an unconfined setting: 1; 
Deflagration with overpressure in a confined setting: 1; 
Detonation in an unconfined setting: 1; 
Detonation in a confined setting: 1; 
Boiling-liquid-expanding-vapor-explosion (BLEVE): 1. 

[End of table] 

If experts answered that "under controlled experimental conditions, it 
is possible to induce this type of explosion in this type of setting," 
they were asked to answer the following question: 

What is the likelihood of a each type of explosion of LNG vapors in 
each setting described occurring following an LNG spill caused by a 
terrorist attack on a tanker? (Finalized in the third iteration.) 

Answer: Highly unlikely; 
Deflagration with overpressure in an unconfined setting: 3; 
Deflagration with overpressure in a confined setting: 6; 
Detonation in an unconfined setting: 1; 
Detonation in a confined setting: 7; 
Boiling-liquid-expanding-vapor-explosion (BLEVE): 4. 

Answer: Unlikely ; 
Deflagration with overpressure in an unconfined setting: 2; 
Deflagration with overpressure in a confined setting: 2; 
Detonation in an unconfined setting: 3; 
Detonation in a confined setting: 3; 
Boiling-liquid-expanding-vapor-explosion (BLEVE): 2. 

Answer: Neither likely nor unlikely ; 
Deflagration with overpressure in an unconfined setting: 1; 
Deflagration with overpressure in a confined setting: 5; 
Detonation in an unconfined setting: 0; 
Detonation in a confined setting: 3; 
Boiling-liquid-expanding-vapor-explosion (BLEVE): 3. 

Answer: Likely; 
Deflagration with overpressure in an unconfined setting: 1; 
Deflagration with overpressure in a confined setting: 4; 
Detonation in an unconfined setting: 0; 
Detonation in a confined setting: 2; 
Boiling-liquid-expanding-vapor-explosion (BLEVE): 1. 

Answer: Highly likely; 
Deflagration with overpressure in an unconfined setting: 0; 
Deflagration with overpressure in a confined setting: 0; 
Detonation in an unconfined setting: 0; 
Detonation in a confined setting: 0; 
Boiling-liquid-expanding-vapor-explosion (BLEVE): 0. 

Answer: No answer/ not checked; 
Deflagration with overpressure in an unconfined setting: 0; 
Deflagration with overpressure in a confined setting: 1; 
Detonation in an unconfined setting: 0; 
Detonation in a confined setting: 0; 
Boiling-liquid-expanding-vapor-explosion (BLEVE): 1. 

[End of table] 

LNG Hazards-Is BLEVE the Worst? 

A BLEVE is the worst potential hazard of an LNG spill. It would result 
in the rupture of one or more LNG tanks, perhaps simultaneously, on the 
ship, with potential rocketing debris and damaging pressure waves. What 
is your level of agreement with the above paragraph? (Finalized in the 
first iteration.) 

Count: 2; 
Percentage: 10.53%; 
Label: Completely agree. 

Count: 16; 
Percentage: 84.21%; 
Label: Do not agree (Please explain in the textbox below.) 

Count: 1; 
Percentage: 5.26%; 
Label: No answer. 

[End of table] 

Questions About the 2004 Sandia National Laboratories Study[Footnote 
12] 

The Sandia report concluded that the most significant impacts to public 
safety exist within 500 meters of a spill, with much lower impacts at 
distances beyond 1,600 meters even for very large spills. Please choose 
the response that best describes your opinion about these hazard 
distances. (Finalized in the third iteration.) 

Count: 4; 
Percentage: 23.54%; 
Label: They are too conservative (i.e., should be smaller). 

Count: 7; 
Percentage: 41.18%; 
Label: They are about right. 

Count: 4; 
Percentage: 23.53%; 
Label: They are not conservative enough (i.e., should be larger). 

Count: 2; 
Percentage: 11.76%; 
Label: No answer. 

[End of table] 

The Sandia report concluded that large, unignited LNG vapor clouds 
could spread over distances greater than 1,600 meters from a spill. For 
a nominal intentional spill, the hazard range could extend to 2,500 
meters. The actual hazard distances will depend on breach and spill 
size, site-specific conditions, and environmental conditions. Please 
choose the response that best describes your opinion about these hazard 
distances. (Finalized in the third iteration.) 

Count: 4; 
Percentage: 23.53%; 
Label: They are too conservative (i.e., should be smaller). 

Count: 6; 
Percentage: 35.29%; 
Label: They are about right. 

Count: 4; 
Percentage: 23.53%; 
Label: They are not conservative enough (i.e., should be larger). 

Count: 1; 
Percentage: 5.88%; 
Label: Do not have the expertise to answer. 

Count: 2; 
Percentage: 11.76%; 
Label: No answer. 

[End of table] 

The Sandia report concluded that cascading damage (multiple cargo tank 
failure) due to brittle fracture from exposure to cryogenic liquid or 
fire-induced damage to foam insulation is possible under certain 
conditions but is not likely to involve more than two or three cargo 
tanks for any single incident. What is your level of agreement with 
this paragraph? (Finalized in the third iteration.) 

Count: 3; 
Percentage: 17.65%; 
Label: Completely agree. 

Count: 6; 
Percentage: 35.29%; 
Label: Generally agree. 

Count: 6; 
Percentage: 35.29%; 
Label: Do not agree. 

Count: 2; 
Percentage: 11.76%; 
Label: I do not have the expertise necessary to respond to this 
section. 

[End of table] 

The Sandia report concluded that cascading events are not expected to 
greatly increase (not more than 20-30 percent) the overall fire size or 
hazard ranges (500 meters for severe impacts, much lower impacts beyond 
1,600 meters) but will increase the expected fire duration. What is 
your level of agreement with this paragraph? (Finalized in the third 
iteration.) 

Count: 7; 
Percentage: 41.18%; 
Label: Completely agree. 

Count: 5; 
Percentage: 29.41%; 
Label: Generally agree. 

Count: 4; 
Percentage: 23.53%; 
Label: Do not agree. 

Count: 1; 
Percentage: 5.88%; 
Label: No answer. 

[End of table] 

The Sandia report suggested that consequence studies should be used to 
support comprehensive, risk-based management and planning approaches 
for identifying, preventing, and mitigating hazards to public safety 
and property from potential LNG spills. What is your level of agreement 
with this paragraph? (Finalized in the third iteration.) 

Count: 8; 
Percentage: 47.06%; 
Label: Completely agree. 

Count: 8; 
Percentage: 47.06%; 
Label: Generally agree. 

Count: 1; 
Percentage: 5.88%; 
Label: Do not agree. 

[End of table] 

Commodity Comparison: 

In your opinion, what is the risk to public safety posed by an attack 
on tankers carrying each of the following energy commodities? 
(Finalized in the first iteration.) 

Answer: Little to None; 
Liquefied natural gas: 1; 
Crude oil: 2; 
Diesel: 1; 
Gasoline: 0; 
Heating oil: 1; 
Jet fuel: 1; 
Liquefied petroleum gas: 0. 

Answer: Little; 
Liquefied natural gas: 3; 
Crude oil: 10; 
Diesel: 11; 
Gasoline: 5; 
Heating oil: 11; 
Jet fuel: 6; 
Liquefied petroleum gas: 1. 

Answer: Medium; 
Liquefied natural gas: 6; 
Crude oil: 3; 
Diesel: 3; 
Gasoline: 8; 
Heating oil: 3; 
Jet fuel: 6; 
Liquefied petroleum gas: 4. 

Answer: Large; 
Liquefied natural gas: 3; 
Crude oil: 0; 
Diesel: 0; 
Gasoline: 2; 
Heating oil: 0; 
Jet fuel: 2; 
Liquefied petroleum gas: 5. 

Answer: Very Large; 
Liquefied natural gas: 2; 
Crude oil: 0; 
Diesel: 0; 
Gasoline: 0; 
Heating oil: 0; 
Jet fuel: 0; 
Liquefied petroleum gas: 5. 

Answer: No expertise to answer; 
Liquefied natural gas: 1; 
Crude oil: 1; 
Diesel: 1; 
Gasoline: 1; 
Heating oil: 1; 
Jet fuel: 1; 
Liquefied petroleum gas: 1. 

Answer: No answer; 
Liquefied natural gas: 3; 
Crude oil: 3; 
Diesel: 3; 
Gasoline: 3; 
Heating oil: 3; 
Jet fuel: 3; 
Liquefied petroleum gas: 3. 

[End of table] 

Future Research: 

In the first and second survey iterations, you noted areas related to 
LNG spill consequences that need further research. We are interested in 
your thoughts on the relative level of need for research in these 
areas, and also the five areas you think should be of highest priority 
in future research. 

Please indicate the degree to which further research is needed in each 
of the areas listed below. (Finalized in the third iteration.) 

Responses to each part of this question are in the table below, which 
is sorted by mean score so that the highest-ranked research priorities 
appear first. 

Type of research: Large fire phenomena (impact of smoke shielding, 
large flame versus smaller flamelets); 
Very great need (1): 9; 
Great need (2): 5; 
Moderate need (3): 3; 
Some need (4) : 0; 
Little to no need (5): 1; 
Do not have the expertise to answer (6): 1; 
No answer (7) : 0; 
Mean score: 4.17. 

Type of research: Cascading failure; 
Very great need (1): 5; 
Great need (2): 9; 
Moderate need (3): 4; 
Some need (4) : 1; 
Little to no need (5): 0; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 3.95. 

Type of research: Large-scale LNG spill testing on water[A]; 
Very great need (1): 7; 
Great need (2): 7; 
Moderate need (3): 2; 
Some need (4) : 1; 
Little to no need (5): 2; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 3.84. 

Type of research: Large-scale fire testing[B]; 
Very great need (1): 7; 
Great need (2): 6; 
Moderate need (3): 3; 
Some need (4) : 2; 
Little to no need (5): 1; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 3.84. 

Type of research: Comprehensive modeling allowing different physical 
processes to interact; 
Very great need (1): 2; 
Great need (2): 10; 
Moderate need (3): 3; 
Some need (4) : 4; 
Little to no need (5): 0; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 3.53. 

Type of research: Risk tolerability assessments; 
Very great need (1): 5; 
Great need (2): 4; 
Moderate need (3): 3; 
Some need (4) : 1; 
Little to no need (5): 3; 
Do not have the expertise to answer (6): 1; 
No answer (7) : 2; 
Mean score: 3.44. 

Type of research: Vulnerability of LNG containment systems, including 
validating hole size predictions for the double hull ship structure; 
Very great need (1): 5; 
Great need (2): 4; 
Moderate need (3): 3; 
Some need (4) : 5; 
Little to no need (5): 2; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 3.26. 

Type of research: Mitigation techniques; 
Very great need (1): 3; 
Great need (2): 5; 
Moderate need (3): 6; 
Some need (4) : 3; 
Little to no need (5): 2; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 3.21. 

Type of research: Effect of sea water pouring into a hole as LNG flows 
out; 
Very great need (1): 2; 
Great need (2): 6; 
Moderate need (3): 5; 
Some need (4) : 3; 
Little to no need (5): 2; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 1; 
Mean score: 3.17. 

Type of research: Impact of wind, weather, and waves (on pool spread 
size, evaporation rate, pool formation, etc.); 
Very great need (1): 3; 
Great need (2): 4; 
Moderate need (3): 6; 
Some need (4) : 3; 
Little to no need (5): 3; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 3.05. 

Type of research: Improvements to 3-D computational fluid dynamics 
dispersion modeling; 
Very great need (1): 0; 
Great need (2): 4; 
Moderate need (3): 6; 
Some need (4) : 6; 
Little to no need (5): 2; 
Do not have the expertise to answer (6): 1; 
No answer (7) : 0; 
Mean score: 2.67. 

Type of research: Effects of different LNG compositions (on 
vaporization rates, thermal radiation, explosive behavior, etc.); 
Very great need (1): 2; 
Great need (2): 2; 
Moderate need (3): 4; 
Some need (4) : 8; 
Little to no need (5): 3; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 2.58. 

Type of research: Whether an explosive attack will result in immediate 
vapor cloud ignition; 
Very great need (1): 0; 
Great need (2): 5; 
Moderate need (3): 4; 
Some need (4) : 5; 
Little to no need (5): 4; 
Do not have the expertise to answer (6): 1; 
No answer (7) : 0; 
Mean score: 2.56. 

Type of research: Rapid phase transitions: likelihood in various 
scenarios and impact; 
Very great need (1): 1; 
Great need (2): 2; 
Moderate need (3): 6; 
Some need (4) : 6; 
Little to no need (5): 4; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 2.47. 

Type of research: Effects of igniting LNG vapors in containment or 
ballast tanks; 
Very great need (1): 0; 
Great need (2): 5; 
Moderate need (3): 3; 
Some need (4) : 5; 
Little to no need (5): 6; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 2.37. 

Type of research: BLEVE properties of tanks on LNG ships; 
Very great need (1): 1; 
Great need (2): 4; 
Moderate need (3): 3; 
Some need (4) : 4; 
Little to no need (5): 7; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 2.37. 

Type of research: Deflagration/detonation of LNG; 
Very great need (1): 1; 
Great need (2): 0; 
Moderate need (3): 5; 
Some need (4) : 8; 
Little to no need (5): 5; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 2.16. 

Type of research: Effects of a large, unignited vapor cloud drifting 
from the incident site; 
Very great need (1): 0; 
Great need (2): 0; 
Moderate need (3): 7; 
Some need (4) : 5; 
Little to no need (5): 7; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 2.00. 

Type of research: Effect of clothing and obstructions on the radiant 
heat level received by the public; 
Very great need (1): 1; 
Great need (2): 1; 
Moderate need (3): 2; 
Some need (4) : 6; 
Little to no need (5): 9; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 0; 
Mean score: 1.89. 

Type of research: Other[C]; 
Very great need (1): 12; 
Great need (2): 2; 
Moderate need (3): 0; 
Some need (4) : 0; 
Little to no need (5): 0; 
Do not have the expertise to answer (6): 0; 
No answer (7) : 5; 
Mean score: [D]. 

[A] Experts suggested pool sizes of 15 meters up to 1,000 meters, 
though the median response was 100 meters. 

[B] Experts suggested pool sizes of 15 meters up to 1,000 meters, 
though the median response was 100 meters. 

[C] Experts suggested frequency modeling, determination of acceptable 
risk to society, analysis of foam on LNG tankers, risk analysis for 
larger LNG tankers, CFD modeling for pool spreading and evaporation, 
and improvement to existing techniques used for fighting LNG fires. 

[D] Not applicable. 

[End of table] 

[End of section] 

Appendix IV: GAO Contact and Staff Acknowledgments: 

GAO Contact: 

Jim Wells, (202) 512-3841, or wellsj@gao.gov: 

Staff Acknowledgments: 

In addition to the individual named above, Mark Gaffigan, Amy Higgins, 
Lynn Musser, Janice Poling, Rebecca Shea, Carol Herrnstadt Shulman, and 
James W. Turkett made key contributions to this report. 

(360675): 

FOOTNOTES 

[1] The onshore facilities are near Boston, Massachusetts; 
Cove Point, Maryland; 
Savannah, Georgia; 
and Lake Charles, Louisiana. The United States also has one LNG export 
facility in Kenai, Alaska, that ships LNG to Japan. 

[2] Under the Natural Gas Act, as amended, FERC has exclusive authority 
to approve or deny an application for the siting, construction, or 
operation of onshore LNG terminals, including pipelines, and offshore 
facilities in state waters--that is, generally within 3 miles of shore. 

[3] The Coast Guard, along with the Department of Transportation's 
Maritime Administration, has jurisdiction under the Deep Water Port Act 
of 1974, as amended, to approve the siting and operation of offshore 
LNG facilities in federal waters. 

[4] LNG vapors only ignite when they are in a 5 percent to 15 percent 
concentration in the air. If the LNG concentration is higher, there is 
not enough oxygen available for fire. If the concentration is lower, 
there is likewise not enough fuel for fire. 

[5] Sandia National Laboratories. Guidance on Risk Analysis and Safety 
Implications of a Large Liquefied Natural Gas (LNG) Spill Over Water. 
Albuquerque: 2004. 

[6] DOE is also sponsoring additional research that applies the 2004 
Sandia National Laboratories' methodology to LNG tankers larger than 
those previously studied, which is expected to be completed in July 
2007. 

[7] Generally, an explosion is an energy release associated with a 
pressure wave. Some explosions are large enough that the pressure wave 
can break windows or damage structures, while others are much smaller. 

[8] Please note that the information used to develop Sandia's terrorist 
scenarios is classified and will be discussed in GAO's classified 
report. 

[9] We considered experts "in agreement" if more than 75 percent of 
experts indicated that they completely agreed or generally agreed with 
a given statement. Not all experts commented on every issue discussed. 

[10] Two experts did not comment. 

[11] Three experts said that BLEVEs were "neither likely nor unlikely," 
and one expert thought that BLEVEs were likely. 

[12] Since two of the experts were authors of the Sandia study, their 
responses to ALL the questions related to the study below have been 
excluded. For the questions related to the Sandia study, there are 17 
experts responding. 

GAO's Mission: 

The Government Accountability Office, the audit, evaluation and 
investigative arm of Congress, exists to support Congress in meeting 
its constitutional responsibilities and to help improve the performance 
and accountability of the federal government for the American people. 
GAO examines the use of public funds; evaluates federal programs and 
policies; and provides analyses, recommendations, and other assistance 
to help Congress make informed oversight, policy, and funding 
decisions. GAO's commitment to good government is reflected in its core 
values of accountability, integrity, and reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through GAO's Web site (www.gao.gov). Each weekday, GAO posts 
newly released reports, testimony, and correspondence on its Web site. 
To have GAO e-mail you a list of newly posted products every afternoon, 
go to www.gao.gov and select "Subscribe to Updates." 

Order by Mail or Phone: 

The first copy of each printed report is free. Additional copies are $2 
each. A check or money order should be made out to the Superintendent 
of Documents. GAO also accepts VISA and Mastercard. Orders for 100 or 
more copies mailed to a single address are discounted 25 percent. 
Orders should be sent to: 

U.S. Government Accountability Office 441 G Street NW, Room LM 
Washington, D.C. 20548: 

To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202) 
512-6061: 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 

Web site: www.gao.gov/fraudnet/fraudnet.htm E-mail: fraudnet@gao.gov 
Automated answering system: (800) 424-5454 or (202) 512-7470: 

Congressional Relations: 

Gloria Jarmon, Managing Director, JarmonG@gao.gov (202) 512-4400 U.S. 
Government Accountability Office, 441 G Street NW, Room 7125 
Washington, D.C. 20548: 

Public Affairs: 

Paul Anderson, Managing Director, AndersonP1@gao.gov (202) 512-4800 
U.S. Government Accountability Office, 441 G Street NW, Room 7149 
Washington, D.C. 20548: