Our site needs your help!
Site categories

Fire Scenarios on liquefied gas carriers

Join Our Telegram (Seaman Community)

Some common fire scenarios on LNG carriers include engine room fires, cargo tank fires, and leaks leading to fires. It’s important to have proper safety measures in place, such as fire detection and suppression systems, trained emergency response teams, and regular drills to ensure the crew is prepared to handle any fire emergency.


The primary aim of this section is to:

1 Focus on the particular elements in the allotted task, while not neglecting to bring to the other working group members issues that the subgroup considered influential or related to these elements for their consideration.

2 compile any reservations they had on the input information both to:

  • show they had considered these and possibly discounted them;


  • they had prepared important reservations not discounted, if any, for consideration by the wider working group;

3 conclude – finally the subgroup concluded on the most representative scenario.

As was discussed during the first WG meeting, the Sandia report is widely taken as a seminal work on this subject. There are, however, other studies that were used that both reinforce the findings of the Sandia report or question its findings. This ensured that a considered analysis was made, as opposed to simply accepting the Sandia report without question, when it remains clear that there are concerns within the industry over some of its findings. These concerns revolve around quantification of the following points:

  • Large fire scenarios are required to threaten significant parts of a cargo tanks;
  • large fires may occur if an LNG tank is ruptured and LNG dispersed to the sea;
  • the release rate determines both the size of the fire as well as the duration;
  • collision damage causing tank ruptures of, e. g. 1 m2, will fuel very large fires (dia > 300 m) but will have short duration (typically less than 15 min) due to the emptying of the tanks;
  • smaller releases can fuel fires of longer duration with the potential of threatening a large part of one side of the vessel;
  • these fires will have a very large flame height and may affect large part of the tanks above the main deck;
  • experimental data from large LNG fires on water does not exist and the effects of oxygen starvation and possible smoke screening are not known. A conservative approach would be to apply heat loads from large pool fires without smoke screening;
  • such fires will burn as pool fires with radiant heat loads 150 kW/m2 and convective heat transfer from the combustion gases with temperatures 1 000-1 100 °C.
Studies and Documentation Reviewed
Document №TitleOrganisationDate
1Guidance on Risk Analysis and Safety Implications of a Large Liquefied Natural Gas (LNG) Spill Over WaterSandia Report2004
2Results of 40 m3 LNG spills on waterLawrence Livermore National Laboratory(circa 1983)
3Spill tests of LNG and Refrigerated Liquid propane on the sea, Maplin Sands, UK, 1980Shell Research Ltd.1980
4Consequence Assessment Methods for Incidents Involving Releases from Liquefied Natural Gas CarriersABS on behalf of FERCMay 13th 2004
5Consequences of LNG Marine IncidentsDNV – CCPS Conference OrlandoJune 29-July 1 2004
6Consequence modelling of LNG Marine IncidentsDNV – Baik, Raghunathan, Witlox – American Society of Safety EngineersMarch 18-22, 2005
7Potential for BLEVE Associated with Marine LNG Vessel FiresDNV – Dr. Robin PitbladoAugust 2006
8Model of Large Pool FiresJ. A. Fay – MITsept-05
9Large Hydrocarbon fuel pool fires: Physical characteristics and thermal emission variations with heightPhani K. Raj – Technology & Management Systems18th August 2006
10Spread of Large LNG pools on the seaJ. A. Fay – MITOctober 2006
11LNG Properties & Hazards – Understanding LNG Rapid Phase Transitions (RPT)ioMosaic – G. Melhem,S. Saraf, H. Ozog2006
12Report on the Outline of Collision between Japanese tanker Yuyo Maru No. 10 & Liberia Freighter Pacific AresMaru No. 10 & Liberia Freighter Pacific Ares MSA – Japanese GovernmentMarch 1975
13LNG Decisions Making Approaches ComparedDr. R. Piblado, Dr. J. Baik, V. Raghunathan – DNV2005
14FDS LNG pool fire simulations: a preliminary study on the application of FDS to study potential marine tanker accidentsJ E S Venart – University of New Brunswick1st June 2006
15Thermal Response of Gas Carriers to Hydrocarbon FiresDNV For Shell International Marine Ltd.26th November 1980
16Public Safety Consequences of a terrorist attack on a tanker carrying LNG need clarificationUS Government Accountability OfficeFebruary 2007
17Department of Homeland security : LNG Tanker Security – Direct Testimony Dr. Phani K. RajDr. Phani K. Raj – President – Technology & Management Systems, Inc.
18Maplin Sands experiments 1980 – Interpretations and modelling of Liquefied Gas Spills onto SeaShell Research B. V. & Shell Research Ltd.1980 (Pub 1983)
19Maplin Sands experiments 1980 – Dispersion results fro continuous releases of refrigerated liquid propane & LNGShell Research B. V. & Shell Research Ltd.1980 (Pub 1984)
20Maplin Sands experiments 1980 – Dispersion & combustion behaviour of gas clouds resulting from large spillages of LNG & LPG on the seaShell Research B. V. & Shell Research Ltd.1980 (Pub 1982)
21Maplin Sands experiments 1980 – Dispersion results from continuous releases of refrigerated liquid propaneShell Research B. V. & Shell Research Ltd.1980 (Pub 1982)
22Review of published experimental resultsLloyd’s Register of ShippingLast ref. 1992
Open table in new window

Summary of Documentation Review

Listed below are the salient points extracted from the documents submitted or researched that were reviewed by the subgroup. These points were those the group considered relevant to the terms of reference of the WG.

Those papers not referenced were considered not to have input directly related to the remit of the subgroup or the full WG.

Guidance on Risk Analysis and Safety Implications of a Large Liquefied Natural Gas (LNG) Spill Over Water – Document No. 1

The Sandia report is a compilation of previous work carried out in the field of Environmental Impact of Liquefied Natural GasLNG spills. The report summarises the different work and then performs various analyses with the assumption of certain inputs from the different input sources.

The report draws a number of conclusions, some of which are directly related to matters being discussed in this WG and within the remit of the work assigned to sub-group A.

Key results extracted

  • LNG cargo tank hole sizes for most credible threats range from two to twelve square metres, expected sizes for intentional threats are nominally five square metres;
  • the most significant impacts to public safety and property exist within approximately 500 m of a spill, due to thermal hazards from fires, with lower public health and safety impacts at distances beyond approximately 1 600 m;
  • large, unignited LNG vapour releases are unlikely. If they do not ignite, vapour clouds could spread over distances greater than 1 600 m from a spill. For nominal accidental spills, the resulting hazard ranges could extend up to 1 700 m. For a nominal intentional spill, the hazard range could extend to 2 500 m. The actual hazard distances will depend on breach and spill size, site-specific conditions, and environmental conditions
  • cascading damage (multiple cargo tank failures) due to brittle fracture from exposure to cryogenic liquid or fire- induced damage to foam insulation was considered. Such releases were evaluated and, while possible under certain conditions, are not likely to involve more than two or three cargo tanks for any single incident. Cascading events were analysed and are not expected to greatly increase (not more than 20 %-30 %) the overall fire size or hazard ranges noted above, but will increase the expected fire duration.

Cloud fire experiments

Table 1 provides an overview of existing LNG spill testing data.

Table 1. Largest Spill Volumes Tested to Date Giving Pool Radius and/or Distance to LFL
ExperimentSpill Size (m3)Spill Rate (m3/min)Pool Radius (m)Downwind Distance to LFL (m) (Max)
ESSO0,8 – 10,89 – 17,57 – 14400
U.S.C.G.3 – 5,51,2 – 6,6~7,5Not measured
Maplin Sands (dispersion tests)5 – 201,5 – 4~10190 ± 20 m
Maplin Sands (combustion tests)10,354,7~15Not measured
Avocet (LLNL)4,2 – 4,5246,82 – 7,22220
Burro (LLNL)24 – 3911,3 – 18,4~5420
Coyote (LLNL)8 – 2814 – 19Not reported310
Falcon (LLNL)20,6 – 66,48,7 – 30,3Not reported380

Pool fire experiments

Liquid Pool Spreading

Table 2. Largest Spill Volumes Tested to Date Giving Pool Radius and Max. Flux Rate
ExperimentVolume Spilled (m3)Pool Radius (m)Mass Flux (kg/m2s)
Boyle and Kneebone (Shell)0,02 – 0,85 Quiescent water surface (laboratory)1,97 – 3,630,024 – 0,195
Increased with amount spilled & amount of heavy hydrocarbons
Burgess et al.0,0055 – 0,36 (pond)0,75 – 6,060,181
Feldbauer et al. (ESSO)8 – 10,8 (Matagorda Bay)7 – 140,195
Maplin Sands5 – 20 (300 m dyke around inlet)~100,085
Koopman et al. (Avocet LLNL)4,2 – 4,52 (pond)6,82 – 7,22Not reported

Overall summary of results of experiments – pool fire & vapour cloud studies

LNG pool and vapour cloud fire experiments and their results are summarised in Table 3. A detailed description of these experiments is provided in the following sections.

Table 3. Large Scale LNG Fire Studies
StudySpill TerrainSpill Vol. (m2)Spill Rate (m3/min)Pool DIA. (m)Surface Emissive Power (kW/m2)Burn Rate (10-4m/s) or kg/m2sFlame Speed for Vapour Cloud Fires (m/s)
Pool FireVapor Cloud Fire
China Lake Tests
Water3 – 5,51,2 – 6,615 (max)220 ± 50220 ± 304 – 11 (measured)
(1,8 – 4,95)
8 – 17
(relative to cloud)
Maplin SandsWater5 – 203,2 – 5,630 (effective)203 (avg) (178 – 248 range)174 (avg) (137 – 225 range)2,1 (calculated) (0,945)5,2 – 6,0
CoyoteWater14,6 – 2813,5 – 7,1Not measuredNot measured150 – 340Not measured30 – 50 (near ignition sources – decayed rapidly with distance)
Maplin SandsLandNo reportNA20153 (avg)
219 (max)
NA2,37 (measured) (0,106)NA
MontoirLand238NA35290 – 320 (avg narrow angle)
257 – 273 (avg wide angle)
350 (max)
NA3,1 (measured) (0,14)NA
Open table in new window

Results of 40 m3 LNG spills on water – Document No. 2

  • Size of Spill – 25 → 40 m3;
  • outflow rate m3/minute – 10 → 20 m3;
  • varying wind speed;
  • 1 metre depth of water;
  • 50 m diameter pool;
  • visible plume length – 50 metres.

Not much input on specific values for heat flux, duration & distribution.

Consequence assessment methods for incidents involving releases from liquefied natural gas carriers – Document No. 4

  • Volume of spill – 12 500 m3;
  • duration – 33 mins;
  • heat Flux – 265 kW/m2;
  • flame Length (height) – (240 → 310 m);
  • small size spills only as practical input.

Summary results for pool fire calcs: Larger Pool size gives shorter fire duration and higher flame length.

Table 4. Summary of Results for Example Pool Fire Calculations
Hole Diameter3,3 ft (1 m)16 ft (5 m)
Initial Spill Rate11 700 lb/s (5 300 kg/s)290 000 lb/s (130 000 kg/s)
Total Spill Duration33 min1,3 min
Maximum Pool Radius240 ft (74 m)440 ft (130 m)
Total Fire Duration33 min6,9 min
Flame Length (Height)910 ft (280 m)1 400 ft (430 m)
Flame Tilt at Maximum Radius35 deg31 deg

Consequences of LNG marine incidents – Document No. 5

  • Quote: The effect of increased boil-off over water, will be to put more fuel into the same air space as over land without any mechanism for entraining more air.
    • This is likely to make the fire smokier and thus less luminous, with a greater fraction of the combustion energy going into heating the plume and less into thermal radiation.
  • Overall, large pool fires have several areas of uncertainty.
  • The large evaporating pool that is sustainable for dispersion is too thin to sustain combustion at the much higher rate of LNG consumption in a pool fire – LNG cannot flow from the source out to the periphery sufficiently quickly to replenish the material lost to combustion.
  • Another uncertainty is associated with the degree of additional smoke associated with pool fires over water. The smaller diameter pool and the greater smoke generation would tend to reduce the thermal hazard range. This consequence area could benefit from large scale trials on water.
Table 5. Discharge Results – Various Holes Sizes and Locations
Discharge CaseAbove Waterline ReleaseBelow Waterline Release
Hole Size (mm)250   750   1 500250   750   1 500
Initial Rate (kg/s)226   2030   8 130200   1 800   7 220
Duration (hr) †19   2,2   0,5430+†   3+†   0,8+†
Total Release (%)69 %100 %
† Durations are based on the average flowrate from the hole. For underwater cases this is an estimate only as the duration becomes complicated to estimate when the LNG driving force equalizes water back pressure.

Consequence modelling of LNG marine incidents – Document No. 6

  • Hole size: 250 mm → 5 000 mm;
  • discharge coefficient: 0,6 → 1,0;
  • Burn Rate Range: 0,089 → 0,353 Kg/m2s;
  • Pool Radius: 29 m → 253 m;
  • SEP: 178 → 265 Kw/m2.

Potential for BLEVE associated with marine LNG vessel fires – Document No. 7

  • A further protection is the limit on internal tank pressure to 0,28-0,30 barg. It will be shown later (in the paper) this is a major limitation on cargo flash and hence BLEVE potential;
  • this requires further investigation if it transpires that the safety valves are unable to accommodate the anticipated pressure rise.

Input to modelling work of other members of the working group

These papers did not provide any specific input to the work of the subgroup but were considered of possible use as input to the work of other members of the WG.

  • Model of Large Pool Fires – J. A. Fay – MIT (Paper 8);
  • Spread of Large LNG pools on the sea – J. A. Fay – MIT (Paper 10).

Large hydrocarbon fuel pool fires: physical characteristics and thermal emission variations with height – Document No. 9

  • Variation of SEP depending on height.
Statistics of Uncorrected NAR Data Apparent Surface Emissive Power (kW/m2)
Data fromMeanStd. Dev
Top of Fire44,844,0
Mid Height of Fire105,035,9
Bottom of Fire230,210,2
  • It was considered that further work might be required to ascertain the height of the flame, as clearly SEP impingement on the tank cover depends on height;
  • it was concluded that it might be necessary to commission work to further examine this.

Large hydrocarbon fuel pool fires: physical characteristics and thermal emission variations with height – Document No. 11

  • Importance of pool size on SEP.
Comparison of model predicted Assured parameter values r = 17,17 for CH4; β = 0,06; km = 130 m2/kg; Emax = 325 kW/m2; Ta = 293 Kx SEP with experimental date
Fire Diameter (m)Surface on which LNG BoilsFroude Number (Fc) (×10-3)Soot Mass Yield (y) Notarianni et al. [17] correlation for smoke yield.x (%)Soot Concentration (C1)(kg/m3)(×10-4)Fraction at Length of the «Clean Burning Zone» (Ψ)Soot Transmissivity (τ1) (×10-2)Mean SEP Over the Visible Fire Plume Height (Emax)Remarks
Current Model Tests Result (kW/m2)From Field kW/m2
15Water9,60612,73,3280,19666,4172185-224Raj, et al [8]. China Lake tests
20Land8,31913,03,4190,18057,12183140-180Mizner and Eyre [16]
35Land6,28813,73,5950,15035,7177175±30Malvos and Raj [14], Montoir tests
100Land3,72014,93,9260,0934,0113Potential size of future pool fire tests
300Water2,14816,24,2720,0330,002890Estimated pool size due to 1/2 tank content spill (12 500 m3 LNG) from an LNG Assumed optical depth at bottom of LNG fire = 13,8 m.x ship
Open table in new window

LNG properties & hazards – understanding LNG rapid phase transitions (RPT) – Document No. 12

  • Rapid phase transitions are more likely to occur in LNG mixtures containing very high fractions of ethane and propane. LNG composition is a critical parameter;
  • the hazard potential of rapid phase transitions can be severe, but is highly localised within the spill area.

LNG pool fire simulations – Document No. 14

  • The author had assumed that pressure valve settings were increased from 0,25 MARVS when at sea to 3 Bar;
  • the author had not accounted for the presence of inert gas surrounding the tank;
  • the author was not aware that pressure could be reduced by using gas boil-off in the propulsion and or steam dumping;
  • the author had not looked at the options for forced vaporising to reduce pressure;
  • the author was apparently not aware that Moss Tanks were design for emergency discharge under pressure;
  • no account of the heat reducing effect of water deluge systems, required by the IGC Code, had been accounted for;
  • the author had assumed that rate of the evaporation would be similar to the limited spill simulation Montoir Pool tests, where a small amount of LNG (pool) was ignited in a pit in the ground. We concluded that the heat influx would be less and so evaporation rate lower. However, in their calculation no consideration was given to the dispelling of oxygen due to a 600:1 ratio expansion and their modelling was of a fully mixed oxygen – methane cloud along one side of a vessel;
  • the author had assumed that a similar scenario to that of a Spanish LNG road tanker, that spilt LNG which then ignited. The tank was heated to the point of rupture and resulted in a BLEVE.

LNG decisions making approaches compared – Document No. 13

Proposed Baseline Cases for Models Used in LNG Marine Assessments.

Validation scale:

  • Use Burro 3, 7, 8, 9;
  • Coyote 5, 6;
  • and Maplin Sands 27, 29, 34, 35.

Run models for dispersion only.

Large Scale:

  • Use an amalgam of current cases reported by many current LNG workers.
  • Run these models for dispersion and pool fire results.
  1. Puncture case – leading to near instantaneous release of 500 m3 LNG.
  2. Maximum credible event case (accidental release) – 750 mm hole above waterline releasing all the cargo that can flow from the single largest tank.
  3. Maximum credible event case (terrorism) – 1 500 mm hole above waterline releasing all the cargo that can flow from the single largest tank.
  4. Maximum credible event case (jettison) – 10 000 m3/hr for 60 minutes.
  5. Worst case event (single tank) – 5 000 mm hole above waterline releasing all the cargo that can flow from the single largest tank.
MaterialPure liquid methane
WeatherCases should be run for D 5 m/s and F 2 m/s
Surface roughnessUse 0,3 mm and 10 mm
Relative Humidity70 %
Temperature20 °C air and water
LFLBase an latest data from AIChE DIPPR = 4,4 % vol


  • Source term – discharge rate duration and total amount, pool diameter and thickness (maximum and event average), boil-off rate (maximum and event average).
  • Dispersion – distance to LFL.
  • Fire – sustainable pool diameter for pool fire (maximum and event average), duration, and distance to thermal radiation predicted of 5 kW/m2.
  • ABS 2004a and Pitblado et al 2004 list example areas of LNG uncertainties that will probably require large scale experimental trials to resolve;
  • these will need to involve both the LNG vessel itself, its response to mechanical damages and to the physical consequences of large LNG spills onto water.

Thermal response of gas carriers to hydrocarbon fires – Document No. 15

  • The principal contribution of this report was considered to be as a verification control;
  • it closely follows some of the work objectives of the WG and, as such, may be compared against work carried out at the behest of the WG;
  • an unknown mentioned previously is the effect of SEP based on height and location to any fire. This is proposed in the paper as:
In Summarising the Results of Diagrams 1-9, We have for Condition with no Wind:
SurfaceFlux in kW/m2
Location on Deck
Fire SideMiddleSide Opposite Fire
Horizontal at deck level15040-15016-150
Vertical at deck level30060-15030-150
Horizontal at dome level35-150
Vertical at dome level120-200

Maplin sands experiments 1980 – dispersion results from continuous releases of refrigerated liquid propane & LNG – Document No. 18

  • Measurements in this paper were taken from sensors mounted at a range of heights;
  • wind speed and direction was identified in being critical as this affects the flame shape, direction, heat transfer, exposure of any adjacent structure etc.
  • «Further analysis of the heat and water content of the plumes is necessary before heat transfer relations may be applied with confidence».

Maplin sands experiments 1980 – dispersion & combustion behaviour of gas clouds resulting from large spillages of LNG & LPG on the sea – Document No. 19

  • A maximum scaled volume of 25 000 m3 was considered to be covered by tests using 20 m3;
  • The two most prominent features of the dense gas spill are the gravity spreading of the gas («Slumping») and the inhibition of vertical mixing by the density gradients formed («stratification»).
  • Table 6 – Combustion Trials at Maplin.
Table 6. Combustion Trials at Maplin
Trial NumberMaterial and Spill TypeVolume (m3/min or m3)Wind Speed Wind speeds given are those relevant during the combustion studies.x (m/s)Comments
17LNG Continuous2,88Flame failed
27LNG Continuous3,76Cloud fire
38LNG Continuous5,85Cloud fire
39LNG Continuous4,74Cloud and pool fire
22LNG Instantaneous125Cloud fire
23LNG Instantaneous8,55Flame failed
24LNG Instantaneous125Cloud fire
49Propane Continuous2,16Cloud fire
50Propane Continuous4,37Cloud and pool fire
51Propane Continuous5,67Cloud and pool fire
68Propane Instantaneous5-106Cloud fire
Table 7. Surface Emissive Powers
Surface Emissive Power (kW/m2)
MaterialCloud FirePool Fire
LNG173 ± 26203 ± 31
Refrigerated liquid propane173 ± 2043 ± 9


Results of documentation review

The figures below should be considered as representative ballpark figures that were derived based on a summation of the overall review of available documentation.

Read also: Fire protection and Fire extinction on Liquefied Gas Carriers

There are undoubtedly contradictory values appearing in the work submitted for review, however those below were felt to provide a representation of the values proposed in the different works.

  • Heat Flux:
    • worst case direct flame impingement – 325 kW/m2 (at source of flame hottest part);
    • this value is maximum at base of flame & reduces with height;
    • applies to pool fire (not cloud fire).
  • pool fire sizing v SEP;
Small Pool Fire35 m220
Large Pool Fire300 m90
  • Fire duration:
    • release of 12 500 m3 = 8,1 minutes;
    • for 330 m (cloud fire) pool diameter ≈ 90 kW/m2.
  • Fire height:
    • not currently possible to define – not addressed by work to date;
    • pool size 35 → 300 m diameter.

Note: 35 m is considered to be unrealistically low due to probable structural failure.

  • impingement:
    • pool fire – worst case – side;
    • flame impingement.
  • deck:
    • cloud fire – 170 kW/m2;
  • gives a lower flame height;
  • cloud is advised to be max 10 m in height:
    • BLEVE – Depends on a number of factors;
  • further consideration of this should be undertaken after modelling.


The result of the document review was that considerable uncertainty exists over the height of any LNG fire resulting from an accidental or deliberate release on water.

The height of the fire has a direct impact on the fire impingement and heat flux to which an Special Requirements for LNG and LPG gas carriersLNG carrier would be subjected. The conclusion is that further definition of the flame height may only be achieved with appropriate modelling, for which the group would be in a position to provide a range of values for the different inputs required. It is only from such work or, alternatively, from full scale tests of larger LNG pool fires on water, that the consequential heat fluxes could then be derived.

The latter full scale tests would also perhaps confirm the shift in opinions over Surface Emissive Power, where current expert opinion is now in favour of incomplete combustion in an LNG pool fire on water, reducing the SEP as the size of any pool fire grows.

Further work by the subgroup is, therefore, pending completion of both modelling of an LNG pool fire on water (CFD and/or solid flame) for a range of values, and output from full scale testing of a large LNG pool fire on water. The latter is considered not to be the remit of this group but nonetheless is an important input for a conclusive output to the work of the WG.


Did you find mistake? Highlight and press CTRL+Enter

Апрель, 03, 2024 78 0
Add a comment

Text copied