Geoff Chamberlain, Shell Global Solutions, UK Martin …members.igu.org/html/wgc2006pres/data/wgcppt/pdf/PGC Programme... · ... Shell Global Solutions International, Netherlands

Management of Large LNG Hazards

Geoff Chamberlain, Shell Global Solutions, UK

Martin de Groot, Shell Global Solutions International, Netherlands

CONFIDENTIAL

Outline

• HEMP

• Scenarios• Pool spread and evaporation models

• Rapid Phase Transitions• Pool fires

• Vapour Dispersion• Vapour Cloud Explosions

• Conclusions

CONFIDENTIAL

Hazard and Effect Management Process - HEMP

Activity under assessment

Identify (potential) hazardous events

Analyse Risk from Hazardous eventslikelihood per event / of exposure / consequences

Determine Overall Risk to People (ALARP), Assets (Cost) and Environment (Impact)

Risk Reduction -- option selection / change decisions / emergency response

Feed BackFeed Back

Major Hazard Scenarios. LNG.

Release

More obstaclesGreater confinementFlame acceleration

Ignites?

Jet fireInternal

explosionPool fire Cloud fire Fast flame

Safedispersion

Dispersing cloud

Ignites?

No

Yes No

Yes

R.P.T

Pool Spread and Evaporation

• The source term for fire and explosion.

• Scope for improved physics and prediction.• On land, substrate important after initial film-

boiling phase. Porous substrates (sand or gravel) show greatly enhanced evaporation rates.

• On water, waves/currents can modify the evaporation rate, but not known exactly how. Some ice or hydrate may form in shallow water, eg Maplin Sands.

Model Demonstration of LNG pool spreading with wall

Spill of 0.1 m3/s LNG for 2s

Model Demonstration of LNG pool spreading with wall and tank

Spill of 0.1 m3/s LNG for 2s

Rapid Phase Transitions RPTs

• Can occur, particularly if LNG has “aged”, and did occur at Maplin Sands.

• Spreading and evaporation become very different compared with gently spreading pools.

• RPTs create blast.• Calculations suggest deformed hull but no

rupture. • E.g.150m3 spill in 6m between FPSO and

LNG carrier, blast energy on hull 60kJ/m3.

Pool Fires

• 35m LNG pool fire.

• Montoir1987

Model flame SEP versus pool diameter & fuel

020406080

100120140160180

0 20 40 60 80 100

Pool diameter, m

LNG

LPG, propane

Butane, gasoline, kerosene

Surface Emissive Power (SEP) in kW / m2

diameter, m

Not validated

LNG Pool Fires

• Thermal hazard depends on size and brightness.• Average brightness [surface emissive power] peaks at

about 180 kW/m2

• Larger LNG pool fires expected to have lower SEP. The limit is 20kW/m2 due to smoke.

• Flame height depends on upward momentum and buoyancy. Froude number scaling.

• Scaling predicts large LNG pools, say100m diameter, burn as “intermittant” fires. i.e. no change in burning regime. Does not become a “mass” fire (forest type fire).

• Steady burning rate on water around 0.22 kg/m2/s.

LNG Vapour Dispersion

• Many models but many uncertainties remain for large spills.

• Uncertainties – roughness, strong temperature gradients over sea, humidity, heat transfer.

• Largest tests 100kg/s continuous, 20m3

instantaneous.• Releases over the sea. Unlikely that the flammable

cloud becomes buoyant. LFL in visible cloud.• Releases below sea. Plume becomes buoyant before

LFL. • CFD modelling and Random Walk modelling (DICE) –

ones to watch.

DICE (Dispersion In Congested Environments)

• Model dispersion using particle based random walk model

• Particle velocity =background flow velocity +semi-random turbulent component +velocity due to jet release momentum

• Calculate trajectories of 1000s of particles and determine gas concentration

LNG vapour cloud explosion - Schelkin feedback loop

Gas burns

LargerVolume

Turbulence

IncreasedGas Flow

BLAST

Gas expands

Larger volume pushesunburnt gas ahead

Unburnt gas pushedaround obstacles

Flame front wrinkled,burning surface greater,

increased mixing, faster burning

CONFINEMENT?Yes Expansion resisted,

pressure builds upNo Less pressure build-up

CONGESTION?High More turbulence

More pressure build-upLow Less turbulence

Less pressure build-up

The size of the flammable cloud or the congestion determines the explosion severity, whichever is the

smaller.

Overpressure and impulse predictionCAM, SCOPE, EXSIM

Shell’s Explosion models

Probabilistic approach - ‘Exceedance’

• Method for determining the probability (frequency) of exceeding a certain overpressure (and impulse) at a given location.

Pro

babi

lity

of e

xcee

ding

pr

essu

re

Pressure

typical “worst case”

Sustainable pressure, e.g. dictated by structural strength

Flammable gas mapping and explosion exceedance applied to a Floating LNG concept Flammable Gas FrequencyFlammable Gas Frequency

Explosion Overpressure FrequencyExplosion Overpressure Frequency

Explosion Overpressure Exceedance over the deck of a Floating LNG ship.

0

1000

2000

3000

4000

5000

Life

boat

TR-

rear

TR-

front

20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 410 420 Turret

Ove

rpre

ssu

re, m

bar

Frequency = 1.0E-03Frequency = 1.0E-04Frequency = 1.0E-05

Experimental study of the critical gap to limit explosion severity between process units

Example of explosion exceedance for safe placing of trailers on a plant

e.g. Control Room

Conclusions

• HEMP studies can account for major LNG hazards.• A hierarchy of modelling tools can be brought to bear

with increasing accuracy as the safety criticality increases.

• But extrapolation to large scale places large demands on the physics and experimental validation is preferably required.

• A few areas for further study - Pool spreading and evaporation, effect of substrates, the burning regime for large pool fires, vapour cloud explosion sub-grid modelling.