verification and validation of consequence and risk modelsukelg.ps.ic.ac.uk/40hw.pdf · dispersion...

Verification and validation of consequence and risk models

1

Verification and validation of consequence models

UKELG, 19-20 September 2007Cardiff, Wales

Henk W.M. Witlox, DNV Software

© Det Norske Veritas AS. All rights reserved Slide 226 October 2007

Verification and validation of consequence models

� Introduction

� Discharge

� Dispersion

� Flammable effects


2


Continuous release with rainout

jet fire

pool fire

explosion or flash fire

droplet trajectory

flashing two-phasedischarge from vessel vapour-plume

centre-line

pool

point ofrainout

SUBSTRATE


Quality procedure for model development

� Define model- Literature review- Formulate physical and mathematical model- Solution method and algorithm

� Design (preliminary, detailed) and coding

� Model testing

� Documentation

� Review

� Model integration into overall product


3


Model testing

� Verification - code solves correctly mathematical model- Against analytical solution- Against parallel ‘verification’ Excel spreadsheet- Against other model

� Validation against experimental data – justify model assumptions- Small-scale experiments (isolating phenomenon)

- Large-scale experiments

� Sensitivity analysis – overall robustness and effect parameters- Base case- Parameter variations (single or multiple parameters)


Key references for consequence and risk modelling

� Dutch Yellow Book (1997)

� Loss Prevention Process Industries - Lees (1996)

- Updated Mannan (2005)

� Perry Chemicals Engineering Handbook (1999)

� CCPS guidelines- Dispersion (1996)- Flammable effects (1994)

- QRA (2000)


4


Discharge model

pipe

flow

expansion zone

leak orifice

atmosphere

vessel (stagnation)


Discharge model

� Range of scenarios- Leak from vessel, short or long pipes, instantaneous, relief systems, …

- Sub-cooled liquid, flashing liquid, or gas release

� Data - Flow rate, velocity and liquid fraction (both orifice and post-expansion data)- Droplet size

� Literature survey- Numerous discharge models

- No up-to-date overview of experiments (benchmark tests)- No published established systematic model evaluation

� Literature review to establish experimental dataset

� Application to Phast discharge models


5


Discharge model verification and validation

� Verification- Analytical flow-rate equations

- incompressible liquid (Bernoulli)

- ideal gas (choked and un-choked)- Process simulators- Worked-out examples in literature (e.g. CCPS publications)

� Validation- Subcooled and saturated water jets

- Sozzi and Sutherland (varying pipe length)

- Uchida and Narai (varying pipe length and stagnation pressure)

- Many other experiments- Hydrocarbon releases

- Full-bore and orifice releases of liquid propane (Shell)

- Orifice releases of butane (Shell)- Long pipe - validation against Isle of Grain (full-bore and partial leaks – LPG)


Discharge model validation for sub-cooled water release

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

0 0.5 1 1.5 2 2.5

Pipe length (m)

Rat

io o

f p

red

icte

d t

o o

bse

rved

flo

wra

te (

-)

PHAST6.53-Sozzi&Sutherland

PHAST6.53-Uchida&Narai


6


Validation against P66: mass expelled against time (45.3% breach)

P66 - Inventory

0

20

40

60

80

100

120

0 10 20 30 40 50 60

t(s)

Mas

s (k

g)


Dispersion (example PHAST Unified Dispersion Model UDM - continuous dispersion)

z

x

TOUCHINGDOWN

ELEVATEDPLUMEy

GROUND-LEVELPLUME

circularcross-section

truncated cross-sectionsemi-elliptic cross-section

� Elevation phases: elevated, touching down, ground-level, lift-off, elevated, mixing layer

� Dispersion phases: jet, heavy, passive

WIND

jet release


7


Dispersion module verification and validation

� Near-field elevated/jet dispersion- Verification: analytical solution (horizontal jet), vertical jet correlations

� Heavy dispersion- Verification: analytical solution (2D), HEGADAS (3D)- Validation: McQuaid (2D isothermal), HTAG (3D isothermal)

� Passive dispersion- Verification: analytical solution, TNO Gaussian concentration profile

� Finite duration- Verification: SLAB/HGSYSTEM [finite-duration correction model – FDC]- Validation: Kit Fox [quasi-instaneneous model (QI) and FDC]


Dispersion module verification and validation (continued)

� Thermodynamics- Verification: analytical, HEGADAS (pure component, mixture, HF)

- Validation: Schotte experiment (HF)

� Pool spreading/evaporation- Verification: GASP- Validation: spills on water/land, wide range of chemicals (LNG, propane, …)

- spreading (non-volatile chemicals)- evaporation (confined pools)- simultaneous spreading and evaporation


8


Dispersion – Validation against large scale experiments

� Continuous passive dispersion- Prairie Grass

� Continuous elevated two-phase jet- Ammonia (Desert Tortoise and FLADIS)- Propane (EEC)- HF (Goldfish)

� Continuous dispersion from pool - LNG (Maplin Sands, Burro, Coyoto)- LPG (Maplin Sands)

� Un-pressurised instantaneous- Freon-12 (Thorney Island)

� Continuous and finite-duration dispersion from area source- CO2 (Kit Fox)


Dispersion – Validation for Kit Fox experiment(20 second release of CO2, experiment KF0706)

0.01

0.1

1

10

100

1 10 100 1000

Downwind distance (m)

Cen

trel

ine

con

cen

trat

ion

(m

ol%

)

Phast 6.53 (FDC)

Experimental


9


Fireballs, jet fires and pool fires� Mathematical model

- Empirical correlations for fire geometry and surface emissive power- Fireball (sphere)- Pool fire (tilted cylinder)- Jet fire (cone)

- Radiation at given location by means of integration along fire surface

� Verification- Simple hand calculations and/or spreadsheet- Comparison against other models

� Validation- Pool fire:

- LNG (Montoir, Johnson)- Hexane (Lois and Swithenbank)

- Jet fire: - Vertical natural gas (Chamberlain)- Horizontal natural gas (Johnson, Bennett et al.)- Two-phase LPG (Bennett et al.)- Horizontal liquid crude oil (Selby and Burgan)


Pool fire model (Phast model POLF)

H

θ

D

WIND


10


Validation against Johnson LNG pool fire experiments

0

1

2

3

4

5

6

7

8

0 1 2 3 4 5 6 7 8

Measured heat flux (kW/m2)

Pre

dic

ted

hea

t fl

ux

(kW

/m2)

0% deviation

-15% deviation

15% deviation

-40% deviation

40% deviation

Johnson (1992)

PHAST 6.53


Jet fire - Chamberlain pure-vapour model (no crosswind)

Z

X

W2

θj

W1

b

RL

Lb

α

� cone

� tilt by wind

� lift-off distance

� widths, length

WIND


11


Validation against Johnson LNG jet fire experiments

0

5

10

15

20

25

30

35

0 5 10 15 20 25

Measured heat flux (kW/m2)

Pre

dic

ted

hea

t fl

ux

(kW

/m2)

0% deviation

-15% deviation

15% deviation

-40% deviation

40% deviation

Johnson (1994)-Simulated Data

PHAST 6.53


Explosion modelling

� Comparative study by Fitzgerald - Key models:

- TNO Multi energy (MULT)- Baker Strehlow (BSEX)

- Shell Congestion Assessment Model (CAM)- Validation:

- LNG, LPG (EMERGE - TNO)

- LNG (BFETS -SCI)


12


Explosion - Validation against EMERGE 6 experiment(LPG, medium scale)

0.00E+00

1.00E+04

2.00E+04

3.00E+04

4.00E+04

5.00E+04

6.00E+04

0.00E+00 1.00E+01 2.00E+01 3.00E+01 4.00E+01 5.00E+01

Distance from edge (m)

Ove

rpre

ssu

re (

Pa)

Test Average

Test Maximum

BSEX Results

MULT Results

verification and validation of consequence and risk modelsukelg.ps.ic.ac.uk/40hw.pdf · dispersion...

Documents