SYMPOSIUM SERIES No.

147 © IChemE

THERMAL RADIATION FROM FIREBALLS ON FAILURE OF

LIQUEFIED PETROLEUM GAS STORAGE VESSELS
T Roberts*, A Gosse** and S Hawksworth*
* Health and Safety Laboratory, Health and Safety Executive, Buxton, Derbyshire SK17
9JN
** BG Technology, Ashby Road, Loughborough, Leicestershire, LE11 3GR
© Crown Copyright 2000. Reproduced with the permission of the Controller of Her Majesty’s Stationery Office.

Fire impingement on vessels containing pressure liquefied gases can result in

catastrophic failure of the vessel resulting in a Boiling Liquid Expanding Vapour
Explosion (BLEVE). If the gas is flammable, this can result in the formation of very
large fireballs. In safety assessments where catastrophic vessel failure is identified
as a real possibility, the risk of death from a fireball tends to be higher than that from
missiles or blast.

Since many of the physical processes which take place in a BLEVE are scale
dependent, a series of tests were undertaken at a large scale where 2 tonne propane
vessels were taken to failure in a jet fire and the vessel response, mode of failure and
consequences of failure characterised. The measurements taken by the Health and
Safety Laboratory and BG Technology relating to fireball formation are described.

Keywords: fireball, propane, LPG, BLEVE, thermal radiation, jet fire.

INTRODUCTION

Within the petrochemical industries, many flammable gases are stored as liquids under
pressure. To ensure safe working practices, safety assessments are undertaken to consider the
hazards posed by accidental releases from such storage vessels in order to assess the risk
presented to the installation, personnel and the local population. Hence, a wide range of
release scenarios needs to be considered, including those of low probability but with high
consequential hazard, such as catastrophic failure. One such scenario, which may give rise to
failure, is jet-flame impingement onto the storage vessel, perhaps resulting from an ignited
release from a leaking flange or failed pipework. Such fires can result in catastrophic
consequences e.g. the Mexico City disaster1.

Fire impingement on a vessel containing pressure liquefied gas causes the pressure to rise
within the vessel and the vessel wall to weaken and may ultimately lead to catastrophic failure
and total loss of inventory. If such an event occurs, the liquefied gas is released to the
atmosphere, where it boils and flashes back to its gaseous state because of the sudden
reduction in pressure. The gaseous material expands to occupy a much greater volume than
the liquid material and a pressure wave is generated. These types of events are known as
Boiling Liquid Expanding Vapour Explosions (BLEVEs). If the released material is
flammable, ignition is likely to occur so that in addition to missile and blast hazards, there is
also a thermal radiation hazard from the fireball produced. It is the thermal radiation hazard
which tends to dominate the near-field risk assessment2.

Since many of the physical processes taking place in events such as BLEVEs are scale
dependent, in order to gain an understanding of the hazard posed, it is necessary to conduct

1
SYMPOSIUM SERIES No. 147 © IChemE

tests at a large scale. In 1989, a series of tests3 were conducted as part of the Commission of
the European Community (CEC) co-funded research initiative on Major Technological
hazards. Seven tests were conducted to study the BLEVE failure of vessels containing up to 2
tonnes of either propane or butane. The effect of the different fuels, the vessel size, the vessel
fill ratio and the pressure at failure were investigated. During these tests, the vessel contents
were heated using an internal electric heater and failure of the vessels was initiated artificially
using an explosive charge. The resulting missile throw, overpressures and fireball
characteristics were investigated.

Following on from these experiments, also as part of the CEC Science and Technology
for Environmental Protection (STEP) programme, a co-funded project was set up to
investigate the hazard consequences of Jet-fire Interaction with VEssels containing
pressurised liquids (JIVE). The Health and Safety Laboratory (HSL) was contracted by the
CEC and HSE’s Technology Division to investigate the thermal response of propane vessels
when subjected to jet-fire attack and to assess the effectiveness of mitigation techniques. As
part of this work, a series of four experiments were undertaken to look at the response of
vessels up to the point of BLEVE failure and to gather some understanding of the mode of
failure when a vessel is exposed to a real jet-fire impingement situation. BG Technology
(BGT) were invited to make measurements of thermal radiation characteristics of any
resulting fireball produced during these four tests. This paper describes the measurements
made by HSL and BG Technology relating to the fireballs formed on catastrophic failure of
the vessels. A brief comparison of the results with the normal empirical models used is also
made.

EXPERIMENTAL ARRANGEMENT

JET FIRE

The jet-fire scenario considered was liquid discharge through a hole in an adjacent punctured
vessel or damaged pipework. The jet-fire size and location was chosen so that the 4.5 m
long by 1.2 m diameter target vessels were at least three quarters engulfed in fire and the
effects of wind were minimised. The jet fire consisted of ignited, flashing, liquid propane at a
flow rate of about 1.8 kg s-1 from a nozzle equivalent to a 12.7 mm diameter hole. The two
tonne target vessels were placed at approximately the still-air lift-off position of the flames
i.e. 4.5 m. The mean heat flux density around the vessels was 179 kW m-2.

PROPANE VESSELS

Each two tonne vessel was fitted with a pressure relief valve (PRV), protected by thermal
insulation during the trials, set to relieve at 17.24 barg. The vessels were instrumented with
thermocouples in the liquid and vapour space and protected from direct flame impingement
by 3 mm plates on the outside of the shell. Pressure transducers were fitted to take-offs from
the liquid and vapour space. The target vessels were mounted on a frame resting on load cells
so that the fuel mass could be monitored.

2
SYMPOSIUM SERIES No. 147 © IChemE

THERMAL RADIATION

Instrumentation was deployed by BG Technology to measure the thermal radiation

characteristics of any resulting fireball. Medtherm wide angle, slow and fast response
radiometers, Land slow response radiometers and International Research and Development
wide angle, fast response radiometers (WIRD) were set up along radial lines (see Figure 1)
around the vessel to measure the incident thermal radiation at different distances from the
fireball. An Agema 900 Infrared Thermal Imaging System (AGEMA) was used to record the
surface emissive power distribution of the fireball throughout its duration. An International
Research and Development narrow angle, fast response radiometer (NIRD), with a total field
of view of 1°, was also deployed alongside the AGEMA. HSL used a LAND CYCLOPS
Ti35sm thermal imaging camera, at approximately 90o from the BGT camera, to measure the
thermal image of the fireballs. The HSL thermal imaging camera viewed crosswind and the
BGT camera viewed up or downwind. A sonic anemometer was deployed to measure the
wind speed and direction.

FAILURE MODE TRIALS

INITIAL CONDITIONS

Four unprotected vessels, containing different quantities of propane (20%, 41%, 60% and
85% of the water capacity), were engulfed in a jet fire until they failed. The ambient
conditions, degree of fill and mass flow rate of the propane jet used are summarised in Table
1. Normally the vessels were unrestrained but, because rocketing occurred in the 60% full
vessel trial, the 85% full vessel was restrained.

Table 1. CONDITIONS FOR FAILURE MODE TRIALS

Degree of fill (%) 20 41 60 85

Propane mass (kg) 455 929 1,364 1,932

Propane jet mass flow rate (kg.s-1) 1.5 1.4 1.6 1.7
Ambient temperature (oC) 19 20 17 18
Wind direction relative to direction of jet (deg.) 180 180 0 180
Approximate wind speed (m s-1) 4 3 5 2.5
Relative humidity (%) 80 60 95 90

GENERAL OBSERVATIONS

In each trial, propane was immediately ignited by a pilot light when ejected from the nozzle.
It burnt with a bright yellow flame that became slightly darker as the proportion of liquid
ejected increased. This gave a flame that almost enveloped the target vessel but, occasionally,
the left side (in relation to the jet) of the vessel could be seen indicating that the flames were
slightly skewed to the right. After 1 to 2 minutes, the PRV opened releasing gas that ignited
to give a jet of flame. All the vessels failed catastrophically within about 3 minutes of the
PRV opening giving a large fireball. On failure, three vessels split longitudinally and opened

3
SYMPOSIUM SERIES No. 147 © IChemE

out flat and the other (60%), after initially splitting longitudinally, split circumferentially and
rocketed.

CONDITIONS AT FAILURE

All the vessels failed within 5 minutes of commencing jet-fire impingement and at pressures
ranging from 16.5 to 24.4 barg. In all cases except for the 20% full vessel, after the initial
pressure drop on PRV opening, the pressure increased until, at failure, the pressure was higher
than the respective initial PRV opening pressure. For the 20% full vessel, the pressure fell
from 18.6 barg to 16.5 barg.

There was considerable variation in the vessel wall temperatures depending on whether
the wall was in contact with liquid propane or not. In every trial, the wall temperature just
above the liquid level was much higher than that just below the liquid level suggesting that
there was relatively little level swell with consequent cooling of the wetted wall. The
temperatures at the back were lower than those at the front except for the 60% full vessel trial,
which was the only trial in which the wind was in the same direction as the jet. In all cases,
the wall in contact with the vapour space reached the highest temperatures. The pressures and
highest shell temperatures at the time of failure are summarised in Table 2.

Table 2. PRESSURE AND HIGHEST SHELL TEMPERATURE AT FAILURE

Degree of fill Failure time Pressure Highest shell temperature

(%) (s) (bar g) (Celsius)
20 250 16.5 870
41 286 21.3 704
60 217 18.6 821
85 254 24.4 848

PARAMETERS AFFECTING FIREBALL FORMATION

When the target vessels failed, the contents were released very rapidly. As the temperature of
liquefied propane was at least 100 oC higher than its boiling point (- 42 oC) at atmospheric
pressure, it rapidly boiled with a considerable proportion of the liquid immediately flashing
to vapour on loss of containment, giving a BLEVE. In every case, the vapour clouds ignited
almost immediately giving large fireballs. The parameters affecting fireball formation are
summarised in Table 3.

4
SYMPOSIUM SERIES No. 147 © IChemE

Table 3. PARAMETERS AFFECTING FIREBALL FORMATION

Fill Propane Wind Wind Tank failure mode

percentage released speed direction
(%) (kg) (m s-1)
20 279 3 Westerly Opened out flat
41 710 4 Westerly Opened out flat
60 1,272 5 South-easterly Split circumferentially
85 1,708 2 Westerly Opened out flat

THERMAL RADIATION MEASUREMENTS

The fireball from each trial is shown at its maximum size in Figure 2. Table 4 shows a
summary of the incident thermal radiation results (Medtherm radiometers along lines A and
C) from the 85% full vessel trial and Figure 3 shows the thermal images of fireball growth
during this trial. The variation of incident thermal radiation received at 100 m is illustrated in
Figure 4 and the dosage against distance in Figure 5. In general, there was good agreement
between the results from the narrow angle radiometer and the AGEMA, with the AGEMA
giving values within 10% of those of the NIRD. Because of cloud drift, the crosswind (line
C) results are higher than the upwind (line A) results.

Table 4. INCIDENT THERMAL RADIATION AND DOSAGE DATA (85% TRIAL)

Line A (upwind) Line C (crosswind

Distance Maximum Dosage Maximum Dosage
radiation radiation
(m) (kW m-2) (kJ m-2) (kW m-2) (kJ m-2)
50 36.72 132.00 47.78 169.12
75 20.03 81.08 28.13 114.75
100 14.03 55.47 18.12 73.73
125 9.70 37.65 11.96 49.98
150 8.03 28.68 9.51 36.99
200 4.38 16.56 5.05 21.27

The fireball dimensions and surface emissive power data, measured at maximum
projected area, are summarised in Table 5. There is some uncertainty in the durations given as
the fireball breaks up into small flame eddies just before extinction and there are different
perceptions of extinction. The flame heights and widths are calculated from the number of
pixels, the pixel size, the distance of the camera from the fireball and the field of view of the
camera lens. The error in the surface emissive powers is estimated to be ± 15% and errors in
dimensions to be ± 5%.

5
SYMPOSIUM SERIES No. 147 © IChemE

Table 5. FIREBALL DATA AT MAXIMUM PROJECTED AREA

Trial 20% 41% 60% 85%

Propane released (kg) 279 710 1,272 1,708
Cross or up/downwind View
Duration (s) CW 3 5 6.5 7
U/D 3.8 4.6 5.9 6.6
Time to maximum CW 1.56 <2 2.6 3.12
power (s) U/D 1.38 2.07 3.04 2.96
Time to maximum CW 2.08 2 3.12 3.64
projected area (s) U/D 2.21 2.84 4.18 3.69
Area at maximum CWa 1221 < 2040 2907 4244
power (m2) U/Db 837 1406 3176 3267
Maximum projected CWa 1300 2040 3150 4600
area (m2) U/Db 960 1460 3840 3520
Height (m) CW 43 67 72 105
(at maximum area) U/D 42 59 98 103
Width (m) CW 45 41 65 85
(at maximum area) U/D 41 43 74c 71
Average SEP CW 403 > 195d 314 312
(kW m-2) U/D 294 242 245 285
(at maximum power)
Average SEP CW 295 > 195d 287 312
(kW m-2) U/D 188 196 117 212
(at maximum area)
Maximum average CW 413 (1110 > 195d 320 (2670 m2) 360 (3210 m2)
2)
SEP (kW m-2) U/D m 278 (1060 m ) 270 (2470 m2) 314 (1860 m2)
2

333 (650 m2)

e
Maximum SEP CW 650 > 195 482 556
(kW m-2) U/D
f 554 484 486 523
Maximum power CW 492 > 398 913 1451
(MW) U/D 246 340 778 931
a o -2
670 C (cf. 45 kW m ) fireball contour used to define area
b
40 kW m-2 fireball contour used to define area
c
Estimated width as flame outside AGEMA field of view
d
Instrument not set for maximum range
e
Point maximum
f
10% maximum

The data show that the maximum average surface emissive power occurs before the
fireball reaches its maximum size. The variation of surface emissive power and projected
area with time is illustrated for the 85% full vessel trial in Figures 6 and 7 respectively. In
general, the maximum surface emissive powers measured crosswind were higher than those
measured up/downwind. The average up/downwind surface emissive power, at maximum
power, was in the range 242 to 294 kW m-2 and, at the maximum projected area, was 117 to
212 kW m-2. The corresponding crosswind ranges were 312 to 403 kW m-2 and 295 to
312 kW m-2, respectively. Both sets of results indicated that the highest average surface
power was from the release of propane from the vessel with the lowest (20%) degree of fill
and that failed at the lowest pressure.

6
SYMPOSIUM SERIES No. 147 © IChemE

DISCUSSION

The trials have confirmed that, when a BLEVE of a vessel containing flammable pressure
liquefied gas occurs, the main effects are thermal radiation, fragmentation and blast. In terms
of potential to kill, the thermal radiation from the large fireball formed is by far the most
dominant hazard for personnel near to the vessel.

When the vessels fail, the rapid vaporisation and expansion results in a cloud of vapour
and small droplets, with the concentration of vapour within the cloud being above the upper
flammable limit. After ignition, turbulent burning occurs mainly from the outside inward.
Burning of the vapours and droplets increases the buoyancy within the burning cloud and
hence increases its tendency to rise. The volume of the fireball also increases. The
turbulence maintains rapid mixing within the cloud and hence an increased rate of burning
with high flame temperatures, little soot formation and the emission of higher levels of
thermal radiation. The amount of thermal radiation emitted from the fireball depends, as a
function of time, on the surface area and surface emissive power and, for a given size of
fireball and surface emissive power, the radiation dosage received by a target will depend
upon the distance from the source, atmospheric transmissivity and the fireball duration.

There have been numerous papers on evaluation of fireball hazards, some deriving
empirical relationships, some theoretical relationships and some a combination of the two. In
general, for the purposes of risk assessment, empirical relationships are used. Comparison of
data obtained with published empirical models is restricted to HSL work2&4 and to recent
comprehensive reviews, with guidelines, published5&6 by the Center for Chemical Process
Safety (CCPS) of the American Institute of Chemical Engineers.

FIREBALL DURATION

Both Roberts4 and the CCPS6 suggest using the expression:

1/3
t = 0.45 . M (M < 30,000 kg) (1)

to calculate the time, t (s), when visible radiation from the fireball ceases from the released
mass of fuel, M (kg). However, Prugh7 suggests that the expression in the CCPS Guidelines
for Chemical Process Quantitative Risk Analysis5 be used viz.
0.26
t = 0.825 . M (2)

The results are compared in Table 6. The expression suggested by Prugh7 gives a slightly
better fit to the data although both expressions appear to underestimate the duration of the
largest fireball.

7
SYMPOSIUM SERIES No. 147 © IChemE

Table 6. MEASURED AND PREDICTED FIREBALL DURATIONS

Propane Predicted duration Measured duration

released 1/3 0.26 Crosswind Up/Downwind
0.45 . M 0.825 . M
(kg) (s) (s) (s) (s)
279 2.9 3.6 3 3.8
710 4 4.5 5 4.6
1,272 4.9 5.3 6.5 5.9
1,708 5.4 5.7 7 6.6

FIREBALL DIAMETER

Although the shape of the fireball formed on vessel failure depends on the failure mode,
fireballs are usually considered to be spherical in shape and for which a representative
diameter can be calculated. Both Roberts4 and the CCPS6 suggest using the expression:
1/3
D= 5.8 . M (3)

to calculate the maximum fireball diameter, D (m), from the mass of fuel, M (kg), released.
Again Prugh7 suggests that the expression in the CCPS Guidelines5 be used viz.
0.325
D= 6.48 . M (4)

The results are compared in Table 7. Again the marginally higher values given by the
expression suggested by Prugh gives a slightly better fit.

Table 7. MEASURED AND PREDICTED FIREBALL DIAMETERS

Propane Predicted diameter Measured diameter

released 1/3 0.325 Crosswind Up/Downwind
5.8 . M 6.48 . M
(kg) (m) (m) (m) (m)
279 38 40 45 41
710 52 55 45 43
1,272 63 66 75 74
1,708 69 73 85 71

FIREBALL LIFT-OFF TIME AND ELEVATION

In general, hazard calculations assume that fireballs are spherical and touch the ground. In
practice, the fireballs start to lift off when buoyancy and entrainment are dominant and drift
with wind. Both Roberts4 and the CCPS6 recommend using the Hardee and Lee8 expression
for lift-off time, tlo (s)

tlo = 1.1 . M1/6 (5)

8
SYMPOSIUM SERIES No. 147 © IChemE

Prugh suggests using the CCPS5 expression for the height, H (m), of the centre of the fireball
above the ground at the time of maximum diameter, D (m):

H= 0.75 . D (6)

The results, with D as predicted by Equation 4, are compared in Table 8.

Table 8. PREDICTED AND MEASURED LIFT-OFF TIMES AND ELEVATIONS

Propane Lift-off time Elevation

released 1/6 Measured 0.75 . D Measured
1.1 . M
(kg) (s) (s) (m) (m)
279 2.8 2.2 30 22
710 3.3 3 41 33
1,272 3.6 3.4 50 31.5
1,708 3.8 3.7 55 49

SURFACE EMISSIVE POWER AND FIREBALL RADIATION

The radiation received by a target is usually calculated using either a point-source model or a
solid-flame model.

Point source model: In the point-source model, it is assumed that a certain fraction (F) of the
heat of combustion is radiated in all directions. Prugh suggests using the CCPS5 relationship
between the surface emissive power, SEP (kW.m-2), heat of combustion, Hc (J.g-1) and the
fuel mass, M (kg) :
2
SEP = F . M . Hc / [ π. D .t] (7)

If the expressions (Equations 2 and 4) for D and t are substituted and the heat of combustion
of propane is taken as 46,000 J g -1 then this expression reduces to:
0.09
SEP = 423.2 . F . M (8)

Prugh gives an example where F is taken as 0.25 for propane. Roberts suggests that the
radiation fraction is related to the pressure, P (MPa), by the expression:
0.32
F = 0.27 . P (P < 6 MPa) (9)

F is usually taken to be in the range 0.25 to 0.40. Roberts' expression is not derived from the
same data set as Prugh and, strictly speaking, should only be used with corresponding
expression derived from the original data. The calculated values, using Prugh's expression,
are compared with the average values at maximum projected area in Table 9.

9
SYMPOSIUM SERIES No. 147 © IChemE

Table 9. PREDICTED AND MEASURED SURFACE EMISSIVE POWERS

Amount Pressure Fraction Calculated SEP = Average SEP at

of at radiated 0.09 maximum area
423.2 . F. M
propane failure F = 0.27 . P0.32 F = 0.25 F = 0.40 Crosswind U/D wind
released -2 -2 -2 -2
(kg) (MPa) (kW m ) (kW m ) (kW m ) (kW m )
279 1.65 0.32 176 281 295 188
710 2.13 0.34 191 306 > 195 196
1,272 1.86 0.33 201 322 287 117
1,708 2.44 0.36 207 331 312 212

Nearly all the data is covered by the radiated fraction range of 0.25 to 0.40. However, it
should be noted that the upwind and downwind average surface emissive powers at maximum
output, are in the range 242 to 294 kW m-2 ,i.e. much higher than at maximum projected area,
and that it could be misleading to use the values at maximum projected area. In general, the
measured average surface emissive power is only near its maximum value for about 50% of
the fireball duration. The CCPS6 suggest that a reasonable value for the surface emissive
power associated with large scale releases of hydrocarbon fuels is 350 kW m-2. Currently, a
value of 270 kW m-2 is utilised in LPG RISKAT2. British Gas3 measured surface emissive
powers of 320 to 370 kW m-2 from 1000 and 2000 kg releases of butane and propane at 0.75
and 1.5 MPa.

Solid-flame model: In the solid-flame model, the radiation received is calculated from the
surface emissive power (SEP, kW.m-2) of the flames, the relative geometry of the target and
fireball and the atmospheric attenuation. The incident radiation, I (kW m -2) is given by:

I = V . SEP . τ (10)

where V is the view factor and τ the atmospheric transmissivity.

The solid-flame model is more realistic than the point source model, particularly for
distances less than five fireball diameters away. However, the view factor calculations are
more complicated and the atmospheric conditions need to be known if a reasonable
atmospheric attenuation model is to be used.

CONCLUSIONS

The following conclusions are drawn:

• The resulting size and shape of the fireball following the BLEVE failure of a vessel was
dependent on the amount of fuel in the vessel at the time of failure and the mode of failure.
• The resulting external radiation field, and hence received dosage, are dependent on these
factors and also on the wind speed and direction as these effect the fireball trajectory during
its formation.

10
SYMPOSIUM SERIES No. 147 © IChemE

• The duration of the fireball was seen to be, as expected, dependent on the mass of fuel in the
vessel at the time of failure, varying from 3 seconds when 279 kg of propane was released, to
7 seconds when 1708 kg of propane was released.
• The maximum average surface emissive power measured during the experiments ranged
from 270 to 333 kW m-2 up/down wind and 278 to 413 kW m -2 crosswind. The highest values
were for the smallest release.
• The resultant fireballs gave their maximum power output before the fireballs reached their
maximum volume and close to the lift off time.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the support of Mr S Wright (Technology Division, HSE)
in the preparation of this paper.

REFERENCES

1 Pietersen, CM, 1988, Analysis of the LPG disaster in Mexico City, J.Haz. Mats, Vol.
20, pp. 85-108.
2 Hurst N and Trainor M, 1992, Quantified risk assessment for liquefied gas
installations, IBC conference on The Safe Handling of Pressure Liquefied Gases,
November 26/27, London.
3 Johnson D M, Pritchard M J and Wickens M J, 1990, Large scale catastrophic
releases of flammable liquids, CEC report, Contract No: EV4T.0014.UK(H).
4 Roberts A F, 1982, Thermal Radiation Hazards from Releases of LPG from
Pressurised Storage, Fire Safety Journal, Vol. 4, pp. 197-212.
5 CCPS, 1989, Guidelines for Chemical Process Quantitative Risk Assessment,
CCPS/AIChemE, ISBN
6 CCPS, 1994, Guidelines for Evaluating the Characteristics of Vapour Cloud
Explosions, Flash Fires and BLEVEs, CCPS/AIChemE, ISBN 0-8169-0474-X.
7 Prugh R W, 1994, Quantitative Evaluation of Fireball Hazards, Process Safety
Progress, Vol. 13, pp. 83-91.
8 Hardee H C, Lee D O and Benedick W B, 1978, Thermal hazard from LNG fireballs,
Combust. Sci. Tech., 17, pp. 189-197.

11
SYMPOSIUM SERIES No. 147 © IChemE

Land
Cyclops
camera
250 m 0

C 355 N
200 m
APPROXIMATE WIND CONDITIONS

B 310 D 45
N
2 m/s WESTERLY
150 m

150 m
125 m

150 m
100 m
100 m
100 m 75 m

50 m
15

100 m
150 m
E 90
50 m
100 m
75 m VESSEL
125 m
150 m Land radiometer
200 m 100 m Medtherm radiometer
270 NIRD radiometer and Agema camera
A Medtherm and WIRD radiometers
265
Field Control Room and sonic
anemometer
350 m

Figure 1. INSTRUMENT PLAN FOR 1708 KG RELEASE

12
SYMPOSIUM SERIES No. 147 © IChemE

279 kg 710 kg

1272 kg 1708 kg

Figure 2. MAXIMUM SIZE FIREBALL FROM EACH TRIAL

(not to scale)

13
SYMPOSIUM SERIES No. 147 © IChemE

1.51 Seconds After Vessel Failure 2.60 Seconds After Vessel Failure

3.69 Seconds After Vessel Failure 4.83 Seconds After Vessel Failure

Figure 3. THERMAL IMAGES FOR 1708 KG FIREBALL GROWTH

14
SYMPOSIUM SERIES No. 147 © IChemE

30

IN C ID E N T TH E R M A L R A D IA TIO N (kW m )
-2

25
Line A
Line B
Line C
20
Line D
Line E
15

10

0
0 1 2 3 4 5 6 7 8 9 10

TIM E (seconds)

Figure 4. VARIATION OF IRRADIANCE AT 100 METRES

(FOR 1708 KG RELEASE)

200

180
Line A
160
Line B
140
Line C
D O S A G E (kJ m )
-2

120 Line D

100 Line E

80

60

40

20

0
0 50 100 150 200 250 300

D IS TA N C E FR O M V E S S E L (m etres)

Figure 5. VARIATION OF DOSAGE AGAINST DISTANCE

(FOR 1708 KG RELEASE)

15
SYMPOSIUM SERIES No. 147 © IChemE

600

SURFACE EMMISSIVE POWER (kW/m2)

500

400

300

200
Crosswind maximum
Upwind maximum
100 Crosswind average
Upwind average

0
0 1 2 3 4 5 6 7
TIME (Seconds)

Figure 6. VARIATION OF SURFACE EMISSIVE POWER WITH TIME

(FOR 1708 KG RELEASE)

6000

Crosswind
5000 Upwind
PROJECTED SURFACE AREA (m2)

4000

3000

2000

1000

0
0 1 2 3 4 5 6 7
TIME (Seconds)

Figure 7. VARIATION OF PROJECTED AREA WITH TIME

(FOR 1708 KG RELEASE)

16