Professional Documents
Culture Documents
Abstract: High pressure jet fires pose a serious hazard to offshore installations operated by the
oil and gas industry as demonstrated by the Piper Alpha incident. Following the Piper Alpha inci-
dent a major initiative by the offshore oil and gas industry operators led to the production of
Interim Guidance Notes which provided guidance to operators on how to assess jet fire hazards.
However, many areas of uncertainty were identified where no data was available. Areas of par-
ticular concern identified in the Interim Guidance Notes were two-phase jet fires, the effect of
confinement on jet fires and their behaviour with water deluge. Since that time a considerable
body of experimental research has been undertaken. Based on this recent data, this paper reas-
sesses jet fire hazards in an offshore environment and provides updated guidance on the
hazards they pose, including tabulated data and simple calculation techniques for predicting
jet fire hazards.
Keywords: jet fires; two-phase jet fires; offshore fire hazards; water deluge; thermal load.
(formerly British Gas Research and Development) and con- water deluge systems on fire behaviour and consequences
fined condensate jet fires were studied by SINTEF in are also summarized.
Norway (SCI, 1998; Advantica, 1997a). Guidance is then presented for a simplified approach to
Advantica were well qualified to undertake this work since quantifying jet fire hazards (in terms of values for flame
from the early 1980s Advantica had been studying jet fire size, temperature, heat loads, and so on) for the range of
hazards associated with natural gas, principally related to fire sizes likely to be considered in a quantified risk assess-
gas transmission pipelines. Recognizing the importance of ment (QRA). These typical values for jet fire hazards are
studying these hazards at a representative scale, large and based on the experimental data reviewed here sup-
full scale experiments were undertaken at their Spadeadam plemented with predictions by validated models (developed
Test Site in Cumbria to determine the fire characteristics by Shell and Advantica) and can be used to assess the
(Wickens and Lowesmith, 1993; Cook et al., 1987; potential hazard to personnel and the likely affect to fire
Hankinson et al., 2000), although much remains unpublished. impacted obstacles. The approach taken in defining jet fire
Other studies of jet fires (conducted jointly with Shell Global hazards is similar to, but more detailed than, that presented
Solutions) included other fuels such as propane and butane within the Energy Institute guidance on severe fires (Energy
of particular relevance to onshore storage facilities (Bennett Institute, 2003).
et al., 1991; Shell, 1992; Davenport, 1994a; Sekulin and As the focus is offshore operations, the fuels for which gui-
Acton, 1995). dance is required are principally pressurized gas, oil fractions
After Phase 2, Advantica continued work on gas jet fire and their mixtures, sometimes including water. The guidance
hazards (Gosse and Pritchard, 1995; Advantica, 1997b) does not extend to activities involving liquefied natural gas
and Shell undertook large scale experiments studying gas/ (LNG). Experimental studies generally used processed oil
kerosene jet fires (Davenport, 1994b). Attention was also fractions such as diesel or kerosene and these are con-
focused on fire mitigation using active water deluge. Advan- sidered to be representative of the behaviour of light crude
tica initiated two JIPs to study the effectiveness of water oils. Some experimental studies related to LPG (propane
deluge to mitigate jet and pool fire hazards (Advantica, and butane) and, although not generally experienced off-
1997c, 2000a), whilst Advantica and Shell jointly conducted shore except as part of the composition, where appropriate
a JIP to study jet fires involving oil/gas/water mixtures the results of these studies have been included.
(Advantica, 2000b). Only limited information from these
JIPs has been published thus far (Gosse and Hankinson,
NATURE AND CHARACTERISTICS OF JET FIRES
2001; Hankinson and Lowesmith, 2004; Hankinson et al.,
forthcoming). Experiments studying dedicated vessel Jet fires can be produced following the pressurized
deluge with jet fires have also been conducted at Spadea- release of a variety of fuel types. The simplest being a
dam and elsewhere (Shirvill and White, 1994; Shirvill, 2004; pressurized gas giving rise to a gas jet fire. A pressurized
Roberts, 2004). liquid/gas mixture (such as ‘live crude’ or gas dissolved in
The effect of confinement (with and without deluge) on fire a liquid) will give rise to a two-phase jet fire. The gas
behaviour was studied during the BFETS JIP experiments stream atomizes the liquid into droplets which are then
conducted at SINTEF and subsequently in further studies evaporated by radiation from the flame. However, a pressur-
by Shell (Chamberlain, 1994; Chamberlain and Brightwell, ized release of a liquid can also give rise to a jet fire in
1994). which two-phase behaviour is observed if the liquid is able
It was always the intention of the BFETS JIP that the to vaporize quickly. This is most likely to occur when a
IGNs would be updated after Phase 2 and this was liquid is released from containment at a temperature
completed initially in relation to explosion hazards. After above its boiling point at ambient conditions whereupon
some delay, the updating of the IGNs for fire hazards is flash evaporation occurs (e.g., propane, butane) and a
now in progress with the support of the UK Offshore flashing liquid jet fire results. Non-volatile liquids (e.g., kero-
Operators Association (UKOOA, 2006). sene, diesel, or stabilized crude) are unlikely to be able to
sustain a two-phase jet fire, unless permanently piloted by
an adjacent fire; even so, some liquid drop-out is likely
and hence the formation of a pool.
SCOPE OF PAPER
As part of the updating of the IGNs for fire hazards, infor-
Flame Stability
mation on the nature of fire hazards likely to be experienced
offshore and guidance on their quantification is being Whether or not a stable jet fire will arise following the
reviewed taking into account the wealth of large scale exper- releases of a pressurized hydrocarbon will depend principally
imental data that has been generated during and since upon the nature of the fuel, the size of the hole from which the
Phase 2 of the BFETS JIP, as described above. This paper release occurs and the geometry of the surroundings. In the
presents the results of this review in relation to jet fire case of natural gas, it has been found that for free jets (not
hazards which forms the basis of the guidance relating to impacting) some combinations of hole size and pressure
jet fires which is to be included within the updated industry cannot produce stable flames (Birch et al., 1988). Figure 1
guidance notes (UKOOA, 2006). shows that for hole sizes under 30 mm diameter, there is a
Based on data from the large scale experiments outlined lower bound pressure which vertical high pressure releases
above including much data which has not been previously must exceed to produce stable jet fires. In practice this
published, the nature and characteristics of hydrocarbon jet means that most small leaks will be inherently unstable and
fires, especially in relation to offshore oil and gas installa- will not support a flame without some form of flame stabiliz-
tions, are described. The effects of confinement and active ation, such as the presence of another fire in the vicinity to
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207– 220
HYDROCARBON JET FIRE HAZARDS IN THE OIL AND GAS INDUSTRY 209
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207 –220
210 LOWESMITH et al.
For high pressure gaseous releases, the mass release rate radiated, F, for such fires as can be seen in Figure 3. (F is
through an orifice, Ṁ, can be related to the area of the hole, defined as ‘(energy released as radiation from the flame
Ad, and pressure, P, using surface/net energy of combustion)’). As can be seen,
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi F increases with carbon number, reflecting the increased
u ggþ1 ! radiative emissions from soot within the higher hydrocarbon
u g m 2 1
_ ¼ Cd Ad P10
M 6 t (2) jet fires.
RTg Z g þ 1
For gas– liquid fuel mixtures, interpolation based on the
percentage liquid within the mixture provides a reasonable
For natural gas, this can be approximated for circular holes estimate of F as can be seen in Figure 4 for butane/natural
of diameter, d (mm), by gas mixtures.
A special case of interest at some installations is ‘live’
_ ¼ d 2 P 103
M (3) crude that includes a significant quantity of water. Experi-
ments (Hankinson et al., 2007) have shown that mixtures
Substituting into equation (1), the jet fire length for natural with a ‘water cut’ [defined as (mass of water/mass of
gas can be approximated by fuel) 100%) of up to 125% remain flammable, although
not necessarily capable of supporting a stable flame in the
L ¼ 0:933d 0:746 P0:373 (4)
absence of some other supporting mechanism. The inclusion
of water also slightly increases flame length and flame
buoyancy, and significantly reduces the amount of smoke
Flame Radiative Emissions
produced. For water cuts less than 50% there is little
As noted above, the combustion process within a natural impact on the fraction of heat radiated but for higher water
gas jet fire is relatively efficient and produces little soot cuts the fraction of heat radiated is reduced (Figure 5).
(carbon). Consequently, these flames are not as luminous
as higher hydrocarbon flames. Radiation emissions from
Heat Loads and Flame Temperatures
natural gas jet fires arise mostly from water vapour and
carbon dioxide, except for very large releases where soot The thermal load to an engulfed object in a jet fire will be a
production starts to enhance the process. The long thin combination of radiative load and convective load from the
shape may also result in a flame path which may not be opti- hot combustion products passing over the surface. Clearly
cally thick. The net result is that the radiative heat transfer to the total heat flux which is imparted to an engulfed object
the surroundings is lower than for comparable higher hydro- will vary over the surface of the object. In addition, the relative
carbon jet fires. This is reflected in the fraction of heat proportions of convective and radiative heat flux will vary over
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207– 220
HYDROCARBON JET FIRE HAZARDS IN THE OIL AND GAS INDUSTRY 211
the surface, with the highest convective component likely to cylinder (pipe) impacted by a horizontal high pressure gas
be experienced close to the point of impact of a flame jet fire (Pritchard and Cowley, 1991). The cylindrical surface
where the highest velocities occur, whereas the highest is presented flat by cutting along the rear. When the pipe
radiative heat load will be experienced where the more was located at 21 m (towards the end of the flame) the
radiative part of the flame (usually towards the end of the maximum heat fluxes were experienced at the point of
flame) is viewed by the object. As the more radiative part of impact on the front of the pipe [Figure 6(a)]. At 15 m
the flame is closer to the tail, this can result in the highest [Figure 6(b)] the heat loads were relatively uniform around
overall heat fluxes being experienced on the rear surface of the pipe, but at 9 m [Figure 6(c)] the heat loads were greatest
an engulfed object which may seem counter-intuitive. to the rear of the pipe due to radiation from the tail of
Figure 6 shows total heat fluxes experienced by a horizontal the flame.
Figure 4. Fraction of heat radiated for fuel mixtures. Figure 5. Effect of water cut on fraction of heat radiated.
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207 –220
212 LOWESMITH et al.
Figure 6. Variation of total heat flux over the surface: (a) pipe at 21 m; (b) pipe at 15 m; (c) pipe at 9 m.
Due to the radiant soot emissions, the radiative heat trans- phenomenon can be explained by considering the maximum
fer from higher hydrocarbon jet fires is generally greater than time averaged flame temperatures measured in over 180
that from natural gas flames and the generally lower vel- large scale gas jet and flashing liquid jet fires (Bennett
ocities arising from flashing liquid releases (such as propane et al., 1991; Sekulin and Acton, 1995; Advantica, 1997b) pre-
or butane) result in a lower convective flux to engulfed sented in Figure 9. As can be seen, the maximum flame
objects. temperatures are higher for the gas jet fire (on average
For the rear surface of an engulfed object, Figure 7 shows 12808C and up to 15008C), compared to a flashing liquid
that the fraction of the heat flux which is radiative increases fire which has a lower flame temperature (typically 10508C).
from about 0.5 for natural gas to about 0.8 for fuels containing In the case of two-phase jet fires, the flame temperature is
a large proportion of higher hydrocarbons (data from Bennett dominated by the gas content but the flame emissivity will
et al., 1991; Davenport, 1994a, b; Advantica, 1997a). also be enhanced by the higher hydrocarbon content leading
In the case of a pressurized gas–liquid mixture (such as to overall higher radiative fluxes for such mixtures.
‘live’ crude), the high gas velocities may still occur and Figure 8 includes data with mass flowrate between 2.5 and
result in a high convective contribution, whilst the higher 5 kg s21 and shows that for a flashing liquid fuel the maxi-
hydrocarbon content maintains a high radiative contribution; mum heat flux is generally around 200 kW m22 whereas for
making these type of jet fires a ‘worst case’ in terms of total gas only (0% liquid), the maximum heat flux is typically
heat flux to engulfed obstacles. Experimental work 250 –275 kW m22. In experiments involving natural gas at
(Davenport, 1994a, b; Advantica, 1997a) suggests that the higher flowrates (up to 10 kg s21) maximum heat fluxes up
maximum combined fluxes occur for gas –liquid mixtures
which are about 60 –80% by mass of liquid (Figure 8). This
Figure 7. Radiative heat flux as a fraction of total heat flux to rear of Figure 8. Maximum heat fluxes to an engulfed object for gas– liquid
engulfed object. mixtures.
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207– 220
HYDROCARBON JET FIRE HAZARDS IN THE OIL AND GAS INDUSTRY 213
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207 –220
214 LOWESMITH et al.
Figure 11. Performance of different deluge strategies with a two-phase jet fire. (a) Top of vessel; (b) front; (c) rear; (d) bottom.
the surroundings, which protects adjacent plant and in par- Increased deluge rates can further reduce incident radi-
ticular, aids escape by personnel. Similarly water curtains ation levels: 60 –70% at 18 l m22 min21; 80– 90% at 24 l
can be used to protect escape corridors. Figure 12 presents m22 min21 for general area deluge. Nozzles producing smal-
large scale data (Advantica, 1997c; Advantica, 2000a) on the ler droplet sizes can have an enhanced mitigation effect, but
reduction of incident radiation due to water deluge by MV57 there is an increased risk that the droplets will be blown away
type nozzles spaced 2.84 m apart. For this arrangement, by the wind (Advantica, 2000a).
operating at 12 l m22 min21, incident radiation levels can The mitigation of incident radiation is due to attenuation of
be reduced by about 20% for a single row of nozzles, radiation by the water droplets in the atmosphere, that is,
30–40% for two rows and 40–60% for more than two rows effectively reducing the atmospheric transmissivity, t. The
(general area deluge). reduction in incident radiation, Is, with distance, s, due to
this attenuation is characteristically expressed by an equation
of the form
Is ¼ I0 eats (5)
N:V_ W 1
Wf ¼ (6)
AW :UW 60 000
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207– 220
HYDROCARBON JET FIRE HAZARDS IN THE OIL AND GAS INDUSTRY 215
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207 –220
216 LOWESMITH et al.
which have been extensively validated against large scale heat released by combustion as
data. A range of such models are available on a licence or
consultancy basis. However, a simplified approach was pro- EAf
F¼
posed in the Interim Guidance Notes (SCI, 1992), whereby 1000 Q
correlations for flame dimensions were suggested for jet
and pool fires together with guidance on typical heat loadings The incident radiation received in the far field at a distance,
to engulfed objects. Similarly, guidance on typical heat load- s (m), from the fire, Is, is then expressed as
ings from fires was provided in the Energy Institute document _
for severe jet and pool fires (Energy Institute, 2003). 1000 tF MH
Is ¼ 2
(10)
In this paper, a similar approach is taken in order to provide 4 ps
tabulated guidance on assessing jet fire hazards, based on The atmospheric transmissivity will depend upon the pre-
recent knowledge and large scale experimental work. The vailing atmospheric conditions (absolute humidity) and the
guidance values (which are conservative) can then be used path length, but might typically be 0.8 on a dry day for dis-
to undertake an initial simplified QRA. If necessary, problem tances likely to considered. However, fog would significantly
areas can then be identified for more rigorous assessment. reduce this value.
Guidance values for jet fire hazards are provided for the fol- The activation of water deluge on an offshore installation,
lowing range of sizes of leak: producing water droplets in the air, effectively reduces the
. Small—typically 0.03 –0.3 kg s21, represented by transmissivity by absorbing radiation. However, for the pur-
0.1 kg s21; poses of this paper, a revised ‘effective’ fraction of heat
. Medium—typically 0.3–3 kg s21, represented by 1 kg s21; radiated is defined as F0, accounting for the reduced trans-
. Large—typically 3–30 kg s21, represented by 10 kg s21; missivity in the area being deluged, (but not including the
. Major failure—over 30 kg s21. transmissivity of the atmosphere between the fire and the
receiver which is outside the deluged area). The incident radi-
For each size category, the information is provided to allow ation at a distance, s (m), from the fire can be estimated using
assessment of the fire size, the incident radiation field to per-
sons and objects outside the fire, the thermal loading to _
1000 tF 0 MH
objects engulfed by the fire, the smoke hazard presented Is ¼ (11)
4 ps2
by the fire, the likely effects of active water deluge and the
effects of confinement. (It should be noted that for leak This ‘point source’ model can be useful for estimating the
sizes of 10 kg s21 and above, the fire is large compared distance to a given radiation level. In the case of personnel,
with the average offshore module and it may be necessary it is important to consider both the level of radiation and the
to consider the fire to be ‘confined’.) duration of exposure. The thermal dose is often defined
as the product of the incident radiation and the duration
(kJ m22). However, the effect on people correlates better
with a ‘thermal load’, J defined by
Incident Radiation to the Surroundings ð
J ¼ Is(4=3) dt (12)
An obvious hazard presented by a fire is the thermal radi-
ation to the surroundings, in particular to personnel during
escape and evacuation. The incident thermal radiation, I, to Threshold values of dosage associated with serious injury
a person or object from a fire can be described in terms of or death can be found in the literature (Lees, 2005; UKOOA,
the flame emissive power, E, and the view factor, V, of the 2006). However, as a guide, a radiation level of about 5 kW
flame from the position of the receiver as m22 can be tolerated for about one minute and consequently
represents a level from which it is reasonable to assume that
escape, without significant injury, would usually be possible
I ¼ VE t (kW m2 ) (9) for an average employee wearing typical protective clothing.
However, the point source model is not suitable for estimat-
The view factor is a function of the flame shape. Conse- ing incident radiation to locations close to (certainly within
quently, most integral or empirical mathematical models of one flame length) of the flame, where it may be significantly
fires will assume some kind of simplified flame shape which in error. In the near field, a mathematical model which defines
is then used to calculate the view factor. The flame average a realistic flame shape (and ideally a variation of surface
surface emissive power is also a function of the flame emissive power over the flame to allow for smoke obscura-
shape. Therefore, average surface emissive powers used tion) should be used. Typical values of F and F0 are provided
by the model will not necessarily be the same as those in the tabulated guidance for the different sizes and types of
measured during an actual fire. jet fire so that estimates of the far field incident radiation
In the far field, (typically more than two flame lengths away) hazard can be made using the point source model equations
the flame shape is not critical, so a simplified approach can (10) and (11) given above.
be taken using the ‘point source’ model, whereby the difficul-
ties of defining a flame shape and associated average sur-
Thermal Loading to Engulfed Objects
face emissive power can be avoided. In this approach, the
fraction of the heat of combustion of the fuel radiated to the The thermal load per unit area, ql, to an object engulfed by
surroundings, F, is defined in terms of the flame area, Af, fire will be a combination of radiation from the flame (qrl) and
the flame surface emissive power, E, and the net rate of convection from the hot combustion products (qcl) passing
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207– 220
HYDROCARBON JET FIRE HAZARDS IN THE OIL AND GAS INDUSTRY 217
over the object surface. Hence the thermal load can be writ- provided in the tabulated guidance for the jet fire types and
ten as sizes.
However, researchers often quote heat fluxes measured
ql ¼ qrl þ qcl ¼ 1f sTf4 þ tf sTa4 þ h(Tf Ts ) during experiments using calorimeters (for total heat flux)
and radiometers (for radiative flux). These instruments are
However, not all the thermal loading is necessarily designed to have a surface emissivity close to 1 and are
absorbed by the surface, some may be reflected back and maintained at a low temperature throughout the experiments.
the surface will also lose heat by radiation. Hence the total Hence the fluxes measured and reported for calorimeters are
absorbed load per unit area (qa) is given by given by equation (13) with 1s ¼ 1 and Ts 333 K and can be
regarded as a conservative estimate of the total heat flux
qa ¼ 1f as sTf4 1s af sTs4 þ as tf sTa4 initially absorbed by an engulfed object. Radiometers are
designed and calibrated to measure 1f sTf4 . At an early
1s tf sTs4 þ h(Tf Ts )
stage, whilst Ts is low, the flux measured by a radiometer is
approximately (qra =1s ) and so, if 1s is taken as 1, it provides
Assuming that both the flame and surface can be con-
a conservative estimate of the radiative flux absorbed by an
sidered as diffuse grey bodies, then as ¼ 1s, af ¼ 1f and
object. Subtracting the measured radiative flux from the
tf ¼ 1 2 1f. Furthermore, the term involving Ta is small and
measured total heat flux enables the initial convective flux
can be neglected, giving
absorbed by the object to be determined. Using an estimated
or measured value for Tf enables h to be determined at the
qa (kW m2 ) ¼ qra þ qca
measurement location. In this way, experimentally measured
¼ 1s s(1f Tf4 Ts4 ) þ h(Tf Ts ) (13) flux levels can be used to derive input data for transient heat-
up calculations. Consequently, typical values of the total,
radiative and convective fluxes (as defined in the above
When considering an actual object engulfed by a fire, the
with 1s ¼ 1 and Ts 333 K) are also provided in the tabulated
flame temperature exposed to different parts of the object
guidance for the range of fire types and sizes. These rep-
surface may vary; similarly the velocities in the flame (and
resent initial heat fluxes to the fire engulfed object (that is,
hence convective heat transfer coefficient) may vary.
when the object surface is still cold and the temperature
Hence, to determine the total load absorbed by an object
difference is greatest). A simplified method for heat-up calcu-
the above equation should be summed over the area of the
lations can then be used (such as that given in UKOOA,
object.
2006; Energy Institute, 2003) to determine the temperature
Also, as the object engulfed in the flame heats up, the
of the engulfed object with time.
absorbed load will reduce. This is particularly the case with
the convective load which reduces linearly with increasing
object temperature. Therefore, for an accurate transient cal-
Tabulated Guidance
culation of the temperature rise of an object, the parameters
Tf, 1f and h are required, together with the emissivity of the Table 1 provides guidance values for gas jet fires and
surface (which itself may change as the surface heats up). Table 2 for two-phase jet fires. As noted above, the maximum
Where possible, typical values of these parameters are heat fluxes to engulfed objects have been found to occur
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207 –220
218 LOWESMITH et al.
Note: For fraction of heat radiated, Fm, of mixture involving x% liquid by mass: use Fm ¼ (x=100):(FL FG ) þ FG where FG is the fraction of heat
radiated for natural gas as given in Table 1 and FL is the fraction of heat radiated for the liquid fuel involved. Take FL ¼ 0.24 for C3; 0.32 for C4,
0.45 for C6– C25 (including condensate and diesel); and 0.5 for crude oil.
when the fuel mixture is about 30% gas and 70% liquid by these guidance values with the data measured during the
mass. Consequently, the values shown in Table 2 correspond tests for maximum total heat flux and the proportion of radia-
to this worst case condition. tive and convective fluxes measured at locations on the front,
For flashing liquid fires (such as propane or butane) a top and rear surface of the fire engulfed pipe. Clearly, the
lower flame temperature of about 1300 K is likely with an simplified guidance values take no account of the spatial vari-
emissivity of 1, giving a radiative flux of about 160 kW m22. ation of heat loading over the surface of a fire engulfed object.
A convective heat transfer coefficient of about 0.08 kW m22 For a conservative assessment of the response of the pipe
K21 is suggested giving a convective flux of about 80 kW to the heat loading, the temperature of the engulfed pipe with
m22 and a total flux of about 240 kW m22. time can be calculated for a series of time steps (Dt) as
follows:
At time ti ¼ ti1 þ Dt(i 1) the thermal load absorbed by
Validation Exercise the object, qi , is given by
Consider a ‘live’ crude jet fire at 5 kg s21, which is 20%
gas, 80% oil, released at 20 bar. Three experiments with qi ¼ 1s s(1f Tf4 Ti4 ) þ h(Tf Ti ) (i 0) (14)
these test conditions were conducted during the BFETS
Phase 2 project (SCI, 1998; Advantica, 1997a), two of
which impacted onto a pipe (Tests 9 and 11) and the third
being a free flame (Test 3).
Using Table 2, this would fall into the ‘large’ release cat-
egory of 10 kg s21 giving a flame length of 35 m which
gives a conservative assessment compared to the measured
length of 29 m for the free flame and 21 m and 27 m for the
impacting flames.
Using the formula for fraction of heat radiated below Table 2,
with FG ¼ 0.13 and FL ¼ 0.5, Fm ¼ (80=100):(0:5 0:13)
þ0:13 ¼ 0:426, Figure 14 presents incident radiation
measured during the jet fires with calculations made using
the point source model Is ¼ (1000tF MH _ m )=(4ps2 ), taking
t ¼ 0.8 and Hm ¼ 42 MJ kg21.
Table 2 suggests a maximum total heat flux of 350 kW m22
whereas during the two experiments where the flame
impacted a steel pipe, the maximum heat fluxes measured
were 348 and 370 kW m22 . Table 2 suggests that the radia-
tive flux accounted for 65% of the total flux (230 kW m22) with
34% being convective (120 kW m22). Table 3 compares Figure 14. Incident radiation from ‘live’ crude oil jet fire.
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207– 220
HYDROCARBON JET FIRE HAZARDS IN THE OIL AND GAS INDUSTRY 219
Table 3. Comparison of guidance values for heat flux with large scale data.
Maximum
heat flux Front Top Rear Avg Front Top Rear Avg
Data Test 9a 370 41% 61% 68% 57% 59% 39% 32% 43%
Data Test 11a 348 43% 86% 100% 76% 57% 14% 0% 24%
a
Test 9 impacted the pipe at 9 m and Test 11 at 15 m from the release point.
NOMENCLATURE
a constant
Figure 15. Heat up of engulfed object in ‘live’ crude jet fire. Ad area of release hole, m2
Af area of flame, m2
Aw deluged area, m2
Ḃ flowrate of air entrained for combustion, kg s21
and the temperature of the object increases to Tiþ1 ¼ C specific heat capacity of steel, kJ kg21 K21
Cd coefficient of discharge of orifice
Ti þ DTi (i 0) where
Cw water curtain thickness, m
d diameter of release hole, mm
Dtqi E flame emissive power, kW m22
DTi ¼ (15)
Crx F fraction of heat radiated by a flame
H net calorific value, MJ kg21
Figure 15 shows a comparison of the calculated tempera- h convective heat transfer coefficient, kW m22 K21
I incident radiation, kW m22
ture rise of the pipe with data measured by thermocouples
J thermal load to personnel, (kW m22)4/3 s
embedded within the pipe wall at three locations during the Is incident radiation at a distance s, kW m22
two impacting fire experiments. The three locations were: L jet fire length, m
close to the point of impact of the fire (front of pipe); 908 Ṁ mass flow rate, kg s21
from this location on top of the pipe; and 1808 from the m molecular weight, kg
N number of nozzles over a deluged area
point of impact on the rear of the pipe. The data used for P absolute pressure, MPa
the calculations was: C ¼ 520 J kg21 K21, r ¼ 7850 kg Q net rate of heat released by combustion (Q ¼ MH),_ MW
m23, s ¼ 5.6697 W m22 K24, x ¼ 0.0125 m, 1s ¼ 0.7, qa total thermal load absorbed by an engulfed object, kW m22
Dt ¼ 1 s, together with Tf ¼ 1560 K, 1f ¼ 0.7 and h ¼ 95 W qi total thermal load absorbed by an engulfed object at time ti ,
kW m22
m22 K21 from Table 2.
ql total thermal load to an engulfed object, kW m22
qra radiative load absorbed by an engulfed object, kW m22
qca convective load absorbed by an engulfed object, kW m22
qrl radiative load to an engulfed object, kW m22
CONCLUSIONS qcl convective load to an engulfed object, kW m22
Jet fires pose a serious hazard to oil and gas installations R Universal gas constant, J mol21 K21
r mass ratio of air to fuel for stoichiometric combustion
with the potential for escalation to a major incident. Hence it is RI reduction in incident radiation,%
important for operators to have a good understanding of fire s distance, m
hazards posed by their installation and to be able to quantify Ta ambient temperature, K
them as part of their Safety Case. Following the Piper Alpha Tf flame temperature, K
disaster, the Interim Guidance Notes provided guidance on Tg temperature of gas release, K
Ti surface temperature at time ti , K
how to assess fire hazards based on the information avail- Ts surface temperature, K
able at that time. However, many areas of uncertainty were ti time at step i of iterative calculation, s
identified where no data was available, especially in relation Uw water droplet velocity, m s21
to two-phase fires, confinement and deluge. V view factor of flame from position of receiver
V̇w water flowrate per nozzle, l min21
Since that time a considerable body of experimental Wf water volume fraction
research of large scale jet fires has been undertaken provid- x thickness of steel, m
ing an increased understanding of jet fire behaviour. Based Z compressibility factor
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207 –220
220 LOWESMITH et al.
Greek Symbols Gosse, A.J. and Pritchard, M.J., 1995, Large scale jet fire impaction
af flame absorptivity onto a flat surface, Proceedings of the IGRC, Cannes, France.
as surface absorptivity Gosse, A.J. and Hankinson, G., 2001, Use of water deluge to mini-
1f flame emissivity mise hazards of oil and gas fires offshore, Proceedings of the
1s surface emissivity IGRC, Amsterdam, The Netherlands.
g ratio of specific heats Hankinson, G., Lowesmith, B.J., Genillon, P. and Hamaide, G., 2000,
f equivalence ratio Experimental studies of releases of high pressure gas from punc-
r density of steel, kg m23 tures and rips in above-ground pipework, International Pipeline
s Stefan–Boltzmann constant, kW m22 K24 Conference, Calgary, Canada.
t atmospheric transmissivity Hankinson, G. and Lowesmith, B.J., 2004, Effectiveness of area and
tf flame transmissivity dedicated water deluge in protecting objects impacted by crude
oil/gas jet fires on offshore installations, J Loss Prevention in the
Process Industries, 17: 119–125.
Hankinson, G., Lowesmith, B.J., Evans, J.A. and Shirvill L.C., 2007,
REFERENCES Jet fires involving releases of crude oil, gas and water, IChemE
Process Safety and Environmental Protection, 85.
Advantica, 1997a, BFETS Phase 2: Horizontal jet fires of oil and gas, Lees, 2005, Lees’ Loss Prevention in the Process Industries, Vol. 1
unpublished Advantica report. (Elsevier Butterworth-Heinemann Ltd, London, UK).
Advantica, 1997b, Large scale natural gas jet fires impacting on flat Pritchard, M.J. and Cowley, L.T., 1991, Thermal impact on structures
surfaces, unpublished Advantica report. from large scale jet fires, Safety Developments in the Offshore
Advantica, 1997c, JIP: The effectiveness of water area deluge in miti- Industry, Inst of Mech Eng, Glasgow.
gating the effect of fires, unpublished Advantica report. Roberts, T.A., 2004, Effectiveness of an enhanced deluge system to
Advantica, 2000a, JIP: A programme of large scale experiments to protect LPG tanks and sensitivity to blocked nozzles and delayed
study the effectiveness of water deluge in mitigating potential off- deluge initiation, J Loss Prevention in the Process Industries, 17:
shore jet and pool fires, unpublished Advantica report. 151– 158.
Advantica, 2000b, JIP: Large scale experiments to study jet fires of Steel Construction Institute (SCI), 1992, Interim guidance notes for
oil/gas/water mixtures, unpublished Advantica report. the design and protection of topside structures against explosion
Bennett, J.F., Cowley, L.T., Davenport, J.N. and Rowson, J.J., 1991, and fire.
Large scale natural gas and LPG jet fires, final report to EC, Shell Steel Construction Institute (SCI), 1998, JIP on blast and fire engineer-
Report TNER.91.022. ing for topside structures Phase 2, final summary report, SCI 253.
Birch, A.D., Brown, D.R., Cook, D.K. and Hargrave, G.K., 1988, Sekulin, A.J. and Acton, M.R., 1995, Large scale experiments to
Flame stability in underexpanded natural gas jets, Combustion study horizontal jet fires of mixtures of natural gas and butane,
Science and Technology, 58: 267– 280. final report to the EC, Advantica report GRC R 0367.
Chamberlain, G.A., 1994, An experimental study of large scale com- Shell, 1992, Large scale butane jet fires data, unpublished Shell
partment fires, Trans I Chem E, Part B, 72: 211 –219. reports TNER.92.012 to TNER.92.053.
Chamberlain, G.A. and Brightwell, H.M., 1994, Large scale compart- Shirvill, L.C. and White, G.C., 1994, Effectiveness of deluge systems
ment fires, experimental data, HSE report OTO 94 011-024. in protecting plant and equipment impacted by high velocity natural
Cook, D.K., Fairweather, M., Hammonds, J. and Hughes, D.J., 1987, gas jet fires, International Symposium on Heat and Mass Transfer
Size and radiative characteristics of natural gas flares. Part I— in the Chemical Process Industry, Rome, Italy.
Field scale experiments, Chem Eng Res Des, 65: 310– 317. Shirvill, L.C., 2004, Efficacy of water spray protection against pro-
Cowley, L.T. and Johnson, A.D., 1992, Oil and Gas Fires: Character- pane and butane jet fires impinging on LPG storage tanks,
istics and Impact. JIP BFETS Phase 1, published by HSE as OTI J Loss Prevention in the Process Industries, 17: 111 –118.
92 596. UKOOA, 2006, Fire and explosion guidance. Part 2: Avoidance and
Cullen, W.D., 1990, The Public Enquiry into the Piper Alpha Disaster mitigation of fires, UK Offshore Operations Association, final draft
(Department of Energy, HMSO, London, UK). available from www.fireandblast.com\lgn-update.
Davenport, J.N., 1994a, Large scale natural gas/butane mixed fuel Wickens, M.J. and Lowesmith, B.J., 1993, Fire and explosion
jet fires, final report to EC, Shell report TNER.94.030. hazards—large scale experiments to assess and mitigate their
Davenport, N., 1994b, Large scale natural gas/kerosene mixed fuel effects. Presented at Inst Gas Engineers, Eastern Section.
jet fires, final report to the API, Shell report TNER.94.061.
Drysdale, 1985, An Introduction to Fire Dynamics (John Wiley &
Sons, Chichester, UK). The manuscript was received 29 June 2006 and accepted for
Energy Institute, 2003, Guidelines for the Design and Protection of publication after revision 6 October 2006. The paper was published
Pressure Systems to Withstand Severe Fires (Energy Institute, online ahead of print 20 March 2007.
London, UK).
Trans IChemE, Part B, Process Safety and Environmental Protection, 2007, 85(B3): 207– 220