Beruflich Dokumente
Kultur Dokumente
c
(T
1
T
c
)) (8)
where
m = vapour mass fraction
c = liquid specific heat (j/Kg) at T
c
= latent heat of vaporization of liquid (J/Kg) at T
c
T
1
= reservoir storage temperature (K)
Note : If m is negative then liquid flow only will occur and equation (4) should be used
to calculated the flow rate.
(4) Assuming homogeneous mixing and so slip between the phases, the mixture density is
calculated as :
c
m
g
m
+
_
,
1
1
1
(9)
where
g
is the vapour density at T
c
and P
c
(kg/m
3
)
1
is the liquid density at T
c
and P
c
(kg/m
3
)
(5) The standard discharge formula is then used to calculate the critical flow rate :
M = 0.6 A
2
c o c
P P ( )
( 10 )
where M is the mass flow rate ( kg/s)
This calculation should be used for large leaks and ruptures of lines and hoses, where
two-phases flow will form upstream of the break. Note that if the leak is fed by a pump
the flow rate could be determine by the pump capacity rather than the flow regime.
In the case of a pipe rupture the escaping LPG will probably be fed from both sides of
the break, e.g. from the pump as well as from the receiving vessel.
3.3 VAPOUR JETS DISPERSION AND FIRES
3.3.1 Relief valve vapour releases
Single phase (vapour only) emission is assumed from vessel relief valves, as generally
observed under fire engulfment conditions. The case of vessel overfill and subsequent two
phase emission through the pipe is taken as a special case of the two-phase jets and fires
dealt with in the next section.
For the discussions of relief valve dispersions and fires which follow ,dispersion and jet fire
radiation calculations have carried out for wide range of emitted vapour flow rates, which
cover all sizes of tanks in general use within Group operations. The selected flow rates
are:
Propane : 2.5, 5, 7.5,10,15, 20, 25, 30, 40 ( kg/s)
Butane : 2.5, 5, 7.5, 10,15, 20, 25, 30 ( kg/s)
In deriving the various consequences it is necessary to make various assumption in regard to
the particular facility and equipment. In order to carry out calculation to complete Tables 2 and
3,typical tanks sizes, valves sizes, vent stack height and length have been used (see Figure
1) . Where the facility under examination differs from this model facility, interpolation between
the calculated figures will be necessary.
3.3.1.1 Dispersion
As the source of emission under consideration in this section is vapour only, the
PLUMEPATH dispersion package has been used for emissions well above ground level. An
example is shown in Figure 2 for butane emitted from a vertical relief valve vent pipe.
Dispersion plumes have been calculated for the mass flow rates set out above. In all these
cases mixing with air occurs rapidly and it may be assumed that flammable plumes do not
reach down to the ground. Therefore no dispersion data is presented here for vapour
emission from relief valves.
3.3.1.2 Fires
The radiation fields generated by relief valve fires have been modeled with computer
packages which have been developed by Thornton Research Centre and validated by
experimental studies. Tables 2 and 3 present distances to the critical ground level fluxes of
1.5, 5, 8 and 13 kW/m and to the critical tank top level fluxes of 8, 32 and 44 kW/m, as
identified in the Manual/PTS.
A typical wind speed of 5 m/s has been assumed in all calculations. It should be noted that if
wind speeds are considerably higher radiation flux levels close to the stack can be
significantly higher. If this could be critical, further radiation calculations will be required for
higher wind speeds in which case PETRONAS should be consulted.
As may be seen from the Tables the higher flux levels are frequently not achieved. In addition
to the critical distance data, the Tables also present flame lengths, flame lift-off and stack
outlet height above grade.
3.3.2 Other vapour releases
For the dispersion of vapor releases from sources which do not behave as relief valve
releases, in that they may be horizontal and near the ground, dispersion calculations will be
very unreliable. This is because the presence of other equipment in the area will cause
unpredictable air movement. The distance to lower flammable limit (LFL) will be smaller than
the conservative estimates which can be obtained by using the data in Figure 3 for liquid
releases.
For the assessment of vapour fires reference should be made to the vapour fire curves shown
in Figures 4 to 16 (a description is given in section 3.4.2). As propane is less radiative than
butane these results may be used for propane as well.
3.4 TWO-PHASE JETS - DISPERSION AND FIRES
Two-phase jets may arise from a wide range of events in a facility and may lead to a variety of
consequences. As set out in Part One of these Guide-lines, the following situations have been
identified as having the potential for the creation of a serious incident.
flange (joint) leak
pump seal leak
open drain valve
small bore pipe rupture
hose leak or rupture
tank overfill/relief valve discharge
As may be seen from this list, a more disparate collection of release situations is possible in
comparison to the relief valve cases considered in the previous section. Again consequence
information is presented calculated for a wide range of mass flow rates at the source. The only
geometrical factor which has been introduced to help differentiate two phase discharge cases is
the division into horizontal and vertical emissions.
A wide range of flow rates to cover the range and scale of the principal incidents listed above
have been considered.
3.4.1 Dispersion
Before describing the dispersion calculations adopted and results obtained, some words of
caution are necessary. The calculation of highly-turbulent flashing two-phase flows is an active
and difficult research field. The guidance provided here is therefore subject to revision as better
models are developed and validated.
The principal purpose of the dispersion calculations is to assess the extent of the flammable
cloud which will be formed. The distance to the lower flammable limit (LFL) is thus the main
parameter calculated. This then enables a judgement to be made in regard to facilities engulfed
by the flammable plume and the proximity to the site boundary. As mentioned in Part One, the
initial consequence of ignition of this cloud will be a transient flash fire which although of high
intensity will only seriously affect people within it. The major effect of a flash fire will be to
initiate a jet fire or, much less likely, a pool fire.
Dispersion distances for dense gases, such as LPG vapours, are strongly dependent on the
nature of the surface over which the gas disperses. The relevant parameter used in the models,
called the surface roughness length, can be estimated. A reasonable conservative value
relevant to typical LPG facilities is 0.1 m.
The vertical temperature gradient in the atmosphere has a considerable effect on gas
dispersion. Strong surface cooling, under clear skies at night, and a low wind produce stable
conditions. Weather conditions are denoted by letters A to F (after Pasquill - discussed in Ref.
4). The most stable conditions are denoted by the letter F. Neutral conditions, under cloudy
skies or in higher winds, are most common, and are given the letter D. Strong sunshine in the
daytime, with low winds, produces unstable temperature gradients, the most extreme being
denoted by A. In these circumstances, gas will dilute in a shorter distance than for D stability.
The largest dispersion distances are found in F stability weather. Calculations are normally
performed for 5 m/s wind speed and D stability (5D) to give typical average results. 2 m/s wind
and F stability (2F) represents a typical worst case.
3.4.1.1 Horizontal dispersion
The source for the dispersion calculations was assumed to be a flashing jet of pressurized
liquid propane or butane. No dilution due to the jet has been assumed. For horizontal jets
directed downwind, this dilution is counteracted by the fact that the jet pushes the gas
further downwind. For other jet orientations, the results are conservative (predicted
dispersion distances greater than actual).
Figure 3 presents distances to LFL for the 5D and 2F atmospheric conditions for a surface
roughness factor of 0.1 metre.
3.4.1.2 Vertical dispersion
Calculation methods for two-phase vertical releases are still being developed. In the
meantime the horizontal dispersion calculations described above may be taken as a worst
case analysis of the vertical dispersion case. Where this failure mode is critical,
PETRONAS should be asked to advise.
3.4.2 Fires
The radiation fields generated by two-phase Jets have been calculated with computer models
which have now been partially validated by large-scale jet fire trials. A typical worst case wind
speed of 5 m/s has been assumed in all calculations.
Results for varying mass flow rates are presented for butane in Figures 4 to 16. As propane is
less radiative than butane, these results may be used for propane also.
3.4.2.1 Horizontal jet fires
Curves showing the distance to critical flux levels for a range of mass flow rates are
presented in Figures 4 to 9. These give radiation fluxes at ground level. The flame shapes
and lengths are different for liquid and vapour fires. Two limiting curves are shown, the
upper for liquid and the lower for vapour fires. Depending on the precise ratio of liquid to
vapour, actual cases will lie between the two extremes. In the high momentum jet fires
which these Figures represent, jet momentum decreases with increasing distance along
the jet until at a given point buoyancy forces become dominant and the flame lifts off the
ground. This phenomenon enables an effective jet impingement distance to be defined.
Figure 10 shows the horizontally projected flame lengths calculated for the range of flow
rates.
3.4.2.2 Vertical jet fires
In general, vertical jet fires may be expected to have a smaller radiative impact than the
corresponding horizontal jet fire. A set of curves showing the downwind distance to critical
flux levels for vertical jet fires with a wind speed of 5 m/s is presented in Figures 11 to 16.
These do not give exactly the same figures as Tables 2 and 3 because there is no stack
pipe and the emission conditions are different.
4. WORKED EXAMPLE
In order to demonstrate the application of Sections 2 and 3 of the Guidelines, a worked
example of a study at an existing facility has been completed. This part of the Guidelines
follows the assumptions, considerations and calculations involved in that example.
4.1 DESCRIPTION OF FACILITY
The facility is an LPG depot which is supplied by barge via the local river. It comprises a
barge unloading berth, bulk storage for both commercial propane and commercial butane,
and bulk road vehicle loading. The unloading berth is separated from the remainder of the
facility by a motorway. Figures 17 and 18 show the depot layout and a simplified flow scheme,
respectively.
4.1.1 Barge unloading berth
The unloading berth includes two 100 mm dia loading arms, one for liquid service, the
other for vapour. Barge unloading is performed using a shore-based compressor. The
loading arms are hydraulically operated from a control station at the berth. Each arm
contains several swivel joints and the connection with the barge is via a flanged joint. The
arms are not equipped with a breakaway coupling or other quick release device.
On the shore side of the loading arms, the line from each arm bifurcates for either
propane or butane service. The vapour line is equipped with an hydraulically-operated
ball valve before the line bifurcates. This valve is actuated electrically from the
compressor station. The liquid line has no such valve.
As soon as each line bifurcates the four lines (i.e. propane liquid, butane liquid, propane
vapour, butane vapour) are each fitted with two flanged manually-operated ball valves.
Each liquid line is then fitted with a flanged non-return valve. There are several small bore
fittings for instruments, draining and hydrostatic pressure relief in this area.
Once the pipelines are clear of the valve locations, the two vapour lines are lagged to
prevent vapour condensation.
Firefighting facilities at the barge berth consist of dry chemical fire extinguishers. A
hydrant is located on top of the river bank some 30 metres from the loading arms. This
hydrant is partially obstructed from the berth by a building housing fire pumps which take
suction from the river. A fire alarm is located on the berth.
There is no emergency shutdown or leak detection system in place on the berth apart
from the remote operated valve on the vapour line.
4.1.2 Bulk storage
The vapour pipelines to the barge berth are routed via a compressor which takes suction
either from the vapour space on one of four butane spheres (each of 1 025 m
3
capacity)
or from the vapour space of one propane sphere (1 750 m
3
capacity). The liquid filling line
into the bottom of each butane sphere also doubles as a liquid withdrawal pipe. The liquid
filling line into the propane sphere is directed into the top of the vessel.
The four butane spheres each have two flanged connections directly under the vessel.
These connections are liquid inlet/outlet and drain. The liquid inlet/outlet is fitted with a
flanged hydraulically-operated ball valve followed by a Shand and Jurs hydraulic valve.
The valve nearer the sphere includes a fusible link which is designed to cause the valve
to fail closed under fire engulfment conditions. The drain connection is equipped with a
flanged manually-operated ball valve followed by a spring-loaded dead-man ball valve.
The drain line then extends beyond the periphery of the sphere. This line is lagged as is
the liquid inlet/ outlet line.
The top connections into each butane sphere comprise vapour line, pressure relief
valves, Whessoe contents gauge and maximum fill level float gauge. All connections are
flanged. The vapour line is equipped with an hydraulically-operated ball valve, again
actuated electrically from the compressor station. The Whessoe contents gauge is local
readout only. The maximum fill level float gauge is linked to the compressor station such
that in the event of overfill the compressor will cut out.
The propane sphere has a welded line from the base of the sphere to the primary valve
which is located at the edge of the bunded area some ten metres from the periphery of
the tank. The primary valve is flanged and is remotely operated. It is followed again by a
Shand and Jurs valve. A catchpot is located downstream from these valves and a lagged
drain line connected to the pot. The drain line is fitted with a manual valve followed by a
spring loaded valve.
The top nozzles into the propane sphere are as for the butane spheres plus the liquid
inlet line. The liquid inlet line is fitted with an hydraulically-operated ball valve with
electrical actuation from the compressor station as above.
All pressure relief valves discharge direct to atmosphere. There is no flare or vent system.
All spheres are located in separate low-bunded areas. All are equipped with a water
spray sprinkler system.
The nozzle details for the propane and the butane spheres are indicated in the attached
Figures 19 and 20.
4.1.3 Bulk road vehicle loading
Product stored at the depot is pumped into bulk road vehicles. The bulk loading facility
includes a swivel-jointed loading arm for liquid transfer. There is no vapour return line.
Connection with the road vehicle is made using an Acme coupling. The loading arms are
equipped with a manually-operated valve adjacent to the Acme coupling. There is no
breakaway coupling fitted or other driveaway protection/prevention. A remotely-operated
valve is installed in the line upstream from the loading am. This valve is operated from an
emergency stop located directly adjacent to the loading point. Loading is performed using
a preset turbine meter. The loading area is protected from vehicular damage by highway
guard-railing.
Firefighting facilities at the loading area consist of hand held fire extinguishers plus
adjacent fire hydrants. There is no sprinkler system nor are there fire water monitors.
4.1.4 Depot operation
Barge unloading is completed under the control of a shore-based supervisor in
conjunction with barge crew. The supervisor does not stay at the barge berth during the
operation; he is mostly at the compressor station which is located over 100 metres away
from the berth and not in direct line of sight. The supervisor checks the pipelines and the
berth about once per hour during unloading operations. The unloading rates are
approximately 180 m
3
/hour for propane and 140 m
3
/hour for butane. At the completion of
unloading, liquid lines are blown clear of liquid as far as is possible, using the plant
compressor.
The compressor station is equipped with a control panel which enables the supervisor to
set the remotely operated valves on the storage tanks and on the vapour line at the barge
berth. The supervisor thus controls transfer of product into the spheres. As stated above,
there is no remote readout of product level in each tank, only a high level overfill cutout.
Filling level is controlled by observing the local readout from the Whessoe gauges.
The bulk road vehicle loading operation is controlled by the vehicle driver. The depot
supervisor is usually not present during loading operations. The driver does not have
access to storage tank valves, only to valves in the delivery pipework. Loading rate is
approximately 20 m
3
/hour.
4.2 AUDIT OF THE FACILITY AND CHOICE OF SCENARIOS
The audit of the facility was completed by a team of four people including the depot
operations superintendent, a company safety adviser and two LPG engineering specialists.
The team spent a day on site gathering information and inspecting the layout, operation and
maintenance of the facility. Depot staff were able to advise on operations and maintenance
procedures and engineering staff provided layout drawings, flow schemes and technical data
(i.e. pump curves, pressure relief valve data, etc.).
Having assembled all relevant information the team then studied the facility to determine
those situations which might result in leakage of LPG. The approach adopted was to
commence with the barge berth and then follow the pipe track to the compressor station, into
the storage vessels, and then out to the bulk vehicle loading point. Using the procedure set
out in Section 2.1 the team identified a number of leakage scenarios which they considered to
be credible and/or the consequences potentially severe.
4.2.1 Barge berth
4.2.1.1 Scenario 1
There are no protective devices on the loading arms (e.g. breakaway couplings etc.) to
prevent damage or rupture in the event of excessive barge movement. It was therefore
decided that loading arm rupture should be considered. There are no emergency
shutdown valves on the barge so full bore continuous flow driven by the pressure in the
barge tank was adopted. There are similarly no emergency shutdown valves on the liquid
lines to the storage tanks on shore. Although the non-return valve in these lines
decreases the probability of liquid flow from the pipelines, it cannot be relied upon for
emergency shutdown purposes. The pipeline must therefore also be included as an
additional leakage source.
Scenario 1: Loading am rupture. Leak fed by barge storage and lines on shore.
4.2.1.2 Scenario 2
The barge berth area contains many flanged joints, the loading arms contain several
swivel joints, and there are numerous small bore connections in the area. The number of
joints and connections warrants the consideration of a flange or joint leak.
As the liquid line is blown clear after each delivery, there is no need to consider the
effects of barge impact during berthing.
Scenario 2: Flange, swivel joint, or small bore connection leak during barge
discharge.
4.2.1.3 Scenario 3
The pipelines from the barge berth to the storage vessels run parallel and adjacent to the
motorway for some distance before crossing under the motorway. The pipe track is lower
than the motorway and it is conceivable that either a vehicle or goods from a vehicle
could leave the motorway and impact upon the pipelines.
Scenario 3: Damage to pipelines due to vehicular or vehicular goods impact.
4.2.2 Pipe track
4.2.2.1 Scenario 4
The pipelines from the barge berth to the storage tanks are flanged along the entire
length. Where the pipelines enter the depot after crossing under the motorway there are a
number of redundant valves and connections which, as above, warrant the consideration
of a flange leak in this area. The pipelines enter the depot at the south west corner of the
property. A concrete block wall in excess of two metres high separates the depot from the
neighbouring property.
Scenario 4: Flange leak on pipe track at entry point into depot.
4.2.3 Storage vessels
4.2.3.1 Scenario 5
The pipelines at the barge berth and throughout the depot are not clearly marked to
indicate product carried. There is no interlock in the valving system to prevent propane
being delivered into the butane pipework. Given the above, it is quite conceivable that
propane could be delivered into the butane spheres. The butane spheres are not
designed for propane vapour pressure.
Scenario 5: Propane delivered into butane rated spheres.
4.2.3.2 Scenario 6
The butane spheres are fitted with flanges on the sphere side of the primary valve on the
liquid inlet/outlet and on the drain line. Given the inability to control a leak from these
flanges and the sphere inventory, leaks from these flanges must be considered.
Scenario 6: Leaks from flanged joints on butane spheres on sphere side of
primary valve.
4.2.3.3 Scenario 7
Based on Scenario 6, a fire fed from a flange leak underneath the butane spheres must
be considered. This fire may in turn lead to overpressurisation of the vessel and
consequent vapour discharge through the vessel relief valve(s).
Scenario 7: Vapour release from butane sphere pressure relief valve due to fire
engulfment.
4.2.3.4 Scenario 8
The Whessoe level gauges on the butane and propane spheres are only of the local
indicator type. They are not equipped with alarm or emergency shutdown features. The
lack of a remote readout at the control point in the compressor station is likely to lead to
an estimation of ullage by the depot supervisor and the possibility of error.
The maximum fill level gauge can only be tested by exceeding the safe filling level. The
gauge is not tested and its reliability is therefore questionable.
The level instrumentation as described above on both the butane and propane spheres is
inadequate. Overfill of the spheres must therefore be considered.
In the event of overfill, liquid will flow through the vapour suction line to the compressor.
The knockout drum adjacent to the compressor is not equipped with any liquid level
alarms or trips. Liquid could therefore enter the compressor giving a liquid stroke. This
would cause a major leak.
Scenario 8: Overfill of propane/butane spheres leading to major leak at
compressor.
4.2.3.5 Scenario 9
The propane sphere is fitted with a flanged connection on the liquid outlet line on the
sphere side of the primary valve at the edge of the bunded area. As for the butane case a
leak from this flange must be considered.
Scenario 9: Leak from flanged joint on propane sphere on sphere side of primary
valve.
4.2.3.6 Scenario 10
A fire fed from the flanged joint leak on the propane sphere in Scenario 9 may lead to
overpressurisation of the vessel and consequent vapour discharge through the vessel
relief valve(s).
Scenario 10: Vapour release from propane sphere pressure relief valve due to
fire engulfment.
4.2.4 Product transfer
4.2.4.1 Scenario 11
The pumps for transfer of product from the storage spheres to the bulk vehicle loading
point are located at the edge of the bunded area approximately 10 metres from the tank
shell. Each pump is fitted with a mechanical seal but not with a throttle bush. Total failure
of a pump seal is not a common event but neither is it rare enough to discount.
Scenario 11: Leak from pump seal due to total failure of seal.
4.2.4.2 Scenario 12
The compressor station is an area of complicated pipework again with many flanged
joints and small-bore connections.
Scenario 12: Flange or small-bore connection leak at compressor station.
4.2.5 Bulk road vehicle loading
4.2.5.1 Scenario 13
As for the barge berth there are no protective devices on the loading am at the bulk road
vehicle loading point to prevent damage or rupture in the event of a driveaway. Loading
am rupture is therefore included. The only emergency stop button in this area is located
adjacent to the loading point. It would probably be inaccessible in the event of loading am
rupture.
The bulk vehicles are equipped with emergency stop buttons on each corner of the tank
frame. However, given that a positive action is required to activate these buttons and that
such action is required adjacent to a large leak. the vehicle should still be considered as a
leakage source.
Scenario 13: Loading arm rupture. Leak fed by delivery pump and bulk road
vehicle storage tank.
4.2.5.2 Scenario 14
The loading area also contains many flanged and swivel joints and small-bore fittings.
Scenario 14: Flange, swivel joint or small bore connection leak at bulk road
vehicle loading point.
4.2.5.3 Scenario 15
The coupling between delivery pipework and vehicle is a vulnerable area in that the driver
may not make the coupling correctly or a sealing ring may be damaged or missing. The
coupling must be made at each delivery. The frequency of connection warrants the
inclusion of possible leakage.
Scenario 15: Leak from incorrect or damaged coupling.
4.3 CONSEQUENCE ASSESSMENT, ANALYSIS AND PROPOSED ACTION
In this section each of the leakage scenarios is considered in turn and a consequence
assessment is completed as described in Section 3. The acceptability of the consequences are
then considered and possible actions suggested to reduce the probability of the scenarios,
reduce their impact by minimising the leakage rate, and/or provide additional protection to
people, property and equipment.
The possible actions suggested for this example should not be seen as 'correct answers' for
these scenarios. Neither are they exhaustive. They are merely typical of the types of action
which might be relevant to a particular situation. Actions should be selected according to
specific site requirements.
With the exception of vapour release from pressure relief valves, the consequence
assessments have only been completed for leaks from liquid pipelines since these leaks give
more severe consequences than for the vapour case.
If the proposed actions for offending scenarios in liquid lines are not to be also applied for the
vapour cases then these cases must be considered in their own right.
4.3.1 Scenario 1: Barge berth loading arm rupture
Calculations have only been completed for the propane case since the vapour pressure,
release rate and consequent hazards are higher than for butane.
Leak = Two phase leak from 100 mm dia hole
From Section 3.2.2.2
Adopt L/D >12
P
c
= 0.55 P
o
For P
o
choose equilibrium vapour pressure at 15C =
9 x 10
5
N/m
2
(Ref Fig 01.02.08.01 of Manual /PTS)
P
c
= 0.55 x 9 x10
5
= 4.9 x 10
5
N/m2
Tc = 268K
m = 1 exp [
c
( T
1
T
c
)]
with c = 2 407 J/kgK (Ref Fig 01.02.11.01 of Manual /PTS)
and = 383 000 J/Kg (Ref Fig 01.02.10.01 of Manual /PTS)
m = 1 - exp
1
]
1
2 407
383 000
288 268 ( )
= 0.12
Mixture density
c
=
m m
g
+
_
,
1
1
1
c
=
012
93
088
535
1
.
.
.
+
_
,
(Ref Tables 01.02.04.01 and 01.02.0.03. of Manual /PTS.)
= 69kg/m
3
Critical flow rate
M = 0.6 A ) ( 2
c o c
P P
= 0.6 x
x
x x x
01
4
2 69 9 10 49 10
2
5 5
.
( . )
= 35.4 kg/sec
From Figure 3 estimate distance to LFL as follows:
5D conditions - 105 m
2F conditions - 190 m
Probability of early ignition is high. Vehicles on motorway most likely ignition source.
From Figures 4 to 16 distances to radiation flux levels from the ignited mixture are as follows:
Horizontal Jet Fires Vertical Jet Fires
1.5 kW/m 140 m 120 m
5 kW/m 110 m 65 m
13 kW/m 90 m 40 m
This scenario is clearly unacceptable, even though the leak considered above is fed by the
barge storage alone. The leakage rate would be much greater if the shore based pipelines
were also to contribute to the leak.
The consequences of this incident are so severe that the leakage rate and duration should be
reduced. The probability of the incident should also be reduced.
Possible means of achieving the above are:
Emergency release couplings on loading arms.
An inter-related system of remotely-operated fail-safe emergency shutdown valves at
the termination of the barge pipework and on the berth on the shore side of the loading
arms.
Inclusion of fusible links in emergency shutdown systems.
Supervision during entire unloading operation at the berth.
Improved fire protection/fire fighting and gas detection/ gas dispersion facilities.
Construction of a water curtain adjacent to the motorway.
4.3.2 Scenario 2: Flange., swivel joint, or small-bore connection leak at barge loading
berth
In this scenario, typically consider one segment of a flange gasket blown out. Based on a
number of pipe sizes from 50 mm to 300 mm diameter. the average leakage path area
equates to an effective hole diameter of 12 mm. A small-bore connection leak would give a
similar case. As above, the calculation is only completed for the propane case.
From Section 3.2.2.2
Adopt L/D = 0
M = 0.6A 2 ( ) P P
o a
= 0.6 x
x
x x x
0012
4
2 510 9 10 1 10
2
5 5
.
( )
= 2.0 kg/sec
From Figure 3 estimate distance to LFL as follows:
5D conditions - 25 m
2F conditions - 35 m.
The elevation of the barge berth is lower than the motorway and the berth is elevated above
the river. The above distances will therefore be conservative because LPG vapour is heavier
than air and the vapour will tend to fall to the river.
From Figs 4 to 16 distances to radiation flux levels from liquid fires are as follows:
Horizontal Jet Fires Vertical Jet Fires
1.5 kW/m 55 m 45 m
5 kW/m 40 m 30 m
8 kW/m 40 m 25 m
13 kW/m 35 m 20 m
The scenario is unacceptable. The radiation flux levels on the motorway particularly are
excessive.
The probability of leakage should be reduced.
The pipework in the barge berth area contains many joints and small bore connections.
Possible action to reduce the probability of leakage includes:
Rationalisation of pipework to reduce number of joints/ connections.
Replace flanged joints with welded joints.
In addition, possible action to prevent the vapour cloud reaching the motorway and/or to
protect the motorway from the effects of an ignited leak includes:
Construction of vapour barrier wall.
Installation of water curtain with or without automatic actuation triggered by gas detectors
on the berth.
4.3.3 Scenario 3: Damage to pipelines due to vehicular or vehicular goods impact
In this scenario the range of damage caused by impact could range from shearing the
pipelines to springing a flange. Typical leakage rates for these two extremes have already
been considered in the above two scenarios.
As both the above scenarios are unacceptable then this scenario is also unacceptable.
Possible actions are:
Install reinforced highway guard-railing adjacent to pipe track.
Bury pipelines in this area.
Install remote operated emergency shutdown valves at either end of vulnerable pipework.
4.3.4 Scenario 4: Flange leak on pipe track
The leakage rate in this scenario will be as for the similar case at the barge berth (i.e. M =
2.0 kg/sec). Where the pipelines enter the depot there is significant local confinement,
especially in a disused compressor shed. Confinement raises the possibility of vapour
cloud explosion in the event of ignition. The probability of ignition is high due to the
adjacent motorway and offices. This scenario is unacceptable.
Possible actions:
Reduce probability of leakage as for Scenario 2.
Examine need for disused building and demolish if possible and/or investigate other
means to reduce confinement.
4.3.5 Scenario 5: Propane delivered into butane rated spheres
In order to calculate the vapour pressure of a mixture created by delivering propane into
the butane spheres, a worst case of assuming 100 per cent propane at a receipt
temperature of 25C into an empty sphere has been adopted.
The vapour pressure of propane at 25C is 11.5 x 10
5
N/m which is higher than the
design pressure of the butane sphere of 7.3 x 10
5
N/m . This is also the start-to-
discharge pressure of the relief valves. The fully open pressure (accumulated pressure) is
8.0 x 10
5
N/m.
The scenario is unacceptable. If product contamination is a credible scenario, then the
relief valves should be sized to relieve sufficient propane vapour to avoid over-
pressurization of the vessel.
Possible actions to prevent product contamination are:
Clear and positive identification of pipelines throughout the depot and particularly at the
barge berth.
Introduction of a valve interlock system to ensure only one set of valves is open at the
barge berth.
Installation of in-line densitometers linked to an alarm/ shutdown system to detect
incorrect product in pipelines.
4.3.6 Scenario 6: Leaks from flanged Joints on butane spheres on sphere side of primary
valve
The leakage in this scenario will again be based on leakage from an equivalent hole
diameter of 12 mm.
From Scenario 2
M = 0.6A 2 ( ) P P
o a
= 0.6 x
x
x x x
0 012
4
2 577 31 10 1 10
2
5 5
.
( . )
= 1.1 kg/sec
From Figure 3 estimate distance to LFL as follows:
5D conditions - 15 m
2F conditions - 25 m
From Figs 4, 5, 6, 7, 9 distances to radiation flux levels from liquid fires are:
Horizontal Jet Fires
1.5 kW/m 40 m
5 kW/m 30 m
8 kW/m 25 m
13 kW/m 25 m
44 kW/m 20 m
Vertical jet fires will impinge on the vessel shell in all cases. No radiation distances are
therefore given.
In the event of ignition the radiation effects on the sphere and on adjacent spheres exceed 44
kW/m and are therefore unacceptable.
Possible actions to reduce the probability of this scenario are:
Modification to sphere pipework and provision of correct valve to remove flanged joint
upstream of primary valve.
4.3.7 Scenario 7: Vapour release from butane sphere pressure relief valve
From Section 3.2.1, for fire-engulfed butane tanks
W =
367
KAP
kg/sec
Spheres are equipped with two pressure relief valves although only one valve is lined up at
any time. The valves are labelled as 6R10.
From Table 1c for R orifice,
Discharge area = 103.23 x 10
-4
m
W =
09 103 23 10 73 12 10
367
4 5
. . . . x x x x x
= 22.2 kg/sec.
Section 3.3.1.1 describes that flammable vapour plumes from vertical relief valves may be
assumed not to reach ground level. The distance to LFL is therefore not applicable in this
instance.
From Table 3, distance downwind to radiation flux levels are:
Tank Top Radiation Flux
8 kW/m - 45m
44 kW/m - -
Ground Level Radiation Flux
1.5 kW/m - 95 m
5 kW/m - 40 m
13 kW/m - -
The above flux levels indicate that the 5 kW/m radiation contour crosses the site boundary.
Whilst this boundary should probably be described as an urban area, the region adjacent to
the boundary is not developed and a higher radiation intensity could be tolerated. Given that
the flux level at the plant boundary is not much greater than 5 kW/m, this scenario is
considered to be acceptable and the existing location of the butane sphere closest to the
plant boundary does not warrant any action. Any future development outside the plant should
be monitored as construction of a hospital, school or other facility difficult to evacuate at short
notice would require a reappraisal of the situation.
4.3.8 Scenario 8: Overfill of propane/butane spheres
Overfill of the propane or butane spheres will not result in product leakage from the
spheres but will result in liquid flow through the suction line to the compressor.
A liquid stroke in the compressor is unacceptable.
Possible actions are:
Installation of remote readout level gauge in transfer control area.
Installation of level gauge capable of registering low level, high level and high-high level,
with alarms and inlet valve shutdown devices attached. Gauges should permit regular
testing to ensure satisfactory operation.
Regular testing of maximum fill level gauge.
Installation of liquid level emergency alarm on knockout drum with compressor trip.
4.3.9 Scenario 9: Leak from flanged joint on propane sphere on sphere side of primary
valve
This scenario as per Scenario 2:
M = 2.0 kg/sec.
Distances to LFL:
5D conditions - 25 m
2F conditions - 35 m
Distances to radiation flux levels from liquid fires are:
Horizontal Jet Fires Vertical Jet Fires
1.5 kW/m
5kW/m
8 kW/m
13 kW/m
44 kW/m
55 m
40 m
40 m
35 m
30 m
45 m
30 m
25 m
20 m
10 m
The radiation flux level on the propane sphere in the event of an ignited leak is excessive,
although the location of the primary valve remote from the sphere mitigates the impact on the
sphere. Possible actions are as for Scenario 6.
4.3.10 Scenario 10: Vapour release from propane sphere pressure relief valve due to fire
engulfment
From Section 3.2.1 for fire engulfed propane tanks:
W =
419
KAP
kg/sec
The propane sphere is equipped as per the butane spheres with two pressure relief valves
although only one is lined up at any time. The valve is labelled 6R10.
From Table 1C for R orifice:
Discharge area = 103.23 x 10
-4
m
W =
09 103 23 10 155 12 10
419
4 5
. . . . x x x x x
= 41.2 kg/sec
As for the butane case, the distance to LFL is not applicable as the flammable plume may be
assumed not to reach ground level.
From Table 2, distance downwind to radiation flux levels are:
Tank top Radiation flux
8 kW/m - 50 m
44 kW/m - -
Ground level radiation flux:
1.5 kW/m - 110
5 kW/m - 40
13 kW/m - -
As for the butane case, the 5 KW/m
2
radiation contour crosses the site boundary, The figures
are similar for both the propone and butane case and the points set out for the butane case are
therefore also applicable to this case.
4.3.11 Scenario 11: Leak from pump seal due to total failure of seal
Adopt L/D = 0
For pumps without throttle bushing, adopt effective diameter hole =17mm
M = 0.6A 2 ( ) P P )
o a
= 0.6 x
x
x x x
0076
4
2 69 9 10 49 10
2
5 5
.
( . )
= 20.5 kg/sec
From Figure 3 estimate distance to LFL:
5D conditions - 75 m
2F conditions - 140 m
Probability of ignition is high as LFL extends well beyond site boundary and into housing area
on eastern side of depot.
From Figures 4 to 16 distances to radiation flux levels from the ignited mixture are:
Horizontal Jet Fires Vertical Jet Fires
1.5 kW/m
5 kW/m
13 kW/m
110 m
85 m
70 m
95 m
55 m
30 m
This scenario is unacceptable. As for Scenario 1, the probability of the incident and also the
leakage rate from the incident should be reduced.
Possible means of achieving the above are:
Breakaway couplings.
Driveaway prevention device(s) on the bulk road vehicle or installed as part of fixed
facility (e.g. boom).
Remote-operated emergency shutdown system capable of operating valves on both
delivery pipeline and bulk road vehicle.
Fusible links in emergency shutdown system.
Increased supervision during loading.
Improved fire-fighting/fire protection and gas detection/ dispersion facilities (e.g.
automatic sprinkler system over loading bay).
4.3.14 Scenario 14: Flange, swivel joint, or small bore connection leak at bulk vehicle
loading point
This consequences of this scenario are similar to those of Scenario 2.
The distances to LFL for this scenario maintain the flammable vapour cloud from a
leakage mostly within the site boundary. The probability of ignition is therefore low.
However, in the event of ignition, the radiation flux on the store to the west is
unacceptable.
Possible actions include:
Reduce probability of leakage as per Scenario 2.
Install improved fire-fighting/fire protection and gas detection/gas dispersion facilities.
Construct fire wall on store wall adjacent to loading point.
Install water curtain on store wall.
4.3.15 Scenario 15 : Leak from incorrect or damaged coupling at bulk road vehicle loading point
Assume equivalent hole diameter = 25 mm
Adopt L/D = 0
M = 0.6A 2 ( ) P P
o a