You are on page 1of 24

I. CHEM. E. SYMPOSIUM SERIES NO.

85

THE DESIGN AND OPERATION OF OFFSHORE RELIEF SYSTEMS


F.K. Crawley* and D.S. Scott*

The design of pressure relief systems for offshore operation


provides a demanding challenge to the process designer, which if
poorly handled will cause numerous operating difficulties.
This paper therefore consists of two parts. The first, which
deals with the design of relief systems, will discuss the sizing
and routing of relief piping and includes two phase flow,
estimation of relief rates and alternatives to venting.
Items
of particular importance to the offshore industry (such as total
system weight and flare location) will be examined briefly and
contrasted with onshore practice.
The second part of the paper will examine problems encountered
in the operation of relief systems and includes flare radiation
levels, segregation of flare systems and, of particular
importance, drainage and cross communication in vent systems.
The paper aims to provide a practical basis for the design of
relief systems (equally valid for offshore, and onshore
equipment) and suggests a number of improvements to existing
design practice.
INTRODUCTION
Pressure relief systems are one of the most important and difficult to
design piping systems on any process plant - onshore or offshore.
The design of relief and blowdown facilities usually takes place toward
the end of the process design phase when any flexibility within the
design has been limited.
This is unfortunate, but probably
inevitable, as process flow sheets, control valve sizes, vessel
dimensions and many other factors must be agreed before detailed design
of the vent system can proceed. Few other systems require the same
detailed attention and whenever possible the routing of the vent pipework
should be considered at an early stage, even if pipe sizes are unknown.
This can ease, or possibly prevent, layout problems at a later stage.
By considering a number of points as early in the design process as
possible, later problems can be minimised.
The most important points
are:

The relief rates from each source.


Composition, temperature and pressure of the relieved material.
Routing of the relieved material to flare (or other means of
disposal).
Radiation levels from the flare.

These factors, and others, will be examined later, first however, it


is worth comparing some aspects of onshore process plant with offshore
practice.
* Britoil plc,

150 St. Vincent St.,

291

Glasgow.

G2 5LJ

I. CHEM. E. SYMPOSIUM SERIES NO. 85

COMPARISON OF ONSHORE AND OFFSHORE PRACTICE


At the present time the only process commonly found offshore is oil and
gas production from fixed, as opposed to floating, installations.
A
typical flowsheet for oil production is found in figure 1.
Oil
production from floating facilities and chemical production from barge
mounted plant are being considered,
but are not yet widespread.
In
addition to the challenges inherent in the design of any major industrial
plant a number of particular problems exist offshore and are highlighted
in table 1.

DESIGN OF RELIEF SYSTEMS


Estimation of Relief Rates
There are a relatively small number of overpressure causes present in
offshore facilities: these are generally covered, sometimes in a
superficial manner, in API-521 (1) which is usually taken as the standard
reference for pressure relief systems, even outside the oil and
petrochemical industry.
The major areas of concern are:

'Blowby' or 'Blowthrough'
Fire
Thermal expansion
Heat exchanger tube failure
Pump or compressor deadhead
Mal-operation

These are examined below.


It should be noted that runaway chemical
reaction, which is a difficult relief case to analyse in chemical plant,
does not occur offshore.
Additional relief cases may be identified
during hazard and operability studies, safety reviews or other procedures.
A typical example would be failure of control valves on service systems.
For example failure in the open position of a heating medium control
valve may allow a high flow of hot fluid into heat exchanger causing
thermal expansion or boiling of the process fluid.

'Blowby' or 'Blowthrough'
As all but the last separator in a production train operate at elevated
pressure, loss of the liquid level will allow gas at high pressure to
enter downstream vessels which are designed to operate at low pressure.
This is termed 'blowby' or 'blowthrough'.
Low level trips in each
vessel should operate to stop production and close emergency isolation
valves, if the trip system fails relief valves will provide the last
line of defence.
The relief rate required can be calculated from the
control valve size and the pressures upstream and downstream of the
relief valve.
The downstream pressure will be either atmospheric, if
the downstream equipment vents directly to atmosphere, or the set
pressure of the relief device in question.

292

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Wellhead fluid is usually at high pressure and is letdown, or choked, to


a relatively low pressure before processing. Elaborate instrument
systems are generally used to close off the wellheads upstream of the
choke valves in the event of a serious process upset.
It is not normal
practice to design the flare system to cope with a simultaneous failure
of a number of choke valves which, due to the use of high integrity
protective systems, is a very remote possibility.
A more credible event is the failure of a level control valve in a
gas/water/oil separator.
There is an initial temptation to take the relief valve set pressure,
plus accumulation of the separator relief valves as the upstream
pressure.
However, this may result in extremely large relief valves and
header sizes. As separators usually have a high pressure alarm and trip
plus one and usually two high level trips, it is unlikely that level will
be lost at the same time as the pressure trips do not operate.
Therefore it is usually permissible to use the pressure trip setting for
the upstream pressure in the calculation.
If necessary, a fault tree
can be constructed and quantified to justify the choice of upstream
pressure.
Fire
One of three distinct situations will exist when fire impinges on a
vessel heating the contents and causing an increase in internal
pressure. Each situation described below requires a different approach
to relief valve sizing.
.

Gas filled Vessels


There are a number of vessels on any platform where a liquid phase is
unlikely to be present, for example instrument air receivers and fuel
gas heaters.
In a fire, overpressure will be caused entirely by
thermal expansion of gas.
A relief valve for this duty can be sized from (2):
A = 1.824 103

F ' A3
...(1)
P1

293

A =

effective discharge
area of the relief valve
(M2)

F =

factor related to gas


temperature (which can be
determined from fig. 2).

A3 =

exposed surface area


(M2).

P1 =

upstream relieving
pressure
in Bar(a), (e.g.
atmospheric pressure plus
relief valve set pressure,
plus overpressure).

I. CHEM. E. SYMPOSIUM SERIES NO. 85

The use of this equation ignores the fact that as the metal wall of
the vessel heats up the yield stress of the wall reduces ultimately
reaching a point where the wall will rupture below the relief valve
set pressure (2,3). It is therefore unwise to rely on relief valves
as a protection against fire in a gas filled system, depressuring
the vessel via a control valve is generally preferable and is
discussed further below.

Vessels containing liquid with a vapour space


Calculation of relief rate in this situation is straight forward
using the method outlined in API-520 and API-2000. In general there
is little difference between the methods which relate heat input to
wetted surface area, however API-2000 is more conservative for lower
wetted surface areas. A comparison of the two is shown in figure 3.
There is growing evidence to suggest that heat input rates in large
fires may be more than 100% in excess of the figures given by either
API 520 and API 2000 (4,5).
It should be noted that crude oil is not a single component substance
and the vessel contents will fractionate during a fire, resulting in
a variable boil off rate. The highest boil-off rate usually occurs
at low temperatures when low molecular weight materials are present,
however it is prudent to calculate boil off rates at higher
temperatures particularly near the critical point where the enthalpy
of liquid and vapour phases may be similar. Again the dangers of
overheating the metal walls of pressure vessels should be examined
and the alternative of depressuring considered.

Liquid filled vessels


Behaviour of the contents of a liquid filled vessel under fire
conditions is difficult to predict and the commonly applied (6)
method of sizing the relief valve to pass a volume of liquid
equivalent to the vapour generated is probably conservative.
A more realistic approach where relief valves are installed on the
top of vessels is to consider three distinct stages:1. thermal expansion of the fluid up to the boiling point or bubble
point.
2. boil-off of vapour with entrained liquid until a vapour cap is
formed, which allows liquid droplet disengagement.
3. continuous vapour relief.
The required relief rate for stage 1 can be calculated from the
thermal expansion of the liquid and vapour relief rates for stages 2
and 3 can be calculated in the same manner as above. The major
problem is the estimation of the liquid entrainment in stage 2, in
trial calculations this has been assumed to be the same as the vapour
flowrate in mass terms. The relief valve is then sized on the
largest of the three orifice areas calculated. Usually the only
liquid filled vessel on the platform is the crude pig launcher which is
isolated and depressured when not in use.

294

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Estimation of orifice areas for relief valves on two phase flow


duties presents difficulties. Most relief valve manufacturers
suggest that separate calculations be done for the vapour and liquid
components of the flow and the separate orifice areas added together.
This is commonly felt to under-estimate the required orifice area as
it does not allow for flashing through the valve or differences in
sonic velocity between a vapour and a two-phase mixture (7,8).

Thermal Expansion

Protection of liquid filled vessels and piping against thermal


expansion of the contents from high ambient temperature or heat
tracing is often necessary.
Frequently small thermal relief valves
which do not comply with API-526 (9) are installed without estimates
of the required capacity being made.
This is unfortunate as relief
valve capacity is easily calculated from (10):

Where V =

flow rate at the flowing


temperature M3/S.

B =

cubic expansion coefficient


for the liquid at the
temperature under
consideration.

H =

total heat transfer


rate (kw).

G =

specific gravity related


to water at 15C.

C =

specific heat of the trapped


fluid in (KJ/kgK).

A particularly severe problem may occur in heat exchangers if the


cold fluid is isolated, with the hot fluid flowing.
Temperature
rise may be very rapid and if the bubble point of the cold fluid is
exceeded boiling will occur.
Theoretical calculations can be
performed to estimate the relief rate, which will occur in a similar
fashion to fire relief of liquid full vessels, but usually at a much
higher rate; in some cases, particularly with plate type exchangers,
the pressure drop imposed by the exchanger may cause unstable
operation of the relief valve.
In this situation rigorously
enforced operating procedures and valve interlocks are a better form
of protection than relief valves.

295

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Burst Tubes
Although plate heat exchangers are preferred offshore for reasons of
space and weight savings, shell and tube type exchangers are still
required for high pressure duties on gas compression and fuel gas
systems. Almost invariably gas flow is on the tube side with
cooling on the shell side.
In common with accepted practice (11) relief protection against a
burst tube is only fitted if the hydrostatic test pressure of the
lower pressure side of the exchanger can be exceeded. Unfortunately
this logic only applies to the first stage or first two stages of a
compression train and overpressure protection is required for the
high pressure stages. Here bursting discs rather than relief valves
are used due to the high venting rate required. Reverse buckling
discs are favoured for this duty as they are thicker and hence less
likely to pinhole or rupture spuriously than conventional or
composite slotted discs. Bursting disc sizing methods are less
standardised than relief valves and disc manufacturers should be
approached for advice. Note that obstructions in the pipework such
as knife blades or catcher bars will be present and the line size may
have to be increased to ensure that sufficient flow area is
available (12).
If thermal expansion of the cooling medium occurs it may be relieved
by a partial buckling of the bursting disc, which will either cause
spurious rupture or, more probably, an increase in the pressure which
may be attained before the disc ruptures. For this reason a
separate thermal relief valve should be installed on the shell side
with a set pressure plus overpressure below the lower burst tolerance
of the disc. Note that the pressure rating of the bursting disc
must allow for back pressure on the downstream side of the disc.

Pump and Compressor Deadhead


Mal-operation may result in isolation valves on pump and compressor
discharge lines being inadvertantly closed, allowing equipment to
'run-up' its curve. Downstream piping or equipment with a design
pressure below the deadhead pressure must therefore be protected.
The relief rate can be estimated, as shown in figure 3, by finding
the pump capacity at the relief valve set pressure. It should be
noted that relief valves on liquid duties may have to be set below
the vessel (or piping) design pressure if the valve overpressure
exceeds 10% of relief valve set pressure. When estimating relief
rates for compressors it should be noted that variations in molecular
weight of the compressed gas will affect the discharge pressure.

296

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Mai-operation
Operator error may cause any of the situations discussed above, with
the possible exception of fire.
Probably the most common examples
of operator error are the unauthorised opening and closing of
isolation valves or incorrect adjustment of control valve set
points. For this reason essential valves are often locked or
interlocked and their operation governed by a permit to work
system (12).
Where valves are accidently opened or closed problems may result from
the overpressure of a lower pressure vessel from higher pressure
source.
The relief rate required is estimated by locating the flow
limiting component of the piping system (usually a globe valve,
control valve or an orifice plate) and then calculating the flow
coincident with the highest postulated upstream pressure and lowest
downstream pressure.

ALTERNATIVES TO PRESSURE RELIEF VALVES


In some situations, both onshore and offshore, the installation of
relief valves and their associated pipework may prove unrealistic either
for economic reasons or due to the difficulty of disposing of large
quantities of relieved material. Here alternative solutions must be
found with a reliability equal to, or greater than, a relief valve.

Instrumented Trip Systems


Trip systems to close emergency isolation valves or to shut down pressure
producing equipment are commonly found on any offshore facility as part
of the Emergency Shutdown System (ESD).
Instrumented systems of this
type require careful design, components liable to failure have to be
duplicated or perhaps triplicated; voting systems may be necessary to
prevent frequent spurious operation of the trip (13).
Proof test
procedures and test intervals must also be defined at the design stage,
and the proof test must cover as much of the trip system as possible.
Frequently, great attention is paid to testing and self checking of the
trip controller and comparatively little to the less reliable sensor
element and actuator mechanism.
Another popular misconception is that
reliability of a trip can be increased to any desirable level by
increasing the frequency of testing.
As the frequency of testing
increases the proportion of time the trip is unavailable increases and
hence, during this period, a demand on the trip will not initiate the
remedial action required (14, 15).

297

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Depressuring Systems
The advantages of depressuring systems for the protection of vessels
exposed to fire have been discussed above; similar systems are
frequently installed on the platform compression train and equipment is
depressurised as a precautionary measure under certain upset conditions.
Operating pressures in the compression train will normally range from
atmospheric up to perhaps 150 bar g and frequently several vessels with
different design pressures will discharge to the same header.
It is
therefore essential to size the depressuring valves to prevent excessive
flow and hence excessive backpressure in the header.
High uncontrolled
flows can cause a number of problems (16, 17):

Excessive cooling, causing materials of construction and pipe


stressing problems.
Large flaring rates with attendant thermal radiation problems.
Noise caused by sonic velocity through the depressuring valve.
Reverse flow from high pressure sources into low pressure vessels.

Design of depressuring facilities is a complex task, based on an initial


specification to reduce pressure to a given level within a set time.
The criteria proposed by API (18, 19) may be inadequate for large
hydrocarbon fires where metal in contact with the vapour space may yield
after 5 - 1 0 minutes exposure (4).

PIPING DESIGN
On oil platforms all vented material, except that from low pressure small
volume vents, is fed to a flare system.
On gas platforms, gas is not
released during normal process operations and flammable material is
'cold' vented rather than flared during process upsets.
The discussion
of flare systems below therefore relates to oil production platforms.
Most platforms have separate high and low pressure headers to prevent
vessels with a low design pressure being subject to excessive
backpressure.
As most present generation platforms use low emissivity
type flares, which require significant pressure at the flare tip, it is
essential to ensure that vessels are connected to the correct header.
Another important segregation, in high pressure headers, is between
(water) wet and cold streams.
Wet streams usually originate in the
gas/oil separators or the cooling medium side of gas compression
coolers.
Pressure control valves in the separation train will also
contaminate the flare header with wet gas in normal operation.
Due to
the thermodynamic properties of hydrocarbon gases, very low temperatures
may result when they are relieved or blow down from high pressures.
The
mixing of cold and wet streams in pipework may therefore cause problems
of blockage by ice or hydrate formation.
Ideally, segregation should
occur with cold and wet streams being run in separate headers.

298

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Sizing of Relief Headers


The correct sizing of relief piping particularly downstream of the relief
valve or other relief device requires more care than almost any other
section of piping on an offshore facility.
A number of apparently
conflicting requirements have to be reconciled until an acceptable
compromise is reached.
This process is best achieved by following a
rigorous stepwise procedure as show below.

Step 1

Decide which combinations of relief valves will operate


simultaneously; realistic estimates, not worst assumptions are
required. For example, it is reasonable to expect all the fire
relief valves from one module to operate together, it is not
reasonable to expect every fire relief valve on the platform to
operate at once.

Step 2

Make an approximate estimate of the relief line sizes.

Step 3

Calculate the back pressure that will exist at the discharge of


each relief valve for the combination in question.

Step 4

Check that the backpressure imposed on the relief valves does


not exceed 10% of relief valve set pressure (RVSP) for
conventional valves or 50% RVSP for balanced bellows valves (20).
(Note that some relief valve vendors are unhappy with back
pressures as high as 50% RVSP on balance bellows relief valves
and may recommend lower figures).
It is generally advisable
to specify a maximum velocity of approximately Mach 0.3 to limit
noise.

Step 5

Repeat steps 2 to 4 until all line sizes are satisfactory.

Calculation of back pressure in relief valves is most conveniently


performed by flashing all the relief streams involved to a known
downstream pressure at the discharge of the header.
This will be
atmospheric pressure in the case of a 'cold' vent or the pressure
required for efficient operation of the flare tip. Where low emissivity
flare tips are installed the pressure is a function of relief rate.
At this downstream pressure physical properties can be estimated and used
to calculate the pressure drop and hence the upstream pressure for the
section of pipe in question.
The upstream pressure then forms the basis
of another flash calculation and the whole process is repeated until the
entire header has been examined.
More sophisticated line sizing formulae are required for relief systems
than are normally used for pressure drop calculations.
In particular
use of the Fanning equation with average vapour properties is not
applicable as this only calculates the frictional pressure drop. The
other components of the total pressure drop (gravitational and
accelerational) must also be considered, in many relief situations the
accelerational pressure loss predominates (21).
The complexities of
relief line sizing requires the use of computer software, preferably with
a physical properties generator, for complete solution.

299

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Manual calculations may be performed if required and in simple cases may


be adequate.
They should always err on the side of caution; for
example, isothermal and not adiabatic pressure drop calculations should
be performed.
Lapple's method for calculating pressure drop (which is
recommended in API-521) is not an accurate method for sizing relief
headers, particularly where changes in pipe diameter occur, and it should
be avoided if possible (22).
Special problems arise where two, or in some cases three, phase flow
(vapour, hydrocarbon liquid and water) is present.
Here vaporisation of
the liquid stream or retrograde condensation of the vapour may occur.
A
number of short cut methods are available (23, 24), but rigorous solution
is only practical with a computer.
Equilibrium conditions should be
used to predict vapour fractions.
This will produce a conservative
result as there is evidence to suggest that under conditions of rapid
depressuring a metastable condition exists which inhibits complete
formation of the equilibrium vapour fraction.

Sizing of Relief Valve Inlet Lines


The pressure drop between the relief valve and the item being protected
must be limited to a maximum of 3% of the relief valve set pressure, and
preferably less.
Higher pressure drops will effect the lift and
blowdown characterisitics causing 'chatter' which will damage the
valve (20, 25).

Sizing of Depressuring Lines


In many cases depressuring valves are piped into the relief header: this
may be termed 'uncontrolled depressuring'.
It requires some additional
calculation to ensure that backpressures at the start of depressuring do
not affect operation of relief valves which may be simultaneously
discharging. Similarly when the pressure in the depressured vessels is
low, relief valve operation should not interfere with the satisfactory
completion of the depressuring operation.
Where weight limitations pose a problem there are advantages in running a
completely separate depressuring header feeding directly into the flare
drum or other disposal unit, with a pressure control valve at the
downstream end.
By maintaining the gas flow at high pressure upstream
of the flare knockout drum the volumetric flow rate can be controlled,
resulting in a more constant flow of gas; with careful sizing the time to
depressure equipment will not be increased.
The concept is shown
graphically in figure 4 (26).
By keeping pressure in the discharge header relatively high, the
depressured gas density will also be high and volumetric flow rates will
be low, compared to those at atmospheric conditions.
This can lead to
significant savings in piping cost and weight.

300

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Obviously the control system must be designed with care to ensure its
reliability, and it is usually desirable to provide additional equipment
for online testing as an integral part of the design.
This system is
probably best suited to depressuring groups of vessels of similar design
pressure where one vessel cannot overpressure another if the control
valve fails closed.

Layout of Piping
All relief piping should be designed to minimize pressure drop by the use
of long radius bends, swept tees and gradual changes in diameter.
Although it is undesirable to install isolation valves in relief lines it
may be necessary in certain situations (see below). A frequent mistake
here is the use of standard pattern ball valves: these have a reduced
flow area through the ball which may have a significant effect on
pressure drop. Full bore ball valves must always be specified in vent
lines.
Mach numbers should be limited to 0.3 or below to avoid noise problems
and limit the possibility of sonic velocity at elbows and other fittings
where the flow area may be reduced.
Relief valves are often fitted with ASA 150 flanges on the discharge
nozzle and if significant backpressure can be generated the pipe flange
rating and wall thickness must be checked to ensure suitability.
All tail pipes should feed into the top of the header to ensure that
liquid relief from upstream valves do not flood the tail pipes of
downstream relief valves.

Discharge piping should be self draining and have a fall in the direction
of flow (27). On production platforms, where equipment is installed on
several levels, this usually results in the flare knockout drum being
located at the cellar deck level. Occasionally low points are required
in the header and, as a last resort, if they are unavoidable, must be
provided complete with drainage and heat tracing facilities. A vertical
leg or 'boot' should be provided to store liquid without reducing the
flow area. Drainage from the boot should be controlled by an
instrumented level control system, float valves are unreliable and
should be avoided.

Location of the Flare


Flare location is a major problem for platform designers.
The flare
must be located in such a position that heat radiation will not pose a
hazard to equipment, personnel or helicopters.
Older platforms have
used pipe flares either free standing above the production modules or
supported off the side of the platform.
In some cases more than one
flare stack is provided, the stack in use at any time being dictated by
wind speed and direction.

301

I. CHEM. E. SYMPOSIUM SERIES NO. 85

More recent platforms use low emissivity type flares with high gas
velocities at the flare tip, resulting in improved gas-air mixing,
combustion and improved heat radiation levels (29).
The heat envelope from the flare can affect radio communications to the
platform and this produces an extra constraint on flare location (28).

OPERATION OF RELIEF SYSTEMS


The operation of offshore relief systems differs in many respects from
those onshore.
Differences occur for economic and strategic reasons in
addition to these cited in table 1.

Isolation of Relief Valves


In onshore installations duplicated relief valves or bursting discs are
commonly installed; there is little cost implication and weight
limitations do not exist.
Offshore, the situation is very different;
duplicated relief systems are a weight penalty and are not installed
unless operating advantages can be demonstrated.
In practice duplicated
relief devices are installed in locations where isolation and removal of
relief devices would cause a process shutdown.
In other locations
single relief devices are provided with locked - not carsealed - open
isolation valves.
Where duplicated systems are installed a system of lock and key
assemblies are used in preference to interference discs, three way valves
or ganged valves.
Their main advantage, apart from simplicity, is that
they present no obstruction to flow; with ganged valves for instance,
there is the possibility that the mechanism will jam at some midpoint,
obstructing both relief routes.
On oil production platforms crude oil
is the major revenue earning product, the associated gas represents the
icing on the cake and can if necessary be sacrificed.
In any event it
is standard operating practice to flare part of the produced gas, as a
means of pressure control.
Passing relief valves are of relatively minor concern.
Fluids leaking
past relief and depressuring valves will collect in the flare knock out
drum and will be recycled to the crude production train.
Ignition and pilot gas supplies
Offshore platforms rarely have the luxury of a guaranteed gas supply for
the flare pilots.
Gas is available from the oil production train and
hence a shut down of oil production will remove the gas supply.
This
system however is of no use during commissioning or start up after a
prolonged break in production.
Bottled gas is required for these
situations and careful assessment is required to ensure that sufficient
but not excessive supplies of bottled gas are available.
Problems with
condensation in the fuel gas line are not uncommon; therefore, heat
tracing and/or some form of pretreatment is not unusual.

302

I. CHEM. E. SYMPOSIUM SERIES NO. 85


Snuffing
The worst conceivable event on a platform is a 'blowout' or uncontrolled
release of hydrocarbons to atmosphere from a well. Fortunately these
are rare events, the best known cases in the North Sea being on Ecofisk B
and more recently Forties D. When a blowout does occur large quantities
of flammable material are released to the atmosphere and a serious risk
of fire and explosion exists.
Blowouts generally give some warning of their occurrance and this
'breathing space' may be used to try and eliminate sources of ignition,
among which the flare features prominently.
Consideration is therefore
often given to flare snuffing by an immediate trip of oil production and
the injection of large quantities of halogenated hydrocarbons ('halons')
into the flare.
The usefulness of this is debatable, as experience
suggests that many blowouts will ignite from sources of ignition other
than the flare.
Purging
During periods where there is no flow in the flare system, there is a
possibility that air may enter the flare either by contraction or
condensation of flare gas causing a volume shrinkage; or by diffusion of
air down the flare from the flare tip.
To prevent this a flow of purge
gas is required,
for short durations fuel gas may be used but during
commissioning or after a prolonged shutdown an inert gas usually nitrogen
should be used.
Husa (30) has developed the following equation for
estimating purge rates on pipe flares up to 48" diameter.
This may
over-estimate the requirement for low emissivity type flares where the
cross-sectional area at the tip is less than the pipework flow area.
However, Husa's work suggests that anti-diffusion devices, such as
fluidic seals, are of very limited value and a conservative calculation
of purge rate is advisable until more information is available.

Where

(M 3 /S)

q =

Purge rate

D =

stack dia (M).

Ci =

volume fraction of gas

Ki =

constant (values for some common


gases are given in table 2).

Whenever maintenance work is necessary purging with an inert gas is


required before breaking into the flare header; purging should continue
after the flare has been extinguished and the purge gas and pipework has
cooled down, this will minimise buoyancy effects (31).
Similarly
purging with an inert gas is required before the system is recommissioned.

303

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Flare radiation
For many years flare radiation calculations have been based on API-521
which advocates a point source formula developed by Hajek and Ludwig (32):

Where

minimum distance from mid point


of flare to object being
considered (M)

fraction of heat radiated;


(this is often mistakenly
called the emissivity)

Heat released (kw)

allowable radiation (kw/M 2 )

The F factor is thought to be dependent on the gas moledular weight


and/or the flow aerodynamics, quoted values range from 0.2 to 0.5.
This relationship was developed for use with pipe flares and not the low
emissivity flare types usually found offshore.
Although this formula
provides an accurate estimate of the radiation level at a distance from
the flame, it is inaccurate close to the flare tip.
The designer is therefore left with four choices:
1.
2.
3.
4.

Reduce flare rates.


Increase the distance from the flare to the receiver.
Reduce the fraction of heat transmitted from the flare.
Increase the allowable radiation level at the receiver.

1. This is not usually a practical proposition, the maximum flaring


rate is a function of crude production and can only be altered by
reducing the production rate and hence affecting project economics.

2. This can be done by increasing the length and hence cost of the flare
boom or by installing a separate platform for the flare.
This is
not a desirable way to operate.

3. This has some attractions and is one reason for the adoption of low
emissivity type flares.
(Another is the greater degree of flame
stability resulting in less flame distortion in high winds).

304

I. CHEM. E. SYMPOSIUM SERIES NO. 85

4. This initially has some attractions but needs careful study,


the frequently assumed figure of 1.39 kw/M2 still appears to be a
reasonable figure.
Particularly as an allowance must be made for
solar radiation which may reach 0.95 kw/M2 during summer in the
U.K. (34).
The possibility of high radiation levels in situations
where personnel may escape before serious injury occurs, is sometimes
considered; however this principle is not valid offshore where space
is limited and rapid escape from the flare is frequently highly
impractical.
The prediction of flare radiation levels is a less than exact science and
a better understanding of radiation mechanisms could result in
significant savings in the structural steelwork required for flare
booms (35).

Gas Dispersion
On gas platforms cold venting of gas is an accepted practice and on
platforms with flare systems it is prudent to check how gas would
disperse if the flame was for some reason extinguished.
In particular
it is important to ensure that the plume of flammable gas will not affect
helicopter flight paths.
The rate of gas dispersion and dilution below
the lower flammable limit is dependent on the buoyancy and momentum of
the vented gas and wind conditions.
Calculation methods have been
developed for this situation, generally on the assumption that gas is
vented through a pipe to atmosphere (36).
So far no work has been done
on the likely dispersion patterns from low emissivity type flare tips;
however it should be possible to calculate dispersion rates assuming
dilution to the stoichiometric fuel/air mixture at the source.

CONCLUSIONS
The design of the pressure relief and disposal system is one of the most
complex parts of platform design.
The designer has little control over
the streams feeding the relief system and is constrained by environmental
and operational factors in the design of the disposal system.
It is
therefore worthwhile to highlight very briefly some developments which
would make the designer's task easier:

Sizing Method for Two Phase Flow through Relief Valves.


A reliable method to estimate the capacity of relief valves in two
phase flow, particularly flashing flow, situations would be
extremely useful. A number of companies have developed individual
methods but at present there is no commonly accepted method.

305

I. CHEM. E. SYMPOSIUM SERIES NO. 85

Computer Program for rating Relief Valve Headers


There are a number of commercially available computer programs for
estimating pressure in pipe networks; however, for one reason or
another they are not entirly suitable for venting applications,
where pressure drops may be very high and the vapour acceleration
pressure loss is significant.
This situation becomes even more
complex if two phase mixtures are present.

Improved Flare Radiation Calculation method.


Although there is now a better understanding of
flare radiation than ever before the calculation
still presents problems forcing designers to err
caution.
Perhaps it is now time to look at new
based on the Stefan-Boltzmann equation and flame

the mechanism of
of the F factor
on the side of
methods of radiation
dynamics.

Implicit in these recommendations is the belief that API-520 and API-521


are dated and need revising.
There is little doubt that further effort will be spent providing better
design methods for pressure relief systems; on the other hand it is also
likely that design criteria and legislative requirements will become
increasingly strict.

Acknowledgement
We would like to thank Britoil plc for permission to publish this paper.
Opinions expressed are the views of the authors and do not necessarily
reflect the opinion of Britoil plc.

306

I. CHEM. E. SYMPOSIUM SERIES NO. 85

References

1.

API-521

Guide for Pressure Relief and Depressuring System

2.

API-520 Recommended Practice for the Design and Installation


of Pressure Relieving Systems in Refineries, Part 1 Design.
Appendix C, Section 3.

3.

Kletz T.A., 1977,

4.

Personal Communication.

5.

Sallet D.W., 1979, Conference on Pressure Relief Devices,


Paper C274/19, Institution of Mechanical Engineers, London.

6.

API-520 Recommended Practice for the Design and Installation


of Pressure Relieving Systems in Refineries, Part 1 Design,
Section 5.4.2.

7.

Wallis G.B. 1969, One - dimensional Two-phase flow pp144-6,


McGraw-Hill, New York.

8.

Bliss D.G.B., Quackenbush TR, Teske ME, 1982,


Transactions of the ASME 104(4), pp272-7.

9.

API-526

10.

API-520 Recommended Practice for the Design and Installation


of Pressure Relieving Systems in Refineries, Part 1 Design,
Appendix C, Section 4.

11.

API-521 Guide for Pressure Relief and Depressuring Systems


section 3.16A.

12.

Scott D.S., 1980, "Some Seldom Considered Aspects of Pressure


Relief Systems" presented to the Institution of Chemical
Engineers Symposium on Explosions, Fire Hazards and Relief
Venting, Sheffield. (March 1980)

13.

Lees F.P., 1977,


1977, pp17-22.

14.

Rooney J.P., 1983,

15.

Personal Communication.

16.

Pilz V., 1978,

17.

Seebold J.G., 1982,

18.

API-520 Recommended Practice for the Design and Installation


of Pressure Relieving Systems in Refineries, Part 1 Design,
Appendix A.

Hydrocarbon Processing 56(8) p98.

Flanged Steel Safety Relief Valves.

Instruments and Control Systems,

November

Hydrocarbon Processing, 62(1), pp89-92.

German Chemical Engineering


Hydrocarbon Processing

307

(1),

pp63-73.

61(10),

pp75-79.

I. CHEM. E. SYMPOSIUM SERIES NO. 85

19.

API-2000

Venting Atmospheric and Low Pressure Storage Tanks.

20.

API-520 Recommended Practice for the Design and Installation


of Pressure Relieving Systems in Refineries, Part 1 Design,
Section 7.

21.

Crane Company,

22.

Duxbury H.A., 1979,


(351) pp851-858.

Chemical Engineer (350), pp783-787 and

23.

Richter S.H., 1978,

Hydrocarbon Processing,

24.

Landis R.L., 1982,

25.

Van Boskirk B.A., 1982,

26.

Paruit B and Kimmel W., 1979,


58(10), pp117-121.

27.

API-521 Guide for Pressure Relief and Depressuring Systems


Sections 5.3 A4 and A5.

28.

Personal Communication.

29.

Wilkins J., 1977, Witheridge R.E., Mason J.T.M. and Newby N.,
Offshore Technology Conference paper 2822,
Huston.

30.

Husa H.W., 1977, "Purging Requirements of Large Diameter


Stacks" presented at Fire/Safety Engineering Subcommittee
American Petroleum Institute, September, 1977, San Francisco.

31.

Reed R.D., 1972,

32.

Hajeck J.D., and Ludwig E.E., 1960, Petro/Chemical Engineering


Part 1, 32(6) C31-8, Part 2 (7) C44-51.

33.

McMurray R., 1982,

34.

Personal Communication.

35.

Personal Communication.

36.

Lees F.P., 1980, Loss Prevention in the Process Industries,


Butterworth, London.

1957 Crane Technical Paper,

No.410.

57(7) ppl45-152.

Chemical Engineering 89(5) pp79-82.


Chemical Engineering 89(17) pp77-82.
Hydrocarbon Processing

Oil and Gas Journal 70(7), pp91-2.

Hydrocarbon Processing 61(11), pp175-181.

308

I. CHEM. E. SYMPOSIUM SERIES NO. 85

TABLE

ITEM

ONSHORE

OFFSHORE

Layout

Generally spacious,
separation between process
units is not usually a
problem.

Cramped, equipment is usually


stacked in several layers one
above the other.

Transport

Transport by road and


sometimes rail.

All supplies must be transported by supply boat,


personnel transported by
helicopter. These may be
disrupted in bad weather.

Product
Storage

Usually substantial
storage available.

In many cases storage


facilities do not exist or
are very limited.
Processed gas and crude is
pumped directly into
pipelines.

Services

Connections to local
utilities exist
(electricity, water,
gas). These may be
supplemented by on
site equipment.

All services are supplied on


the platform, if necessary
supplemented by bulk
deliveries of drinking water
from supply boats.

Accommodation

Generally none, employees


return to their own homes
after work.

Accommodation and recreation


facilities must be provided
for all staff.

Vibration

Sufficient damping can be


provided by foundation
design.

Heavy foundations for equipment pose a weight penalty.


Design codes give specific
advice on vibration.

Construction

Traditional construction
methods employed.

Modular construction employed


with as much work as
possible being carried out
at onshore construction
yards to minimise hook-up
operations offshore.

Flare
Radiation

Radiation does not present


a major problem as the
stack height can be
increased or the stack
located in a remote area.

Flare location presents a


major problem, remote
location requires an
expensive support structure.
In addition the flare
radiation and turbulent air
movements caused by hot
gases must not effect
helicopter operations.

309

I. CHEM. E. SYMPOSIUM SERIES NO. 85

TABLE 2

K Values for use in Equation 5

GAS

Hydrogen

+ 5.783

Helium

+ 5.078

Nitrogen

+ 1.067 (No wind)


+ 1.707 (wind)

Ethane

1.067

Propane

2.651

Carbon Dioxide

2.651

Butane

6.586

(after Husa, ref. 30)

310

I. CHEM. E. SYMPOSIUM SERIES NO. 85

311

I. CHEM. E. SYMPOSIUM SERIES NO. 85

312

I. CHEM. E. SYMPOSIUM SERIES NO. 85

313

I. CHEM. E. SYMPOSIUM SERIES NO. 85

314

You might also like