2 Flare
2 Flare
FLARE SYSTEM
DESIGN
INDEX
CONTENTS PAGE
Chapter Topic Sub topic
1.0 INTRODUCTION
7.1 Startup
7.2 Shutdown
7.3 Inspection
7.4 Normal operation
10.0 REFERENCES
QUESTIONNAIRE FOR
11.0
VALIDATION
1. INTRODUCTION
1.1. What is flaring?
Many industries generate significant amounts of waste streams, such as
hydrocarbon vapors, which must be disposed of, on a continuous or
intermittent basis. Some of the examples can be like off-spec product or the
bypass streams generated during startup operations. Direct discharge of
waste gas streams and vapors into the atmosphere is unacceptable due to
safety and environmental control considerations.
Gas flaring is a standard operation aimed at converting flammable, toxic
and corrosive vapors into environmentally acceptable discharges. Gas
flaring converts flammable, toxic or corrosive vapor to less objectionable
compounds by means of combustion. Flaring is a critical operation in many
plants where design must be based on strict safety principles.
1.2. Why is flaring required?
In general, proper planning and layout of process plants require that special
consideration be given to the design of various safety facilities to prevent
catastrophic equipment failure. These facilities are designed to prevent
overpressure and to provide for safe disposal of discharged vapors and
liquids. Portions of these facilities are also used as an operational tool for
safe disposal of hydrocarbons - particularly during start-up and shutdown
phases.
Standard pressure relieving devices most often used are safety and relief
valves, rupture disks, pressure control valves and equipment blow down
valves. Direct discharge of waste or excess vapors to atmosphere is
unacceptable either:-
Because of restrictions imposed by local ordinances or plant practices.
Concentrations of the contaminants at ground or adjacent platform levels
may exceed permissible explosion or toxicological threshold limits.
Meteorological considerations such as severe temperature inversions of
long duration may occur, creating hazardous conditions.
Non hazardous vapors such as low pressure steam are usually discharged
directly to the atmosphere, in contrast, hydrocarbon vapors that are
discharged on a continuous or intermittent basis cannot be directly
discharged to the atmosphere and should be disposed of through a closed
system, and burnt in a flare.
1.3. Types of Flare
There are basically two types of flare system namely, Elevated Flares &
Ground Flares.
In an elevated flare system, combustion reactions are carried out at the top
of a pipe or stack where the burner and igniter are located. Relieving gases
are sent through an elevated stack from a closed collection system and
burned off at the top. The flame generated is open in this case.
A ground flare is also similarly equipped except that the combustion takes
place at or near ground level. The flare flame is contained in a flare
chamber.
Three types of ground flares are in general use:-
1. The type that uses water spray to disperse the combustion gases.
2. The venturi type that depends on the kinetic energy available in the
waste gases to inspirate and mix the proper amount of air with the
gases.
3. Multi Jet ground flares where the flow of the waste gas is distributed
through many small burners.
The principle advantages of a Ground Flare system are:-
1. No structural support is required
2. Erection is relatively straight forward and requires light parts
3. Maintenance is easy
4. Operating costs are negligible
5. The flame of the flare is not visible since it is hidden in a box. It requires less
steam to produce a smokeless flame since it produces relatively non-luminous
flame because of more controlled combustion at the multiple burners.
6. Finally, with the exception of the venturi type, it is a fairly quiet system
The multi jet type normally used has high initial costs and capacity limited.
In contrast, an Elevated flare requires less ground area. Because of its high
elevation, it can be located within a process area or on the periphery' of the
plant site, since radiation effects and ground level concentrations of
pollutants can be maintained within allowable limits. Piping costs tend to be
lower due to smaller and shorter pipe runs. Also the distance between the
point of discharge from safety valves and the flare stack is less than that in
the case of ground flares.
A problem with elevated flares is that initial land operating costs are high.
Maintenance is also difficult and tedious. The visibility of the flame is the
most serious disadvantage and sometimes causes objections from local
community. These systems also require more steam to produce a smokeless
flare. A final disadvantage is that noise levels are relatively high.
There are situations when a ground flare is used in conjunction with a second
conventional flare, which may be an elevated system. The ground flare is
designed to handle the normal flaring requirement. In the event of major
failure, excess flow is automatically diverted by a seal to a second flare.
Since, the possibility of a major failure is rather remote, it may not conflict
with pollution or local site regulations.
2. COMPONENTS OF THE FLARE SYSTEM
As we know by now, that in a flare system, relieving gases are sent through
an elevated stack from a dosed collection system and burned off at the top.
3. Knock out (KO) drum located before the flare stack in order to
separate any condensate or liquid from the relieving vapors (it is
hazardous to bum liquid droplets).
The relieving gases from safety relief valves and pressure control valves are
collected in a horizontal or vertical knock-out drum through a flare main
header. Any condensate carried out along with the gases is knocked down
here. A constant liquid level is maintained in the boot/drum. The liquid is
pumped to a slop tank or is reused in oil recovery facilities. If required,
steam is used for winterizing to prevent freezing. The gas from the KO drum
is then sent to an elevated flare stack. At the bottom of the stack, normally a
liquid seal is maintained. Alternately another seal may be located between
the KO drum and the flare stack. A positive water seal is maintained by
controlling the level. In cold countries, the water seal is also provided with
steam for winterizing.
The stack is comprised of a riser section, molecular seal and burner tip.
1. Riser structure:
This normally consists of two or more sections. The flare header enters at the
bottom section, which can also serve as a flare stack knock tut drum where
any condensate carried over from the main knock out drum is collected.
2. Molecular Seal:
This is welded to the riser section. It provides a seal against entrance of air
into the flare stack and minimizes the possibility of an explosive mixture
forming in the flare system. More information on a molecular seal is given in
one of the coming chapters. Briefly, it resembles a bubble cap and creates a
seal by using the buoyancy of the purge gas to create a zone where the
pressure is greater than atmospheric pressure.
At the top of the burner tip, pilot burners, which 'are automatically lighted
from a remote place through the igniter line, are positioned. The steam
connection is also provided for smokeless flares and a purge gas connection
is provided for maintaining an air free system and to prevent flash back by
maintaining pressure at the molecular seal higher than the atmospheric
pressure. This arrangement prevents air from re-entering the stack from
ambient surroundings.
Figure-1 shows a schematic diagram of the entire Flare System.
In the next few chapters, we shall go through the flare system design
guidelines, component wise.
3.2 Estimating relief rates:
Each pressure relief valve should be individually analyzed for any probable
causes of over pressure due to operational failures and plant fire.
The valve should be sized for the case that will require the maximum
relieving rate. If a fire condition is controlling, two separate safety valves,
one for fire condition and the other for operational failure, may be provided
since the fire situation is less likely to occur.
Guidelines for determining individual relieving rates are illustrated with an
example of a column -
Consider a fractionating column where different causes of overpressure may
be analyzed as follows:
1- Blocked outlet condition of the overhead vapor line by inadvertent closure:
In this case, it may be assumed that heat input to the reboiler is normal, and
the reflux will still be maintained since the overhead receiver has the holding
capacity for about 10 minutes. Hence, the relieving capacity of the pressure
relief valve may be assumed as the normal vapor load to the overhead
condenser. The relieving pressure will be the set pressure of the PRV and the
temperature will be the boiling point corresponding to that pressure.
Vaporization rate may have' to be corrected as the latent heat changes with
change in boiling temperature which in turn changes because of change in
column pressure. Also the reboiler duty may get affected due to process side
changes in pressure and temperature.
2- Cooling water failure
The cooling water typically, used as the cold utility in the overhead
condenser may stop because of power failure or some other operational
problem. Under this situation, overhead vapor will not condense in the
condenser and because of the vapor accumulation, the pressure will rise. The
reflux can still be maintained for about 10 minutes because of the holding
capacity of the accumulator. The relieving capacity of the pressure relief
valve will also be the normal vapor rate to the condenser. Vaporization rates
may need correction here also.
3- Reflux failure
This may be associated with the malfunction of reflux control valve, pump
failure, or any other operational problem. In this case, the overhead
condenser becomes flooded with condensate. As a result of this, overhead
vapor cannot condense and pressure starts building up. Once the pressure
reaches the set of PRV the relief occurs. The vaporization rate, here as well,
can get affected by rise in pressure.
4- Abnormal heat input by reboilers
If the reboiler controller malfunctions for any reason, the rate of
vaporization may increase. If the vaporization rate exceeds the rate of
condensation, the pressure will build up. In this case, the relieving rate
should be the difference between the maximum rate of overhead vapor and
the maximum rate of condensation of the condenser. In the absence of data,
the relieving rate may be assumed to be the normal vapor load to the
condenser.
The column can also get subjected to high pressure, if the reboiler is an
exchanger, carrying the hot utility (like steam) at higher pressure than the
column bottoms pressure and the exchanger tube leaks.
For relief loads due to fire:
Q = 21,000 FA0.82
Where;
Q = Total heat absorbed in BTU/hr,
A = Wetted surface in sq.ft.
F = Environment factor.
W= Q /
Where;
W = Vapor release rate in Lbs/hr
Q = Total heat input BTU/hr
= Latent heat of fluid in vessel evaluated at the relief valve
inlet pressure, BTU/Lb.
No credit is normally taken for the sensible heat capacity of the fluid in the
tank
For a fluid above the critical point, i.e. when pressure relief conditions are
near or above the critical point, the rate of vapor discharge depends upon the
rate at which the fluid will expand as a result of the heat input. The latent
heat & vaporization at or near the critical point is almost zero in this case.
More information on the relief rate calculation is available in API 520 and in
the training module on the relief valves.
For the fire case, a cause of fire is normally localized. The whole plant is
divided into different fire zones. The flare load is generally calculated based
up on one or two related zones. However, it is not unusual to consider the
total load.
The method of calculating the time element related to each pressure relief
valve is referred to as "TRANSIENT LOAD ANALYSIS". This is based
upon the non steady state condition in the flare system of a plant during
emergency situations. This calculation is tedious but with simplified
assumptions, it provides an estimate of the relative time delays of the
individual valves.
4.0 DESIGN OF COLLECTION SYSTEMS
The relieving vapors from different PRVs and depressurizing valves must
first be collected in individual flare sub-headers located near each process
area. Sub headers must be interconnected to a main flare header which leads
to a knock out drum. Condensates carried over by vapors are separated in
this vessel. Vapors leaving the KO drum from top move up the flare stack
where they are subsequently burned at the tip.
The no. of main flare headers and the individual sub headers connected to
them depends up on the type of vapors handled, temperature and the back
pressure limitation of PRVs.
The pressure level of the flare header depends on the type of pressure relief
valves used to protect the equipment and the pressure levels of the
equipment connected to the flare system. In the conventional type of PRV,
the performance depends on the back pressure. A maximum back pressure of
10% of the maximum allowable working pressure is a limit for the
conventional type of PRV. For the nonconventional valves like balanced
bellow type, piston type or pilot operated type, the maximum allowable back
pressure may be taken as high as 40-50% of the valve set pressure.
4- The equivalent length of the main flare header is then calculated from
the flare stack to the last safety valve, taking in to consideration the
straight length of the pipe and approximate equivalent lengths for
bends, etc. If the actual location of the flare stack is not known by that
time, it may be assumed to be 500 ft from the last piece of equipment.
5- A trial estimate is made for determining the diameter of the flare
header based up on the max. relieving flare load and considering the
back pressure limitation of 10% for conventional valves and 40% for
balanced type valves. Note, however, a single main header in many
cases, may be too large to be economically feasible.
6- The second trial is required for two main flare headers, one collecting
the Low Pressure (LP) flares (usually 5 to 10 psig) and the other
collecting relatively High Pressure (HP) flares (usually 15 to 20 Psig).
The two headers are connected to their individual KO drums. The
vapor lines from the KO drums are combined into single header
connected to the flare stack.
The load in a sub-header used for the line sizing, need not be same as
the load which is utilized for designing the main header or the flare
stack.
7- The next consideration is the cost of construction materials. This
determines the final no. of flare headers. Vapors that normally require
expensive materials may be listed as:-
8- Wet flare and Dry flare: Some times, relatively hot vapors carrying
condensates may be separated from the dry cold vapors. They do not
run as separate headers but either LP or HP flare headers may be
associated with any one of them. Thus a wet flare header may be in
fact the LP header and the dry flare header may be the HP flare or vice
versa.
9- After the total no. of flare headers has been established, it may be
necessary to recheck the vapor load in individual headers since
introduction of a separate header may allow subtraction of the flow
quantity from earlier header to which it was added initially.
For Example:
Once the relief load is established and the maximum allowable back pressure
has been defined, line sizing is reduced
:
to standard flow calculations.
The flare lines carry the vapors which are compressible in nature. The flare
lines are also normally long and not fully insulated. Hence, the flow cannot
be adiabatic flow. Hence, the flare headers are typically sized based upon
isothermal compressible flow. This also gives more conservative design.
[ ]
( )
Where,
Gci = Max. mass flow or critical mass flow, lb /sec ft2
Po = Absolute upstream pressure, lb/in2
M = Molecular weight
To = Upstream temperature, Rankin
Z = Compressibility factor
The actual mass flow G (lb /sec ft2) is a function of critical mass flow Gci,
line resistance N, & ratio of downstream to upstream pressure. This is
represented by figure 2. In the area below the line in the figure 2, the (G /
Gci) remains constant, which indicates that the sonic flow has been
established. Thus, for sizing flare header, the plotted point must be above
the line.
Line resistance,
Where,
Knowing all other values, the pipe inside diameter (di) is calculated.
Knowing di, the Reynolds no. and friction factor can be calculated.
Assuming a straight length of pipe for L=500, line resistance N is calculated.
G is calculated based on the di. Gci is calculated based on downstream
pressure & is called Gc2. G/Gc2 evaluated & P2/P1 determined from Fig.3
since P2 is known, P1 can be calculated. Pressure at inlet of KO drum is taken
as Pl + 0.5 psi.
The sum of all pressure losses starting from flare stack up to the safety valve
yields the total back pressure. This back pressure must be lower than the max
back pressure allowed in the system & corresponding to the lowest set
pressure of the safety valve.
4.2 Example
The maximum flare load of a system is 1,000,000 lb/hr of vapor. The
pressure at the base of the flare stack is 2 psig, the average MW of vapor is
50 and temp is 200 F. The distance from the drum to stack is 500 ft. The line
has two 90 degree welding elbows and an orifice with Ki factor of 0.2. The
total pressure drop at the knock out drum is 0.5 psi. Determine pressure at
inlet of the knockout drum. Also, given are
K = Cp/Cv = 1.2
Z=1
Solution:
= (M* P ) / ( R * T )
= (50* (2+14.7))/ (10.73 * (200+460)) = 0.12 lb/ft3
[ ] & [ ]
[ ]
[ ] [ ]
N = 4.15
Now, G = W/ ( d2/4)
= (4 * 1000,000) / (3600 * 3.14 * (29/12) * (29/12))
= 60.6 lb/sec ft2
[ ]
( )
Hence;
This is a very high pressure drop: Typically, it should not exceed 3 psi.
Hence, a larger pipe diameter is required.|
The above procedure is repeated for higher diameter pipes. It can be seen
that, when pipe ID is 41.25 (corresponding to standard OD of 42),
N = 3.1
G /Gc2 = 0.49
P2/P1 = 0.87
P1 = 19.2
And the pressure drop (P1 - P2) is 2.5 psi, which is acceptable.
Hence, total pressure drop = Line P
: + KO drum P = 2.5 + 0.5 = 3 psi.
Thus, the pressure at inlet of the KO drum is 16.7 + 3, i.e. 19.7 psia or 5 psig.
5.0 DESIGNING THE FLARESTACK & ACCESSORIES
1. Horizontal drum with vapor entering at one end of the vessel &
exiting at the top of the opposite end (no internal baffling)
2. Horizontal drum with vapor entering at each end on the horizontal
axis & a central outlet.
3. Horizontal drum with vapor entering in the center & exiting at the
two ends on the horizontal axis.
4. Vertical drum with vapor entering at the top on a certain diameter &
provided with a baffle so that the flow is directed downward. Outlet
nozzle is located at the top of the vertical axis.
5. Vertical drum with a tangential nozzle
Selection of the drum arrangement depends on economics. When
large liquid volume storage is required & the vapor flow is high,
normally a horizontal drum is more economical.
.:
Split entry/exit reduces size of the drum for large flows. As a rule
of thumb, when drum diameter exceeds 12 feet, split flow
arrangement is normally economical.
KO drums are usually sized by a trial & error method. Liquid
particles can drop out when the vapor velocity traveling through
the drum is sufficiently low. In other words, the drum must be of
sufficient diameter to affect the desired liquid - vapor separation.
The factors considered while designing the knockout drums are:-
1- The residence time of the vapor should be equal to or greater than the
time required tor a liquid droplet to travel the available vertical height
at dropout velocity of the liquid particle.
2- Sufficient volume should be provided for the liquid accumulation in
the knockout drum.
Where;
V= 0 .4 * ( ( L - G) / G) ^ 0 .5
Where;
Liquid seals
Liquid seals are further classified as seal drums and seal pipes. In the former,
a liquid seal is used in a seal drum located between the KO drum & flare
stack. Seal drums can be horizontal or vertical; the selection mainly depends
on the availability of space. Figure 4 shows a horizontal and a vertical seal
drum. Instead of a drum, sometimes, a piping seal is used as a seal leg
located at the bottom of the stack. This is often an integral part of the stack.
A seal drum maintains a seal of several inches on the inlet flare header,
preferably not exceeding 6 inches. More is the height of the seal, more is the
back pressure. Sealing liquid is usually water with a continuous flow, the
overflow going to the sewer.
The capacity of the seal drum is usually the volume corresponding to 8-10 ft,
of the vapor inlet line. In a vertical drum, the ratio of the inlet pipe
cross-sectional area to the vessel free area for gas flow above the liquid
should be at least 1:3 to prevent upsetting surges of gas flow to the flare,
For this, area for the gas above the liquid surface should be at least equal to
that of a circle having diameter, D = 2d, where d is inlet gas pipe diameter.
The height of the vapor space above the liquid level in a vertical drum should
be app. 2-3 times, the diameter (d) to provide disengaging space for
entrained seal liquid.
Gas seals
A more recent gas seal type of device that has been developed to prevent
flash backs in the flare system is Molecular type seal. It uses a purge gas of
molecular weight of 28 or less (like N2, CH4 or natural gas). Because of the
buoyancy of the purge gas, it creates a zone having pressure greater than the
atmospheric pressure. The molecular Seal is located at the top of the flare
stack immediately below the burner tip; the ambient air cannot enter the
stack because of this high pressure (Figure 6). The recommended purge
velocity through the molecular seal is about 0.1 ft/sec. If a molecular seal is
not used, the recommended velocity is 1 ft/sec, thereby increasing the purge
gas requirement.
The flare burner is located at the tip of the flare stack. The top section is
normally about 12 ft long & is called the flare burner tip. The burner
diameter is sized on a velocity basis. The flame blow out can occur when the
exit velocity of the vapor exceeds 20-30% of the sonic velocity.
Mass flow is given as;
W = 3600 * G * Ac * V
Where;
Vapor density;
Where;
M = Molecular weight
P = Absolute pressure of vapor =14.7 psia
T = Temperature, R.
g = Acceleration due to gravity = 32.17 ft/sec2
R = Gas constant = 1546 ft lb force/R. mol.
K = CP/CV = 1.2 (assumed).
d = Diameter of flare tip, inches.
[ ]
If based on the maximum rate, the diameter may be too large. In such case,
the normal flow is used to arrive at value of d and velocity for the
maximum flow is kept at maximum 40% of the sonic velocity.
5.4 Example
The flare normal load is 800,000 lb/hr whereas max load is 1,000,000 lb/hr.
The vapor temperature is 300 degree F and molecular wt. is 50. What should
be diameter of the burner tip?
Solution:
[ ]
[ ]
Vapor density;
= 50*14.7/ (10.73*760)
= 0.09 lb/ft3
Max Velocity = W/ (3600* n * *d2 /4 )
= 1000,000*4/(3600*0.09*3.14*(48/I2)*(48/12))... based
on max flow = 246 ft/sec
Sonic velocity, V* = (g KRT/M) 0.5
= (32.17*1.2*1546*760/50) ^ 0.5
= 953 ft/sec
Thus, the maximum velocity is 25.8 % of the sonic velocity, which is less
than the max limit of 40%. Hence, the diameter of burner tip should be
48".
The location of flare is a safety related issue. The flare stack is generally
located on the downwind of normally prevailing winds & remote from
operating & traffic zones.
In India, as per Petroleum Rules, 1976 (page 49, point 169 ), no flare shall
be situated nearer than 90 meters to any tank, still, pump-house or any
facility for the refining, cracking, reforming, blending, storage for
handling of petroleum or liquefied petroleum gases other than knock-out
drum and condensate recovery pump attached to such flare.
5.6 Flare stack design
For max stack discharge, a mach number of 0.2 is recommended. From fig. 8
then, corresponding L/D is 118. From the stack diameter D, the flame length
L can be determined. Thus,
L = 118 D (I)
The thermal radiation and escape time can be estimated from the data in
table-2. Values are based on experimental data on the threshold limit of
pain to the human body as a function of the radiation intensity in
BTU/Hr/ft2, generated by a flame,
A safe level of heat radiation intensity for unlimited time exposure has
been found to be 440 BTU/hr/ft2. It is apparent that a time interval with
varying radiation intensity must be allowed, to permit a human to escape
from a suddenly released intense heat source. The varying radiation
intensity results from an individual increasing his distance from the source
of heat.
Assume a person is at the base of a flare stack when heat is suddenly
released. The average individual reaction time is between 3 and 5 seconds.
Hence, during this short reaction time interval, the full radiated heat intensity
will be absorbed. Then follows another short interval (20 ft/sec is normally
assumed to be the average escape velocity of a man) during which
continually decreasing amounts of heat will be absorbed until safe distance
is reached (heat intensity for a safe location is 440 BTU/Hr/sq.ft.)
( )
Where;
ta = t r + t e
The escape time depends on the stack height, H. The following steps
outline the, approach to determining the flare stack height based upon the
radiation intensity.
1- Calculate the radiation intensity using the following equation
.(II)
Where;
Where; hc = net heat value of a gas (LHV) in BTU/scf (60 deg.F, 14.7 psia)
Q = W * hc * 379/M
Where;
Hence; X2 = H (H + L) + y2 ..(III)
Hence, from equations (II) and (III), and for max radiation density (qm) at
flare base where y = 0,
( )
Hence, H is derived as
{[ ( )] - } .. (IV)
[ ] OR [ ]
And we note that y = radial distance from the base of the stack:-
y = [X2 - H(H+L)]l/2
Allowing for the speed of escape (20ft/sec) we have
y = 20 te = [X2 - H(H+L)]l/2 .. (V)
This defines the safety boundary, corresponding to quiescent ambient air.
Thus, the stack height H, the limiting heat radiation qM, and the radial
distance, y can be evaluated with a trial and error procedure, by assuming a
value of te.
The above analysis must be extended to account for the more prevalent
case of wind circulation in the vicinity of the flare. For those sections
where wind intensity is unknown, it is suggested that an average 20 mph
wind be assumed in all directions, which results in increasing the safe
circular boundary by the resulting tilt of the flame (Figure 11). The flame
tilt and its effect on the safety boundary increase may be determined as
follows:
tan = Uw/U
Where;
Uw = Wind velocity
U = Flare exit velocity
U w = [Xm - H] sin , and
Ut = [Xm - H] cos
y = [X2 - (H + (Xm - H) cos )2 ]1/2 + (Xm - H) sin .(VI)
This formula establishes the limiting boundary for wind circulation. When
evaluating wind effects on flame tilt, average wind intensity should be used
in the calculations.
5.7 Alternate method of calculating safety boundary
For high flaring rates, the stack height calculation previously described leads
to a very tall stack. Part of the reason for this conservative estimate is that
calculations are based up on the thermal effect on bare skin. If proper
clothing is provided to personnel before entering the flare stack area and
proper shielding is installed at the stack or at the equipment to reduce the
radiation effects, the required stack height can be greatly reduced, However,
there is a tradeoff in that the safe boundary limit must be increased.
Since heat load of the flare, the flame length, and the safe radiation intensity
(440 BTU/hr/sq.ft) remain the same, decreasing the stack height leads to an
increase in the safety radius. Another important consideration is the type of
support provided for the stack. In general, the higher the stack the greater the
structural support costs.
An alternative method of stack sizing is based on the allowable limit for
radiation intensity. For operating personnel the allowable intensity is 1500
BTU/hr/sq.ft. , and for equipment it is 3000 BTU/hr/sq.ft.
The 1500 BTU/hr/sq.ft criterion is established from the following basis. In
emergency releases, an operation time of 3 to 5 sec. may be assumed.
Perhaps 5 to 10 sec. more would elapse before an individual could escape the
area via an average velocity of 20 ft/sec. This would result in a total exposure
period ranging from 8 to 15 sec. only. The time to pain threshold
corresponding to 1500 BTU/hr/sq.ft. is 16 sec. before the individual could
escape to a safe place. The effect of radiation on equipment is shown in
figure-12. The temperature of metal equipment increases with exposure time
and higher the radiation heat intensity, the greater the temp. Curve I in figure
13 shows the theoretical equilibrium temps for metal equipment, based on a
view factor of 0.5. The actual temp, on surfaces facing the flame the flame
will be between curves 1 &. 2.
The temperature of the vessels containing liquid or flowing vapors may be
lower because of cooling effects; Curve-2 applies to materials having a low
heat conductivity coefficient e.g. wood. In this case, equilibrium
temperatures are reached within a shorter time as compared with metal
objects. Dehydration of wood takes place at about 500 deg.F, decomposition
at 700 deg.F, and ignition at around 800 deg .F, corresponding to heat
intensities of 1300, 3000 and 4000 BTU/hr/sq.ft respectively. This means
that wooden structures and vegetation exposed to heat intensities of 3000 to
4000 BTU/hr/sq.ft. and higher may catch fire and bum. Paint on equipment
also may also be damaged.
Therefore, it is recommended that equipment located in this area be
protected by proper heat shielding or emergency water sprays.
1 From equation II, the radial distance from the flame at Q = 1500
BTU/hr/sq.ft. is calculated.
2 The safe radial distance at Q = 440 BTU/hr/sq.ft. is calculated from
the same equation
3 A suitable value for Q is assumed at the base of the stack. Q = 3000
BTU/hr/sq.ft. is a good start since protective shielding will be
provided in this case at the stack.
5.8 Example
The flare normal load is 800,000 lb/hr whereas max load is 1,000,000 lb/hr.
The vapor temperature is 300 degree F and molecular wt. is 50. Stack
diameter is 48. Average wand velocity is 20 mph and net heating value is
1500 BTU/scf), Calculate the stack height and the safe boundary.
Solution:
Total heat released, Q = W * hc * 379/M .. Equation as given earlier
= 1,000,000 * 1500 * 379/ 50 max flow considered
= 11370 X 106 BTU / hr.
Flame emissivity, f = 0.2 (hc/900)l/2 .. Equation as given earlier
= 0.2 (1500/900)1/2
= 0.258
This is the shortest possible stack height, but is not a practical height as it
assumes te=0.
If a reasonable escape time i.e. te = 30 sec. is assumed, then figure 9 gives qM
= 1330 BTU/hr/sq ft. Then, H = 245 ft (as per equation IV)
Now, let us calculate the wind effect on the safe boundary around the flare
stack.
X m = [ H* ( H + L ) ] 0 . 5 = 4 l 9 f t
X m - H = 4 1 9 - 2 45 = 1 7 4 f t
Now, y = [X2 - (H + (Xm-H)cos )2 ]1/2 + (Xm-H) sin
Substituting the values, we get y = safe radial distance from the base of stack
618 ft.
5.9 Estimating ground level concentrations
After the stack height has been established from radiation intensity values,
the maximum permissible ground level concentration of toxic gases in the
event of a flame blow out should be evaluated. Table 3 represents
toxicological threshold limit as allowed by the environmental protection
agency (EPA).
Xmax = (H/Dz)2/(2-N)
Where;
Cmax = Concentration at grade in ppm (volume).
V = Specific volume of toxic gas, cu ft per lb.
M = Weight discharge of pollutant component in tons per day.
Dz = Vertical diffusion coefficient.
P = Air velocity at grade, mph.
H = Stack height, ft.
Dy = Horizontal diffusion coefficient.
Xmax = Distance from stack to the point of maximum concentration, ft.
N = Environmental factor.
Dz = 0.13,
Dy = 0.13,
N = 0.25
5.10 Stack Support:
Up to 150 to Above
150 ft 200 ft 200 ft.
Capital investment (equipment only)
Least expensive
Derrick type Derrick type Guyed
Self Supporting Guyed Derrick
Guyed Self Supporting Self Supporting
Most expensive
Installation Labor
The pilot is piped to the top of the flare stack via a 2 venturi burner. Nozzles
are provided at the end of the pipe,1 In some designs, nozzles are hooded and
should the flame blow out, the heat of the nozzle will immediately reignite it.
In the pilot igniter system, the gas pipe is connected to a 3 venturi type
burner, which is located at the bottom of the stack. The fuel gas flows
through a nozzle to inspirate air to form a combustible mixture. The igniter
with spark gap is located approx. 3 ft above the burner. When the igniter
button is pushed, the resulting spark ignites the gas air mixture. The flame
front generated travels up the pipe at the top of the flare and ignites the gas
from the pilot nozzles.
OTHER DESIGN CONSIDERATIONS
Stack
Structural members, hardware and bolting Should be hot dip galvanized after fabrication
There are several theories which try to explain the chemistry of smokeless
flares, using steam. One of them assumes that the steam separates the
hydrocarbon molecules, thereby minimizing polymerization reactions and
forms oxygen compounds that bum at reduced rate and temperature, so as to
prevent cracking.
Another theory, claims that steam reacts with carbon particles forming CO,
CO2 and H2, thereby removing the carbon which forms smoke after cooling.
C +H2O CO +H2
CO +H2O CO2 +H2
The latter reaction is also known as water gas shift reaction.
It may be observed from this that the higher the mol. wt., the higher the
required steam. This may be associated with the theory' that the higher the
mol. wt. the lower the ratio of steam to C02 after combustion, resulting in a
greater tendency to smoke. :
Since, steam consumption is rather high (about 0.464 lb/lb of hydrocarbons
with mol. wt.50), it is too expensive to provide for smokeless burning for the
max. flare load. Normally, 20% of the max. flare load is designed for
smokeless burning. This is well supported by the fact that massive failure is
very' rare and in 90% of occurrences, smokeless flares are produced.
6.2Fuel requirement
Fuel gas supply to the pilots and igniters must have high reliability. Since,
normal plant fuel sources may be upset or lost in the plant upsets; it is
desirable to provide a backup system connected to the most reliable alternate
fuel source with provision for automatic cut in on low pressure. The flare
fuel system should be carefully checked to ensure that hydrates are not
present to cause problems. Because of small lines, long exposed runs and
large vertical rises up the stack, use of liquid knock out pot is frequently
warranted to remove condensates that may have collected in the fuel line
especially during winter. It is a good practice to provide a low pressure alarm
on fuel supply after the last regulator, which will warn the operator.
6.4 Purging of flare lines
Any gas or mixture of gases that cannot reach dew point at any condition of
ambient temperature can be used as a purge for flare system. Nitrogen,
Methane or Natural gas are normally used as purge gases.
Purging is normally of two types: Normal purging and emergency purging
Normal purging is used continuously and admitted to the flare system at the
end of each sub header and at the bottom of the molecular seal at the flare
stack. When the molecular seal is used, it is that purge volume which will
create a velocity of 0.1 ft /sec. at the flare tip. When a molecular seal is not
provided, the exit velocity is 1 ft/sec. The purge volume depends upon the
wind velocity at the flare elevation. These velocity criteria as* based on a
wind velocity of 15 mph and vary as the square of the wind velocity.
Emergency purging is used to compensate for thermal shrinkage. After
cessation of hot vent gas flow, the system residual gas will shrink as it cools
to the ambient temperature. It normally takes about 15 minutes to reach
equilibrium. Unless the purge is admitted to the system, the shrink will draw
air back in to the flare header. The shrink problem can be overcome by
sensing the system temperature and adding makeup gas at a rate
commensurate with the system volume and the max. anticipated gas
temperature.
6.5 Noise pollution
Noise pollution from flares has for too long been an inconvenience, accepted
in petrochemical plants as an inevitable byproduct of flaring process. It has
been established that major individual source of noise from flare is usually at
the flare tip itself This is especially true when the flare tip is of the type used
for smokeless flaring of hydrocarbon gases utilizing steam injection.
Basically noise is created because of two reasons, steam energy losses at the
high pressure steam injectors and unsteadiness in the combustion process.
Ground flares are normally quieter than elevated flares. This is probably due
to the fact that the flame contained inside a box is protected from wind
effects and the stabilizing effect of the heat re-radiated from the refractory
walls reduces the random characteristics of combustion. The walls
themselves will absorb some of the sound energy.
Sophisticated designs of flare tips have greatly reduced the noise pollution.
In some designs, combustion efficiency has been greatly increased by
remixing of air with gas before they are combusted. Steam is also premixed
with air and gas before gases leave the flare tip. Some of the turbulent noise
energy is thus shielded by the tip itself:
7.1 Startup
The flare startup and shutdown procedures may differ from a plant to plant
depending on the flare system it has. Here are some general guidelines,
which are followed when starting up or shutting down a flare system.
Initial checkout
1 After completion of construction, the system should be thoroughly
flushed with water to remove scale and debris Pressure testing
should be conducted where required. Special attention should be
given to all flanged joints, valves and connections. All leaks found
should be repaired and re-tested.
2 The flare KO drum pump should be checked for ease of operation
and correct rotation.
3 All instruments should be checked for proper connections and
performance
The total flare system can only be shutdown and isolated after all the process
units are shut down, drained of hydrocarbons, depressurized and purged as
necessary. Then flare system is purged with nitrogen before opening up the
KO drum, molecular seal etc. for any maintenance.
Individual process units or pipes of equipment can be isolated from
operating flare system after they are shutdown by closing block valves and
installing blinds, when maintenance is required.
7.3 Inspection
During the normal operation, the shift crew monitors the flare and ensures
that it is smokeless. The flame length is monitored to identify abnormal
releases in the flare system. In the normal operation,: the amount of vapors
flared can be monitored. As this is the material wasted, efforts are to be
taken to minimize the normal load which is flared.
The KO drum level and the flare header purge gas minimum flow is
ensured during the normal operation. The operations crew also ensures that
the seal liquid rate (and hence, the seal) is maintained for the liquid seal
system. It is also checked that the pump-out pump of the KO drum is
always available.
TABLES AND FIGURES
Table-1 - Resistance coefficient K for various pipe fittings