of

e
n
i
l
t
The las

defence

J. Bellovich, J. Franklin and Dr C. Baukal, John Zink Company, USA, discuss

the importance of a reliable flare pilot system for plant safety.

rior to 1947 in the USA, process vent streams

were directly exhausted to the atmosphere.
After 1947, regulations required that hydrocarbon vent streams be safely burned or
flared. Flares are one of the most important
safety systems in the plant. They are designed to safely
and efficiently combust vent streams of combustible
gases from the plant. Failure to properly ignite the combustible vent gases exiting the flare could produce an
explosion hazard that could cause damage to equipment
and personnel both in the plant and in the surrounding
community. Large fines may also result from the release
of unburned flammable gases. Flares are often the last

line of defence for many refining and petrochemical facilities. When they are not functioning properly, the entire
facility may have to be shut down.
Flares must operate over a very wide range of conditions,
from small purge rate flows to large emergency relief flows,
with vent gas compositions that are often highly variable. Large
relief gas flows can produce large flames. Thermal radiation,
noise, and visible smoke are all important emissions that need
to be minimised. Since flares are integral to the safety of the
entire plant and repairs normally involve shutting down part or
all of the plant, they must have long service lives.
The primary objective of a flare system is safe,
effective disposal of gases and liquids at an affordable

Figure 1. John Zink flare test facility.

REPRINTED FROM HYDROCARBON ENGINEERING APRIL 2006

cost1. This can only be accomplished if the exiting vent

gases are properly ignited. Therefore, the most important safety subsystem for a flare is the ignition system. Depending on the size of the flare, there are often
multiple pilots around the perimeter of the flare outlet.
More pilots are recommended as the diameter of the
flare increases. Multiple pilots are used for redundancy
to ensure ignition of the vent gases and prevent costly

shutdowns and flare outages. Flare ignition systems

should be thoroughly tested under representative operating conditions prior to installation in the field (Figure 1).
This means they should be tested at various orientations
with respect to the wind (upwind, downwind, and cross
wind) and at appropriate wind speeds and rainfall.
The flare ignition system consists of the pilots, the
pilot ignition system and the pilot flame detection system. The goals of the flare ignition system are to reliably light the flare pilots, produce stable pilot flames that
will reliably light the flare gases, maintain pilot flames
even under severe weather conditions, and to be able to
detect if and when a pilot goes out. A recently released
standard, API 5372, provides some guidelines for each
of the components of the flare ignition system.

Flare pilots

Figure 2. CFD results showing wind deflection

over a flare tip.

Figure 3. WindProof flare pilot (US patent

6,702,572).

The primary purpose of the flare pilots is to ignite the

vent gases flowing through the flare. This means the
pilots must be capable of igniting a wide range of vent
gas compositions and flow rates, even under severe
weather conditions. Pilots are located next to the flare
outlet and must be easy to light and have stable flames
that remain lit under high wind and rain conditions.
Most flare pilots are small premixed burners where
the fuel and air are mixed prior to reaching the outlet of
the pilot. The flow of fuel and air must be properly controlled to ensure the mixture is in the flammable range,
as too much fuel or air in the mixture can make it difficult
to light and reduce flame stability. The fuel is normally
supplied to the flare pilot at a high enough pressure so
it can entrain the air needed for combustion using some
type of eductor or venturi. This type of design does not
require a compressed air supply, which is one less thing
that could go wrong and take the pilot out of service.
As the fuel flow increases, the amount of air entrained
increases and vice versa. The fuel/air mixture ratio is
fairly constant over the operating range of the pilot. The
fuel and air must be properly mixed to ensure a strong
stable flame. The pilot flame must be stable over the
entire range of operating conditions, including extreme
weather conditions. The pilot flame must also be properly moulded or shaped to ensure the flare gases are
ignited. If the pilot flame does not intersect the flow of
flare gases, those gases will not be ignited. This could
allow unburned and potentially toxic gases to be released
into the atmosphere. The pilot mixer must be located
far enough away from the flare outlet to keep the mixer
from becoming enveloped in the flare flame combustion
products.
A clean and reliable fuel source is normally used for
the flare pilots, which are designed to remain lit at all
times during plant operation, even when there are no
gases being sent to the flare. API 537 recommends a
consistent and reliable fuel be used in the pilot, such as
natural gas or LPG. Fuel filtration, such as a strainer, is
recommended to ensure the small ports in the pilot mixer
do not get plugged. The strainer is generally located just
upstream of the mixer and should be located near an
accessible platform on the stack for cleaning. Often, a
second strainer is located at grade that can be easily
cleaned and maintained. The second strainer reduces
the amount of particulates reaching the first strainer
near the mixer, which reduces the amount of service
required on the first strainer. Knockout pots mounted at
grade are sometimes used to remove any condensates
in the fuel that could plug up the mixer. Flare pilots must
be specially designed to handle fuels with high flame
REPRINTED FROM HYDROCARBON ENGINEERING APRIL 2006

speeds such as hydrogen or acetylene, because of their

tendency to burn back inside the pilot causing damage.
Pressure gauges and/or flow meters in the pilot fuel
line are helpful for initial setup and troubleshooting purposes.
API 537 states that the pilot should continue to burn
and ignite the flare at wind speeds of 100 mph under
dry conditions and 85 mph when combined with a rainfall of 5 cm/hr. These guidelines should be viewed as
the minimum level for performance. When the height of
most flares is taken into consideration, stability in winds
of 125 mph or more is desirable. API 537 recommends
a minimum pilot heat release of 13.2 kW (45 000 btu/hr)
(LHV) when flaring hydrocarbon gases with a lower
heating value of 11 175 kJ/Nm3 (300 btu/scf) or greater.
Pilots must be able to provide sufficient energy to initiate burning of the waste products as they exit the flare
tip. The energy level required is dependent on a number
of factors, including: the composition of the waste gas
being burned; the gas exit velocity of the flare pilot; the
design of the flare pilot; and the atmospheric conditions
under which the pilot is operating. Good practice recommends a minimum pilot heat release of 45 000 btu/hr
(13.2 kW) for favourable conditions. The use of lower
heat release pilots can compromise the ability to ignite
the gases venting through the flare.
During the course of a pilots service life on a flare,
the pilot can be exposed to a number of external influences including wind, rain, envelopment in the flare flame
or inert gases, smoke suppression media such as steam
Table 1. Comparison of flare pilot ignition systems
Ignition system

Conventional
FFG

Self inspirating
FFG

Compressed air

Required

Not required

Ignition line

1 in. diameter

2 in. diameter

Max distance

1 mile for FFG

line

200 ft for FFG

line

Components at grade

Main

Main

Control

Automatic or
manual

Automatic or
manual

Response time

One to several
minutes

30 - 60 s

Table 2. Comparison of flare pilot detection systems

Method

Thermocouple

Sensed phenomenon

Heat

Able to distinguish between individual pilot

flames?

Yes

Able to distinguish pilot flame from main

flare flame?

Partially

Average response time (sec)

100 - 300

Location of key system components

Pilot tip

Installation while flare is operational?

No

Serviceable while flare is operational?

No

Resistant to:
Wind

No

Rain

No

Fog

Yes

Snow

Yes

Steam

No

Sunlight (direct or reflected)

Yes

and caustic or acidic gases. Computational fluid dynamics (CFD) modelling (Figure 2) and testing have shown
that wind can cause variations in the local static pressure, which affects the ability of the pilot mixer to supply
a combustible mixture to the tip. Surprisingly, the upwind
pilot is the least affected by winds and is the most stable.
The pilots in the cross wind and downwind positions are
less stable due to the low pressure zones created by the
wind. It is important that the cross wind and downwind
pilots remain lit as they are in the best position to light
the flare gases. Due to the potential for envelopment in
either combustion products or inert gases from the flare,
all of the combustion air should be provided to the pilot
without having to rely on secondary air entrainment from
around the flare pilot flame. If a pilot is enveloped by
combustion products or inert gases, then no secondary air would be able to get to the pilot, thus potentially
reducing flame stability or even causing the flame to go
out for a pilot requiring secondary air. This is why fully
premix pilots are generally safer and more reliable than
pilots requiring secondary air. Pilots should be tested in
conditions where the pilot is enveloped by inert gases or
combustion products to ensure reliability.
Flare pilots often have some type of shield or hood
to protect them from high winds and rain that can cause
them to go out. Some type of shielding is also usually
required at the fuel/air mixer where the air is entrained
into the pilot because severe weather can adversely
affect the air entrainment into the mixer.
A new flare pilot called the WindProof pilot
(Figure 3) has been developed3,
which remains lit at wind speeds
of up to 160+ mph and rainfall of
Spark probe igniter
Direct spark
up to 76 cm/hr. It was built using
a combination of CFD modelNot required
Not required
ling and extensive testing. A specially designed test apparatus was
Not required
Not required
developed to generate the high
750 ft for ignition
N/A
winds and rain needed to develop
transformer
this new pilot (Figure 4).
Some
Some
Pilots must also be capable of
Automatic or manual
Automatic or
withstanding the heat produced
manual
during major flaring events. Long
1 - 10 s
1 - 10 s
service life is required because of
the relatively inaccessible locations at the top of tall
flare stacks. Extending the service life of a
Flame ionisation
IR sensor
Acoustic
pilot has become more
important
in
recent
Electrical current
Light
Sound
years
as
facilities
have
Yes
No
Yes
increased the amount
of time between major
Yes
No
Yes
turnarounds. The average service life of a pilot
<5
<5
5 - 10
is around seven years,
Pilot tip and grade Grade
Grade
but it is not uncommon
No
Yes
Yes
for a pilot to last 30 years
depending upon the sitYes
Yes
Yes
uation. The number one
cause of pilot failure is
Yes
Yes
Yes
flame envelopment from
Yes
No
Yes
the main flare flame.
Yes
No
Yes
Temperatures caused
by direct flame impingeYes
No
Yes
ment are far greater
Yes
Yes
Yes
than
the
maximum
Yes
No
Yes
continuous temperature

REPRINTED FROM HYDROCARBON ENGINEERING APRIL 2006

FFG piping system must be drained prior to

attempting ignition. A single FFG system can
be used to light multiple pilots. Note that the
fireball is only used to light the pilots and is not
a continuous ignition source. Proper maintenance is essential to ensure the system operates properly when it is needed.
A variation of the standard FFG method is
referred to as a self inspirating FFG because
no compressed air is needed. Fuel is injected
through a mixer which inspirates combustion
air. A piezoelectric igniter (no electricity supply
required), located downstream of that mixer,
creates a fireball inside a small diameter pipe
connected to the pilot which then lights the pilot.
Again, the lines must be dry and the length of
piping is limited to approximately 66 m.
The other method of flare pilot ignition system
uses a high voltage or high energy spark to ignite
the pilot flame. One type is referred to as a spark
probe. A side stream of flammable gases from
Figure 4. Flare pilot test rig.
the air/fuel mixture going to the pilot is used to
create a flame to light the pilot. A high voltage DC
or medium voltage high energy spark is located
in the side stream, usually located inside the
pilot, to ignite the mixture. This has some advantages over the FFG system. No compressed air
is required as the combustion air is inspirated
by the flare pilot mixer. There are no long FFG
lines from ground level up to the top of the flare
tip. The only connection between the flare pilot
control panel and the flare pilot is the electrical
ignition cable. A spark ignition system can significantly reduce the time to light a flare pilot compared to an FFG system.
A variation of the spark probe system is
usually referred to as direct spark where a
heavy duty spark plug type electrode is located
Figure 5. Schematic of a FFG with a SoundProof flare pilot
inside the pilot head. There is no side stream
flame detection system (US patent 5 813 849).
of gases as the electrode directly lights the
flare pilot fuel/air mixture.
that alloys can take in a reducing atmosphere. Actual
There are advantages and disadvantages
operating conditions will determine what materials and
of each type of ignition system, as shown in Table 1. The
thicknesses to use in a specific application.
best ignition depends on the specific application. Note
that some pilots can be retractable down to grade level
Flare pilot ignition systems
for service, to avoid having to take the flare out of operaBefore the advent of modern flare pilot ignition systems,
tion, assuming there is at least one suitably operating
it was not uncommon to light flares with flare guns, flampilot still in operation. A type of winch or crank is used to
ing arrows, burning oily rags and magnesium or phoslower the pilot down to the ground.
phate pellets. These were not very reliable and were
potentially dangerous to other processes in the plant.
Flare pilot detection systems
They were also not very effective at relighting flares
The flare pilot detection system is used to monitor the pilots
under extreme weather conditions. The first flare pilot
to ensure they are lit at all times in the event of a release
was developed in the 1940s.
of vent gases to the flare. Some type of flare pilot detecThere are two primary methods for igniting flare
tion system is required by law in the USA as designated in
pilots. The oldest and most common method is known as
40 CFR 60.18 section (f) (2). A number of systems are availa flame front generator (FFG) (Figure 5). In this method,
able which have various strengths and weaknesses.
a small and controlled fireball is produced inside a small
The most common detection system is a thermocouple
diameter pipe and sent through the pipe up to the flare
located in the head of the pilot which detects the heat propilot. The fireball is typically produced at ground level
duced by the combustion process in the pilot. The thermoand can travel up to a mile to a pilot on an elevated flare
couple junction is placed in the vicinity of the pilot flame
stack. Fuel and compressed air are mixed in proper proto detect the temperature. In a properly operating pilot, the
portions and fill the FFG pipe. The mixture is then ignited
temperature is often over 540 C. The setpoint is generwith some type of spark igniter such as a spark plug. The
ally set at approximately 150 C. If the temperature falls
fireball or flame front then travels through the pipe until
below that setpoint, the pilot is designated as having gone
it reaches the pilot, where the pilot fuel/air mixture is
out and must be relit. A thermowell and high temperature
ignited. It is important that the FFG line is dry as pockets
alloy sheath material are commonly used to ensure long
of water can extinguish the fireball. Any low spots in the
life. Note that the location of the thermocouple is important
REPRINTED FROM HYDROCARBON ENGINEERING APRIL 2006

so that it does not falsely read the temperature of the flare

flame, instead of the pilot flame, as it is possible for the pilot
to be out while the flare is lit. A potential drawback of thermocouples is their relatively slow response time. In some
cases, it could take up to several minutes or more to detect
a flame out, due to the thermal inertia of the system. Thermocouple maintenance consists of monitoring the output
to make sure it has not failed. An open thermocouple will
normally cause a high positive or negative signal, depending on how the system is wired, and will be easy to detect.
Retractable thermocouples are also available so service
can be performed at grade level while the flare remains in
operation.
A second method used to detect flare pilot flames is
referred to as flame ionisation. In a typical flame, ionised (electrically charged) gases are produced due to
the high temperatures. These ionised gases can conduct electrical energy. A flame rod is located in the
flame, with a small voltage applied to it. There is a gap
in the flame rod so that no current is produced in the
absence of a flame. In the presence of a flame, the ionised gases conduct electrical energy across the gap to
ground, completing the circuit and generating a small
current. Detection of this current signals the presence
of a flame. It is important that the electrode is electrically isolated from the pilot, which is usually done with
some type of ceramic insulator, otherwise the spark may
not jump across the desired gap and the signal will be
compromised.
In some cases, the flame rod is the same device used
to generate the spark in the direct spark ignition system,
known as the spark rod. In the ignition mode, a high voltage is applied which is enough to jump the gap and create a spark. After the flame has been ignited, the system is
switched to the flame detection mode where the very low
voltage is applied. Detection of a current (microamps) signifies a flame is present, while absence of a current (open circuit) signifies no flame is present. One of the disadvantages
of flame ionisation detection is that the flare must be out of
service to perform any maintenance.
A third method of flame detection, referred to as an
optical system, uses light to determine the presence
or absence of a pilot flame, which is normally blue in
colour. Pilot flames are usually easily seen at night by
the naked eye, but can be difficult to see during the
day because of the brightness of the sun, the long distance to the pilot and the relatively small size of the
pilot flame. One technique that is sometimes used to
temporarily visually determine the presence of a pilot
flame is to adjust the flare pilot fuel/air mixture ratio to
produce a fuel rich flame which is orange/red in colour
and easier to see than the normal blue flame. This is
generally not a suitable option as it is a manual operation that temporarily produces a weaker fuel rich flame
compared to the more stable flame operating at the
proper air/fuel ratio.
An infrared (IR) or ultraviolet (UV) sensor, properly
sighted to view the flare pilot flame, can be used to
automatically and continuously detect the presence of
a flame. More sophisticated optical IR sensors use multiple wavelengths to focus specifically on the combustion products from flames such as carbon dioxide and
water. Optical sensors can be mounted at ground level.
They should not be pointed directly at the sun. Due to
the long distances and typical fields of view of these sensors, they usually cannot distinguish between multiple
pilot flames. Also, the presence of any flame in the view
area of the sensor, including the flare flame, will satisfy

the detection system. This means that the flame could

be burning while the pilots have gone out, creating a
potentially hazardous situation. Any optical obstructions
in between the sensor and the flames can also cause
detection problems. These obstructions include rain,
snow, fog, smog, and dust, which could block the field of
view and give a false flame out signal.
A fourth method of pilot flame detection is to use the
sound (acoustics) generated by the flame. Flames generate what is usually referred to as combustion roar,
which has a distinct sound pattern. A new pilot detection
system has been developed to use this characteristic for
flame detection (see Figure 5)4. This system can distinguish between individual pilots, has a very fast response
time, can be mounted at ground level, and can be serviced while the flare is in operation. It has been tuned
to selectively distinguish between the sounds with and
without a pilot flame while ignoring extraneous sounds
such as the main flare flame or steam used for smoke
suppression. The sensor can be up to 115 m from the
pilot and 90 m from the signal processor. It is typically
used in conjunction with FFG ignition systems by using
the FFG line to detect the sound from the flare pilot
(Figure 5). The detection system is remotely located
from the flare. Service can be done on the system while
the flare is operating.
Table 2 shows a comparison of the various methods used to monitor flare pilots. The best choice often
depends on the specific application. Some companies
are even choosing to use more than one method for
added redundancy to further enhance reliability.

Conclusion

Flare ignition systems consist of flare pilots, flare pilot ignition systems and flare pilot detection systems. The flare
pilot is designed to reliably light the gases being exhausted
to the flare. It must have a robust flame to light flare gases
that may have a wide range of compositions. The flare pilot
ignition system is designed to reliably light the flare pilot
under a wide range of severe operating conditions, such
as high winds and rain. The ignition system should have
the capability to rapidly relight the pilot if it were to go out.
The flare pilot detection system ensures that the flare pilot
is lit. Rapid detection is important so pilot flameouts can be
quickly corrected. All of these components of a flare ignition system must be properly maintained in order to ensure
proper operation. A wind proof pilot, direct spark ignition,
and sound proof flare pilot detection is an example of a flare
ignition system with a stable pilot even in severe operating
conditions, fast and reliable ignition and rapid detection of
any flameouts. All of the components in a flare ignition system are required to ensure the vent gases going to the flare
are properly and safely ignited or flared. Failure to do so
can produce unburned gases that pose a potential explosion hazard that could cause damage to equipment and
injury to personnel. Releasing unburned gases can also
lead to environemental concerns and large fines from the
government.

References
1.
2.
3.
4.

REPRINTED FROM HYDROCARBON ENGINEERING APRIL 2006

SCHWARTZ, R., WHITE, J. and BUSSMAN, W., Flares, Chapter 20

in The John Zink Combustion Handbook, edited by BAUKAL, C., CRC
Press, Boca Raton, FL, 2001.
API Standard 537: Flare Details for General Refinery and
Petrochemical Service, American Petroleum Institute, Washington,
DC, 2003.
SCHWARTZ, R., HONG, J. and SMITH, J., 'The Flare Pilot',
Hydrocarbon Engineering, Vol. 7, No. 2, pp. 65-68, 2002.
SCHWARTZ, R., KODESH, Z., BALCAR, M. and BERGERON, B.,
'Improve flaring operations', Hydrocarbon Processing, Vol. 81, No. 1,
pp. 59 - 62, 2002. _______________________________________n