U.S. patent number 3,824,391 [Application Number 05/362,546] was granted by the patent office on 1974-07-16 for methods of and apparatus for flame monitoring.
This patent grant is currently assigned to Central Electricity Generating Board. Invention is credited to Bernard Edward Noltingk, Norman Edgar Robinson.
United States Patent |
3,824,391 |
Noltingk , et al. |
July 16, 1974 |
**Please see images for:
( Certificate of Correction ) ** |
METHODS OF AND APPARATUS FOR FLAME MONITORING
Abstract
For monitoring each individual flame in a multi-burner furnace,
two optical light receiving systems each have a photo-sensor on
which is focused a narrow beam, the two beams intersecting at a
small angle in or near the flame to be monitored. The angular beam
widths are less than the angle of intersection. The alternating
electrical signals from the photo-sensors due to flicker components
in the flame are correlated in a correlator which determines the
degree of correlation whereby the presence or absence of a flame in
the intersection region only of the two beams is detected.
Inventors: |
Noltingk; Bernard Edward
(Dorking, EN), Robinson; Norman Edgar (Croydon,
EN) |
Assignee: |
Central Electricity Generating
Board (London, EN)
|
Family
ID: |
23426531 |
Appl.
No.: |
05/362,546 |
Filed: |
May 21, 1973 |
Current U.S.
Class: |
340/578;
250/208.6 |
Current CPC
Class: |
F23N
5/082 (20130101); F23N 2223/10 (20200101); F23N
2229/08 (20200101); F23N 5/08 (20130101); F23N
2229/16 (20200101) |
Current International
Class: |
F23N
5/08 (20060101); H01j 039/12 () |
Field of
Search: |
;250/208,209,214R,217F
;340/227,228.1,228.2 |
References Cited
[Referenced By]
U.S. Patent Documents
Primary Examiner: Lawrence; James W.
Assistant Examiner: Nelms; D. C.
Attorney, Agent or Firm: Mawhinney & Mawhinney
Claims
We claim:
1. A method of monitoring a selected flame in a multi-burner
furnace comprising photo-electrically viewing the selected flame
along two lines-of-sight which have a region of intersection in the
flame to produce two electrical output signals each having
alternating components corresponding to the varying intensity of
the radiation from the flame and determining the degree of phase
correlation of the alternating components of the two electrical
signals.
2. Apparatus for monitoring a selected flame in a multi-burner
furnace comprising two photo-electric pick-ups arranged for viewing
the selected flame along two lines-of-sight which have a region of
intersection in the flame, each pick-up including a photo-sensor
producing an electrical output signal having alternating components
corresponding to the varying intensity of the radiation from the
flame, and signal processing means connected to said photo-sensors
to be responsive to said output signals and arranged to determine
the degree of phase correlation of said alternating components of
the electrical output signals from the two pick-ups.
3. Apparatus as claimed in claim 2 wherein said each of said
photo-electric pick-ups comprises a photo-sensor mounted in a tube,
and two tubes being arranged to collimate the respective light
paths with angles of acceptance at the photo-sensors smaller than
the angle between the tubes.
4. Apparatus as claimed in claim 3 wherein each said tube contains
at least one optical element to form an optical image forming
system.
5. Apparatus as claimed in claim 4 and having an air supply source
and air bleed means in said tube to cause a flow of air over said
at least one optical element.
6. Apparatus as claimed in claim 2 wherein said signal processing
means comprises means determining, over a time period, the fraction
of the total time for which the two output signals have the same
polarity relative to their respective mean values.
7. Apparatus for monitoring a selected flame in a multi-burner
furnace comprising two photo-electric pick-ups arranged for viewing
the selected flame along two lines-of-sight having a region of
intersection in the flame, said photo-electric pick-ups each
including a photo-sensor said photo providing an electric output
signal corresponding to the intensity of radiation incident on the
pick-up, a processor coupled to said photo-sensors to process the
signals therefrom the arranged to give the quotient: ##SPC4##
where x and y are the amplitudes of the output signals from the
photo-sensors, and x and y are the mean values of x and y, and a
low pass filter for smoothing the output of the processor.
8. Apparatus as claimed in claim 7 wherein said processor comprises
a pair of amplifiers coupled respectively to the two photo-sensors
to produce the outputs (x - x) and (y - y) respectively, a pair of
full wave rectifiers coupled respectively to said amplifiers to
produce the signals .vertline.(x - x).vertline. and .vertline.(y -
y).vertline. respectively, a first multiplier coupled to said
amplifiers to provide an output proportional to the product of the
outputs of the amplifiers, a second multiplier coupled to said full
wave rectifiers to produce an output proportional to the product of
the outputs of the rectifiers, and a divider coupled to said
multipliers to the output proportional to the ratio of the
products.
9. Apparatus as claimed in claim 7 wherein said processor comprises
a pair of filters and amplifiers coupled respectively to the two
photo-sensors to produce the signals (x - x) and (y - y)
respectively, a pair of zero crossing detectors to produce voltages
corresponding to the sign of (x - x) and the sign of (y - y)
respectively, a logic circuit to produce one output if these
voltages are of similar sign and a different output if they are of
dissimilar sign, a digital-to-analogue converter to stabilise these
outputs from the logic circuit and lowpass filter to integrate
these stabilised outputs.
10. A method of monitoring a selected flame in a multi-burner
furnace comprising photo-electrically viewing the selected flame
along two lines-of-sight which have a region of intersection in the
flame to produce two electrical output signals, each having
alternating components dependent on the varying intensity of the
radiation from the flame, and determining, over a time period, the
fraction of the total time for which the two output signals have
the same polarity relative to their respective mean levels.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a method of and apparatus for monitoring
flames in multi-burner furnaces, such as in boilers for large
electrical power generating stations.
2. Prior Art
For a long time the desirability of automatically monitoring flames
in a boiler has been recognised. If fuel continues to be supplied
to a burner after the flame has been extinguished, the fuel may
re-ignite explosively. A human observer can identify a particular
flame but continuous human observation is expensive in manpower,
even if confined to critical periods. The problem arises
particularly with the growing demand for automatic start-up of
boilers. Reference may be made to "Flame Monitoring-Important Key
to Unit Automation" Power 104 Oct. 1960, "Combustion Monitoring of
Large Flames from Flame Spectra" Baumgartel and Hage A.S.M.E. Paper
No. 61-WA-299 and "Photo-Electric Supervision of Oil and Gas
Burners of Steam Generating Plant" Brinke Tech. Uberwach Vol. 5,
No. 4 1964.
Many systems intended for flame detection have been produced.
Simple detectors of optical radiation (infra-red, visible or
ultra-violet) give false indications that a flame is present when
they receive radiation from something else of comparable
brightness, such as the furnace walls or part of a neighbouring
flame. In a large boiler, there are many burners and it is
difficult to have a line of sight through only one flame,
particularly bearing in mind that the form of the flame depends on
the fuel supply and other factors. For these reasons, although
radiation amplitude sensitive devices may work satisfactorily in a
single burner furnace, the high level of background radiation
precludes the use of such devices in a multi-burner furnace.
Detectors have therefore been proposed to utilise particular
properties of a flame. One, for example, is a differential system
in which two photo-electric cells are positioned to view dark and
light areas of the flame respectively; such a detector is
applicable to coal-fired flames in which there is a dark area where
the pulverised coal emerges from the burner. Such a detector
however is very sensitive to changes in combustion conditions which
alter the distance along the flame axis at which the fuel ignites.
Also with such a differential flame detector, blocking of one of
the viewing tubes could cause a differential signal even when the
flame is absent.
Another detection system makes use of the alternating component,
known as flicker, which, in all large flames, is superimposed on
the steady radiation. Flicker amplitude decreases progressively
along the flame axis so that the photo-detection system has to be
aimed near the root of the flame. This gives a considerably
improved discrimination but, with the wide variation in combustion
conditions, as occur for example on start-up and loading changing,
it is not possible to achieve a confident distinction between
flame-on and flame-off conditions.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide an improved
method of and apparatus for monitoring any selected flame in a
multi-burner furnace.
According to this invention, a method of monitoring a selected
frame in a multi-burner furnace comprises photo-electrically
viewing the selected flame along two lines-of-sight which intersect
in or near the flame to produce electrical signal outputs and
determining the degree of correlation of the alternating components
of the two electrical signals.
According to a further aspect of this invention, apparatus for
monitoring a selected flame in a multi-burner furnace comprises two
photo-electric pick-ups arranged for viewing the selected flame
along two lines-of-sight which intersect in or near the flame, and
means responsive to or determining the degree of correlation of the
alternating components of the electrical output signals from the
two pick-ups.
It is to be expected that the alternating components of signals
which arise from sources at the point of intersection will show
high cross-correlation while those arising elsewhere in the furnace
should be uncorrelated. This technique depends essentially on the
existence of alternating components (flicker) in the optical
signals from the flame. The origin of such components is not firmly
established. They may be attributed to turbulence in the flame or
to fluctuations in the fuel supply. At the root of the flame, it is
to be expected that turbulence will distort the flame front, giving
rise to a large variation in speed and direction of propagation and
this may well cause the flame to flicker. Along the length of the
flame, the flicker could be influenced by some resonance in the
furnace. With large turbulence, some of the unburnt fuel may be
carried into the hotter regions of the flame where it suddenly
ignites propagating a region of hotter gas along the length of the
flame.
As indicated above, one purpose of flame monitoring is to determine
if a flame is present. If the flame monitor is used for control
purposes, action may be required within a short time, e.g., 1
second, and this would set a lower limit to the frequencies which
can usefully be correlated. If action is required within one
second, frequencies below 10 Hz would be of little value. More
generally however, the pick-ups may be arranged to provide signals
over a wide frequency band. Flicker signals can readily be detected
up to at least 700 Hz. Filters can be provided for removing
electrical signals at frequencies at which correlation is small
even when a flame is present. The optimum frequency band may be
determined empirically.
The photo-electric pick-ups may be sensitive over a wide range of
optical wavelength or may be made selective.
The divergence of the viewing beams may be chosen in accordance
with the particular conditions to be monitored. A wide angle beam
will give greater light gathering but overlap outside the space
occupied by the flame to be monitored will reduce the possible
correlation. If too large a region of incoherently flickering flame
is observed, the high frequency components of the signals will be
relatively smaller.
The angle .theta. of acceptance of light into each detector and the
angle .phi. between their respective optic axes must be considered
in relation to one another. There are advantages in keeping .phi.
small since it gives a greater length over which correlated signals
can arise. Also it makes it possible to have both optical axes
passing through one port in the furnace wall, so simplifying
installation. On the other hand .phi. must be greater than .theta.
or the region of overlap between the two viewing beams will extend
to infinity. It is thus advantageous for .theta. to be small
provided enough light is gathered so that amplifier noise will not
dominate the optical signal.
Alternatively an optical system with a lens or lenses may be used.
In the case of a single lens, the diameter of the lens, its focal
length, the distance between the lens and an aperture in front of
the sensor, and the shape and size of the aperture control the
geometry of the light path.
Each pick-up conveniently comprises a photo-sensor mounted in a
tube which serves to collimate the light path. In practice, the
photo-sensors may be mounted in long metal tubes which extend into
the furnace; such a tube may for example be 1-2 metres long to give
protection for the sensor and to collimate the light path. A tube
with an internal diameter of 16 mm and 1.5 m long will make .theta.
about 1.2.degree.. Two such tubes can be pointed at a flame near
its root and set to give maximum correlation.
Correlators for correlating electrical signals are commercially
available for giving an output representative of the correlation
factor between two alternating input signals. For a simple monitor
to indicate the presence or absence of a flame, it is not necessary
to measure the correlation factor; it may be adequate to give an
output signal which is related to the correlation.
If the two photo-cell outputs, which are functions of time t, are
x(t) and y(t) respectively, the correlation coefficient .rho.x y
between the two photo-cell outputs may be defined as ##SPC1##
where
T is the time period over which the signals are correlated,
.gamma. is a dummy variable of integration representing time,
and
x and y are the alternating components of the signals from the
sensors having zero average values.
Commercial instruments are available which operate on two inputs x
and y to give an output of .rho. x y. Thus a commercial correlation
instrument might be used, the two photo-sensor outputs being used
as the x and y inputs respectively. If either x(t) or y(t)
continues at zero, the above expression for the correlation
coefficient becomes indeterminate and thus it would be necessary to
check that a finite signal x(t) and y(t) was present in the
respective channels. However for a simple monitor to detect the
presence of a flame, the numerator of the above expression could be
used by itself. The fact that it exceeded some threshold value
would be taken as an indication that a flame was present.
In effecting the computation using analogue techniques,
considerable saving in cost can be made by making a simpler
computation given an approximate measure of correlation. In one
form of the invention, the means for determining the degree of
correlation between the signals from the line photo-sensors
comprises a processor coupled to the photo-sensors to process the
signals therefrom and arranged effectively to give the quotient:
##SPC2##
the output of the processor being smoothed by a low pass filter
(typically with a time constant of 0.5 seconds), x and y being mean
values of x and y respectively. This circuit employs only two
multipliers and a divider. No square rooting operations are
employed. The use of a low pass filter, i.e., a smoothing circuit,
with a single time lag in effect provides a continuous average so
avoiding any necessity for sampling. This processor is independent
of signal amplitudes but gives an approximate measure of
correlation since the circuit only compares the instantaneous
polarity of the input signals. The errors introduced however are
not substantial and can be discounted for flame monitoring because,
in the correlated case, the two waveforms will be identical in
shape because they come from the same source.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a diagrammatic plan view of a flame monitoring apparatus
for a multi-burner furnace; and
FIG. 2 is a block diagram illustrating signal correlating means
used in the apparatus of FIG. 1.
DESCRIPTION OF THE PREFERRED EMBODIMENT
Referring to FIG. 1, there are shown two optical image forming
systems 10, 11 each arranged to form, at a photo-sensor 12, an
optical image of a flame front. Each image forming system comprises
a tubular mount indicated diagrammatically at 13 which supports the
photo-sensor 12, a line 14 and a tubular light stop system 15. This
tubular light stop is arranged to give a light acceptance angle of
just over 1.degree.. Typically for a large furnace the tube is
about 1.5 metres long and 15-20 mm in diameter. The two tubes are
mounted in an aperture in the furnace wall at an angle to one
another, which angle must be larger than the acceptance angle of
the tubes, such that they both accept light from the same portion
of a particular flame which is to be monitored.
For clarity in FIG. 1, the acceptance angles of the two image
forming systems 10, 11 and the angle between their axes have been
greatly exaggerated and the overlap region of the two acceptance
regions has been made close to the ends of the tubes. It will be
seen that the overlap region, by making the angle between the axes
small, can be made to have an appreciable depth into the furnace.
The position of a burner flame in a furnace depends on the air/fuel
ratio and fuel supply. The image forming systems are arranged so
that the overlap region covers the required range of possible
positions of the flame to be monitored. It will be noted, however,
that the proportion of the light acceptance regions which overlap
decreases towards the inner and outer limits of overlap region and
hence the degree of correlation of viewed flames will decrease
towards these limits.
For pulverised coal or oil fired burners, the flames are
appreciably opaque. In these cases, the flame front is readily
observed. A silicon photo-voltaic photo-sensor having a wide
optical bandwidth is conveniently employed with such flames. For a
natural gas flame, it is preferred to utilise the longer wavelength
radiation, particularly in the infra-red region of the spectrum,
where the flame is sufficiently opaque to prevent light
interference from flames behind the flame being viewed. For such
flames, lead sulphide or indium antimonide photo-conductive
elements could be used to give output signals dependent more
particularly on the infra-red radiation.
The electrical outputs from the two photo-sensors 12 are fed to a
signal correlator 16 which is illustrated in further detail in FIG.
2.
Although simple image forming systems have been shown in FIG. 1, it
is possible to use more complex telescopes, for example Newtonian
or Cassegrain systems to give the required narrow acceptance
angle.
In the image forming system, an air supply may be applied to a
bleed hole 17 to give a small air flow through the tube into the
furnace to keep the optical system clear. Other known techniques
however may be employed to reduce or prevent the ingress of any
smoke and solid particles from the furnace into the tube.
It will be appreciated that the two image forming systems may be
arranged in any place such that they can have acceptance regions
overlapping in the position of the flame to be monitored. Separate
systems would be provided for each flame in the furnace which is to
be monitored.
The signal correlator of FIG. 2 is an amplitude insensitive
correlator. It includes a simple processor which processes the
instantaneous values and moduli of the two inputs and which
effectively gives the quotient: ##SPC3##
where x and y are the photo-sensor inputs and where x and y are
their mean values. The output of this processor is smoothed by a
low pass filter, with a time constant, in this embodiment, of 0.5
seconds.
In FIG. 2, the inputs x and y from the photo-sensors are fed via
leads 18 and 19 and scaling potentiometers 20, 21 respectively to
two band-pass filters 22, 23 which typically have a pass band of
20-1000 Hz. The signals from these band pass filters are fed
respectively to two separate amplifiers 24, 25 to produce outputs
having amplitudes (x - x) and (y - y). These two signals are fed,
in the first place to a signal multiplier 26 which multiplies the
amplitudes of the two signals and which is followed by a scaling
amplifier 27 to give the signal (x - x) (y - y). This in turn is
fed to one input of a divider or ratio determining circuit 28. The
outputs from the amplifiers 24, 25 are also separately fed to two
linear full wave rectifier circuits 29, 30, each consisting of two
operational amplifiers 31 and 32, to provide outputs .vertline.(x -
x).vertline. and .vertline.(y - y).vertline.. The amplifier 31
forms a rectifier giving the modulus of the input signal and the
amplifier 32 is a scaling amplifier. From the amplifier 31, the two
signals are multiplied in a multiplier 33 and thence passed via a
scaling amplifier 34 to the second input of the divider 28. The
output of the divider 28 is smoothed by a low pass filter circuit
35 with a time constant of 0.5 seconds to give the required output
on a lead 36. The signal may be fed to an indicator and/or recorder
as indicated at 37. More typically however it would be used for
control purposes as indicated at 38, e.g., to shut-off the fuel
supply if the cessation of a flame is detected.
It will be seen that this correlation circuit employs only two
multipliers. No squaring or square-rooting operations are
required.
It is found that with a circuit such as is shown in FIG. 2, it is
possible to distinguish, by means of the level of the output
signal, not only whether there is a normal flame or no flame but
also the intermediate condition of an abnormal flame. This can
arise, because, as previously explained, the degree of correlation
will depend on the position of the flame. The flame position
depends, for example, on the air to fuel ratio and hence an
abnormal flame due to an incorrect air to fuel ratio may give a
magnitude of output on lead 36 intermediate between substantially
complete correlation and no correlation. Provision may be made
therefore for operating an alarm or control device if an output
signal is an intermediate amplitude range.
Other circuits could also be employed giving outputs related to the
correlation coefficient between the two signals, for instance a
digital processor incorporating zero-crossing detectors and a logic
circuit which produces a positive voltage level if the signs of the
input signals are the same and an equivalent negative voltage level
if the signs of the input signals are not the same. These voltages
from the logic circuit would be stabilised by a digital-to-analogue
converter and then integrated by a low-pass filter.
* * * * *