close

Вход

Забыли?

вход по аккаунту

?

Monitoring of 2-D Combustion Temperature Images in a 670 th Utility Boiler and Simulation of its Application in Combustion Control.

код для вставкиСкачать
Dev. Chem. Eng. Mineral Process., 8[3/4),pp.311-322, 2000.
Monitoring of 2-D Combustion Temperature
Images in a 670 t/h Utility Boiler and
Simulation of its Application in Combustion
Control
H.X. Zhou*, S.4. Zhang, Y.-L. Huang and C.-G. Zheng
National Laboratory of Coal Combustion, Huuzhong University of
Science and Technology, Wuhan 430074, P R CHINA
A simple method of monitoring 2 - 0 temperature images from monochromatic gray
images of flames captured from utility boilers, and reference temperatures measured
by two-color pyrometry, are presented in this papez In a 670 t/h utility coal-fired
boilerfurnace, the magnitude of the combustion temperaturefluctuation can reach as
high as 180 K under normal conditions. Furthennore, a two-loop combustion control
strategy was proposed with steam pressure as its outer-loop control variable and the
radiant energy signal as the inner-loop control variable. Simulation on the singleand two-loop control strategies clearly indicates that the two-loop control strategy
can realize a more economical utilization of fuel. It also ensures steady steam
pressure and safer conditions for the superheater surface tubes (not to be overheated) making the boiler respond quickly to the load change.
Introduction
The efficient utilization of energy resources and the protection of the environment are
currently being given increasing attention worldwide. In fossil-fuel-fued power
plants, the necessity of burning lower grade fuel, the requirements for reduced
exhaust emission, demands for improved plant availability, and the desire to minimize
* Author for correspondence.
311
H.-C. Zhou,S.-S. Zhang, Y.-L. H u n g and C.-G. Zheng
plant and system operating costs, etc., all have a direct bearing on combustion
monitoring and control in utility boilers [l]. The monitoring and measurment of
temperature profiles provides important information on the heat released as a result of
chemical reactions [2-81, although some of the combustion temperatures and the
distribution were measured in pulverized coal flames [5-81. For utility boiler
furnaces, Hirano et al. [9] have stated that no appropriate visualization technique have
been established because the furnaces are too large. However, Zhou et al. [lo] has
proposed a simple and easy way to monitor the twodimensional temperature
distribution in a laboratory-scale furnace.
Generally, the mass flow rate of fuel entering a utility boiler is controlled by the
deviation of stream pressure from its set value. The delay and inertia is very large
when controlling the mass flow rate of fuel by varying the steam pressure. Thus
maximizing the control of combustion systems is difficult. In recent combustion
control studies [ 11-12],image processing techniques have been incorporated taking
measurements of the combustion processes from quantities such as volumetric heat
release rate which were estimated directly from the combustion zones.
This paper fist describes the simple and useful method of monitoring 2-D flame
temperature images for utility boilers, and in-situ monitoring experiments were
carried out in a 670 t/h utility boiler. Second, a two-loop combustion control strategy
using the radiant energy signal as intermediate controlled variable is outlined.
Simulation results on the control strategy are also shown. Finally, a discussion on the
above study is presented.
Monitoring of 2-DTemperature Images
Temperature measurements for pulverized coal combustion processes can be based on
Wien’s law of radiation:
312
Monitoring of 2 - 0 Combustion Temperature Images in a Utility Boiler
where
E,(T) is the radiant energy of the combustion flame under the wavelength
a ,for the radiance E ,temperature T , and constants c, and c,.
The difficulty in obtaining 2-D temperature distributions from 2-D flame
illumination images is the use of the two-colour method [ 131 to obtain simultaneously
two monochromatic flame images under different wavelengths. We propose a simple
way to monitor 2-Dflame temperature images based on Wien’s law of radiation [lo].
As shown in Figure 1, a 2-D monochromatic gray image of the flame,
wavelength
a, is obtained through CCD cameras, and the temperature in
direction of the image-formation processes,
py-rometer.
G i , j ,with
%JO,
one
is measured through a two-colour
GiSiis directly proportional to the radiant energy distribution received by
the image-formation elements.
I , .
flame image Gi,j
Figure 1.
Imuge-fonnation and reference temperature measured by two-colour
pyrometer:
313
H.-C. Zhou, S.4.Zhang, Y.-LHung and C.-G. Zheng
In our method, the 2-Dflame temperature image, q,j , is calculated as [ 101:
where Gio,j~is the gray of the image-formation element corresponding to the
direction from which the reference temperature is measured by the two-colour
pyrometer. A significant advantage of this method is that the effects of the adjustment
of camera aperture or the dust covering on the lens on Gi.1 will not change
provided that
&J,O
T,j,
remains constant. This method has been used to monitor 2-D
temperature images from the top of a laboratory-scale furnace fired with pulverized
coal, and the results have shown that this method is a useful tool for combustion
diagnosis [ 101.
Figure 2. The configuration of the boiler furnace and the measurement system.
2-Dtemperature image monitoring experiments has been performed in the furnace
of a utility 670th boiler. Pulverized coal burners are installed in the four comers of
the furnace. The furnace and the monitoring system configuration are shown in
314
Monitoring of 2 - 0 Combustion Temperature Images in a Utility Boiler
Figure 2. The DZHJm flame TV monitoring system was installed in the upper
section of the furnace. The view angle of the camera was 60°, and the deflection of
the lens axis was 20". In the measurements, a monochromatic filter centered at
632nm was placed before the CCD target. The video signal from the TV system was
recorded by computer through a CA6300 card for color image processing.
We developed this new flame image-detecting device with twotolor pyrometer
for measurement and, for convenience, a thermocouple was used to measure a
reference temperature inside the furnace.
2-D temperature images were then
obtained, and one example is shown in Figure 3. As the image sensor was mounted in
a side wall, the central zone with the highest combustion temperature was not located
at the geometrical center of the images.
o
s
~
a
~
s
m
s
m
Figure 3. An example offlame temperature image detected from the furnace.
The variation of the average temperature is shown in Figure 4. It can be seen that
the maximum amplitude change reached nearly 180 K even if the combustion
condition inside the furnace was steady. Also, when the steam pressure increased, the
mass flow rate of fuel was reduced, and it could be observed that the combustion
temperature clearly decreased. When the combustion temperature increased, the
surface temperature of the superheater tubes became higher and the oxygen
concentration in the flue gas reduced. Therefore, the combustion temperature image
can be used to monitor the combustion condition in utility boilers. Furthermore,
315
H . 4 . Zhou, S.4.Bang, Y.-L. Huang and C.-G. Zheng
radiant energy signals transferred from the flame temperature images have a potential
application in combustion control, as discussed below.
13407
1
2
3
4
6
s
7
8
8
10
Number of image (4
Figure 4. Average temperature variedfrom image to image (interval was I1 sec).
Two-loop Combustion Control Strategy
A representative and traditional single-loop combustion control strategy is shown in
Figure 5(a), where Go(s) is the Laplace transfer function from fuel flow rate F(s) to
steam pressure P(s) with a set value of P*(s). D&) is the Laplace transfer function of
the controller. The large delay and inertia of Go(s)leads to uneconomical utilization
of fuel and unsteady pressure of boiler steam.
We propose a new two-loop combustion control strategy as shown in Figure 5(b),
which takes the radiant energy E(s) as an intermediate controlled variable. So the
overall process from F(s) to P(s) is divided into two sub-processes, Gl(s) and Gz(s).
Gl(s) refers to the process from F(s) to E(s), and Gz(s) refer to the process from E(s)
to P(s). In the two-loop control strategy, an inner-loop controller, Dl(s), controls E(s),
and an outer-loop controller,D2(s),controls P(s).
316
Monitoring of 2-D Combustion Temperature Images in a Utility Boiler
Figure 5. Combustioncontrol strategies; (a)the single-loop; (b)the new two-loop.
The average gray of the flame image is directly proportional to the radiant energy
level in a furnace, but some factors could influence the proportional coefficient
between them. These factors include adjustment of the camera aperture and the dust
covering on the surface of the lens of the camera. Therefore, we should extract a
quantity representing the radiant energy level from the temperature image,
q,j ,
obtained from Equation (2), and using Wien's law of radiation:
The radiant energy signal needed in the two-loop control strategy, E , can then be
obtained from averaging all
Ei,j. It is obvious that the
change the radiant energy signal
change of Gi,j will not
E , provided that the combustion condition in the
furnace remains unchanged.
For simplicity, the process G&) and its sub-processes, GI@)and G ~ ( s are
) , all
approximated by one-order or two-order Laplace functions depending on the
317
H.-C. Zhou, S.4 Zhang, Y.-L. Huang and C.-G.Zheng
coordinated control systems of boiler-turbine-generator units. In-situ measurements
were undertaken in the same 670 t/h utility boiler furnace as shown in Figure 2, and
the sampling period, TO,was 4.44 sec. Because the CCD camera and its aperture had
not been changed during the experiments, the gray of every element in the flame
image is also directly proportional to the radiant energy expressed in Equation (3). So
the average gray value in the flame image was taken as the intermediate controlled
variable. The radiation energy obtained from Equation (3) must be used in the
application of this control strategy. The gray value stored in the computer is g-bits, so
the maximum gray is 255 gray-units (GUS). Then the transfer functions can be
summarized as:
P ( s ) 0.068e43.3s
F(s) - s(l+ 180s)
GO(S) = --
P ( s ) - 0.00015e-88.9J
s( 1 + 20s)
E( s)
Gz(s)=--
From Figure 6, the predicted steam pressure from the gray value approached the
measured value. The lack of agreement was due to the simplicity of the functions
used, although it is still useful to demonstrate the application of the two-loop control
strategy qualitatively.
In order to demonstrate the significance of the two-loop control strategy, an
optimal control algorithm was not determined for the controllers but comparison was
made between the two strategies with traditional PID control algorithms. Initially, PI
control parameters for the single-loop control strategy were determined as Kpo =
0.04426,Tio= 1260. Here Kp is the proportional coefficient, and Ti the integral time.
318
Monitoring of 2-0 Combustion Temperature Images in a Utility Boiler
5.0
1
1
-Y
-measured
steam pressure
predicted steam pressure
-average
gray
112.5
3
’
O
h
f
-4
50
la0
140
130
110
5
a
*
110
g
Y
!?
0)
?!
Q
loo
2
,w
12.0
0
150
i50
200
250
80
300
Time (To)
Figure 6. Responses of the steam pressure measured and predicted to the change of
the gray value of &me image, respectively. The sampling period To was 4.44
seconds.
Then PI control parameters of the inner loop of the two-loop control strategy were
determined with Kpl = 0.044269,Ti, = 160. The existence of the inner loop changed
the characteristics of the whole process, so the PI control parameters of the outer loop
was checked again and obtained as Kpz = 25, Ti? = 12000.
In the simulation, the discrete time, TO,was selected as 4.44 seconds, the same as
the sampling period. The steam pressure response to fuel flow rate disturbances with
the two control strategies were simulated. Figure 7 shows the transient response of
steam pressure by single-loop and two-loop control systems to different fuel flow rate
disturbances of +5%, and +30%,respectively.
From Figure 7 even though the fuel flow rate disturbance increased by several
times, the maximum excursion of steam pressure and the transient time by the twoloop control strategy were still better than that of the single-loop control strategy.
There are two advantages of the two-loop strategy. One is the improvement of the
control quality of steam pressure when furnace combustion is disturbed by changes in
mass flow rate and/or the quality of coal fed into the furnace. The other is the
319
H.-C. Zhou, S.-S.
ulang, Y.-L. Huang and C.-G. Zheng
significant amount of coal saved during the transient processes of the furnace
combustion, thus increasing coal utilization efficiency. In addition, provided excess
coal is not fed into the furnace then the maximum combustion temperature can be
maintained at a relatively low level. Thus the possibility of superheater tube failure
due to over-heating is reduced, and the service time of boilers can be extended.
1.4
A
1.2-
I
I
1
1
2 .
3 1.0E
2
E
(I)
I
I
I
I
I
.
0.8-
.
I
I
0.6-
85 .
0.4-
I
I
I
I
I
I
0
-
-two-loop, +30% fuel disturbance
- - - - singlaloop,+5% fuel disturbance
\
\
%
8
!
t
\
\
\
\
\
\
\
\
!
\
\
lo0
Time (5TJ
Figure 7. Transient responses of steam pressure to +30% and +5% disturbances of
the fie1 mass Jlow rate with the two-loop strategy and the single-loop strategy,
respectively.
Simulation of the steam pressure response to a step increase of +1.0 MPa in its set
value by the two control strategies was performed. The transient responses are shown
in Figure 8. The two-loop strategy changes the steam pressure quickly to a new value
compared to the single-loop, and the maximum excursion can be kept within a
narrower band. Therefore, the process controlled by the two-loop strategy is more
responsive to large load changes in the boiler.
320
Monitoring of 2 - 0 Combustion Temperature Images in a Utility Boiler
6
n
2
0
-two-loop strategy
--_-single-loop strategy
1.4
1.2-
.
1.0-
.
0.8-
2
.
a
0.6-
d
0.4-
v
)
5
.
02
*
-
o.o!b.
0
,
.
200
,
4m
.
,
.
MKI
,
800
.
I
lo00
.
1
1200
Time (51,)
Figure 8. Transient response of steam pressure to the same step increase of +1.0 MPa
in its set value with the two-loop strategy and the single-loop strategy, respectively.
Conclusions
A simple and useful method of monitoring 2-D temperature images from
monochromatic flame gray images, and reference temperatures measured by the twocolour pyrometer, is presented in this paper. The advantage of this method is that it
avoids the difficulty of calibration of the absolute radiant energy from the relative
gray values of the flame images. The fluctuation of combustion temperature can be as
high as 180 K under normal conditions.
A two-loop combustion control strategy is proposed as an alternative for
combustion control in utility boilers. The steam pressure still remains as the outerloop controlled variable, and one radiant energy signal detected directly from the
furnace serves as the inner-loop controlled variable. Simulation on the single-loop
and two-loop control strategies clearly shows that the two-loop control strategy can
achieve more economical fuel utilization.
It also ensures more uniform steam
pressure and safer conditions for superheater tubes (to avoid over heating), and more
rapid boiler response to load changes.
321
H.-C. Zhou, S.-S.Zhang, Y.-LH u n g and C.-G.Zheng
Further work is being undertaken to develop a 3-Dvisualization method. These
techniques for combustion process control based on radiant image processing and the
application of the two-loop control strategy will be applied in practice.
Acknowledgment
The authors express their thanks to the National Natural Science Foundation of China
for their continuing support to the present work (No. 59476026,59606007).
References
Kwatny, H.G., and Bauals, J.W. 1986. Modeling of Low Load Dynamics of Coal Frnd Fwnaccs.
Sixth Power Plants Dynamics.Control and Testing Symposium, Knoxville, Tennessee, USA. April
14-16, pp.32.02.
Chigier, N. 1991. Combustion Measurements. Hemisphen Publishing Corporation, p.4.
Yonemuca, M., and Yamamto, Y. 1987. Combustion Diagnosis by Electronic Speckle Pattern
Interferometry Using a CCD Image Sensor, in: h e r Diagnostics and Modeling of Combustion
(linuma, K.. et al. Us.)Springer-Verlag,
.
Berlin Heidelberg, pp.181-186.
Sato, S. 1987. Measurement of Thne-Dimensional Flame Temperature Fields by Holography
Intefemmetry and Computer Tomography, in: Laser Diagnostics and Modeling of Combustion
(Linuma, K,et al Eds.). Springer-Verlag,Berlin Heidelberg, pp.187-194
Carbol, J.. Poticr, B.. and Mac, 0. 1985. Optical Measurement of Local T e m p c r a ~in a Semiinduseial Pulverized Coal Flame by Infrared Spectroscopy, proceedings of SFTE The International
Society for Optical Enacering. Vo1.1341. Published by the Intemational Society for Optical
Engincuing, Bellingham.Washington, USA pp.384-394.
Mitchell, R.E., and McLean, W.J. 1982. On the 'Rmpaahln and Reaction Rate of Burning .
Pulverized Fuels. 19th Int. Symp. on Combustion, The Combustion Institute, Pittsburgh, p.113.
Young, B.C.. McCollor, D.P.. Wcber, BJ.. and Jones, M.L. 1988. Tunpcraturc Measurements of
Beulah Iignite Char in a Novel Laminar Flow Reactor. Fuel, 67, pp.40-44.
Hardesty, D.R., and Ottcsen, D.K. 1987. Optical Diagnosis in-situ for Measurements in Combustion
Environments Containing Coal Particles. Energy (Oxford), 12, pp.813-836.
Hirano, T., et al. 1994. The Potential of Viualization for Studies on Flames and Fumaccs. Fuel,
73(11). pp.1679-1705.
Zhou. H.C. ct al. 1995. On a Monitoring Method of "b-dimmsional Temperature Dishibution in
Furnaces for Combustion Diagnostics. Proccdm
' gs of 3rd Asian-Pacific International Symposium
on Combustion and Energy Utilization, Hong Kong, December 11-15, pp.626-630.
Allen, M.G. et al. 1993. An Jmaging Neural Network Combustion Control System for Utility Boiler
Application. Combustion and Flame,94, pp.205-214.
Padmanabhan, K.T.;Bowman, C.T. and Powell, J.D. 1995. An Adaptive Optimal Combustion
Control Strategy. Combustion and flame, 100. pp.101-110.
Shimoda. N., et al. 1990. prediction of Method of Unbumt Carbon for Utility Boiler Using Image
Processing Technique of Combustion Flame, IEEE Transaction on Energy Conversion, 5(4) 640-645.
-
Документ
Категория
Без категории
Просмотров
0
Размер файла
564 Кб
Теги
monitoring, simulation, application, image, temperature, boilers, combustion, 670, utility, control
1/--страниц
Пожаловаться на содержимое документа