WO1996034233A1

WO1996034233A1 - Method of measuring the amount of pulverized material in a pulverized fuel fired boiler and method of controlling a combustion process

Info

Publication number: WO1996034233A1
Application number: PCT/FI1996/000219
Authority: WO
Inventors: Juhani Hirvonen; Reijo Lilja; Jarmo Nihtinen; Karl Ikonen; Kari Jääskeläinen
Original assignee: Imatran Voima Oy
Priority date: 1995-04-28
Filing date: 1996-04-23
Publication date: 1996-10-31
Also published as: FI100734B; FI952029A0; FI952029A; AU5401196A; EP0823034A1

Abstract

The present invention relates to a method of measuring the amount of pulverized fuel in a pulverized fuel fired furnace (2) and a method of controlling a combustion process in a pulverized fuel fired furnace. According to the invention, the heat radiation emitted by the flame (32) of a pulverized fuel burner (11) is measured over a predefined area (31), the combustion air feed rate (5, α) to the burner is determined simultaneously, and a burner-specific corrective computation (20) is performed to determine the amount of pulverized fuel.

Description

METHOD OF MEASURING THE AMOUNT OF PULVERIZED MATERIAL IN A PULVERIZED FUEL FIRED BOILER AND METHOD OF CONTROLLING A COMBUSTION PROCESS

The invention relates to a method according to the pre- amble of claim 1 for measuring the amount of pulverized material in a pulverized fuel fired boiler.

The invention also concerns a method of controlling air infeed flow rate of a burner on the basis of measuring the amount of pulverized fuel.

Modern power plants are production plants which are oper¬ ated in an automated and centralized manner. Increasing costs of energy production have placed energy producers in front of needs for improving the efficiency of the combustion process and the availability of the power plant as well as reducing maintenance costs. Demands for environmental protection have reduced the permissible emission rates. All these factors have forced power- generating utilities to pay more attention to the combus¬ tion process in their power plants. Although centralized control, automation and different monitoring systems have helped solving the problems, improved combustion process efficiency is requested and emission regulations will become tighter.

To fulfill such tightening demands, attempts are made to improve the combustion process in pulverized fuel fired boilers. Hence, combustion process control is one of the most critical functions in power generation. Today, control and minimization of emissions from power genera¬ tion have become ever more important, and conformity with the requirements of emission standards is a obligatory precondition of continued power generation. If the co - bustion process itself is satisfactorily under control, also the amount of emissions can be significantly lowered and the need for expensive pollutant removal systems reduced. The outcome of combustion process control and tuning is crucially dependent on the process information rendered by the existing instrumentation. While the information available on the flame body itself is the most important source of feedback, the conventional flame monitoring systems are still at a primitive level and serve principally as protective systems.

Nitrogen oxides have been the subject of very active interest and nitrogen compound emissions into the atmosphere are being curtailed to an increasing degree. The conditions of the combustion process have a strong effect on the amount of NOx compounds generated therein. The quantity of obnoxious nitrogen compounds can be reduced substantially by means of staged combustion tech¬ niques. In staged combustion the air is fed into the burner flame so that at the initial stage of the combus¬ tion process the flame is strongly air-deficient thus creating reducing conditions and preventing the formation of nitrogen oxides. Control of the quantities of fuel and combustion air is of utmost importance in staged combustion, because the establishment of reducing conditions requires the ratio of fuel to combustion air to be exactly correct. On the other hand, the flame must have sufficient amount of air to sustain combustion. Controlling the fuel-air ratio is extremely difficult when solid fuels such as coal and peat are burnt in pulverized form. As the pulverized fuel has a varying density and is difficult to feed evenly into the burner, the amount of fuel entering the furnace cannot be controlled in pulverized fuel firing even closely as accurately as when using liquid or gaseous fuels. Because the fuel feed in pulverized fuel firing generally occurs as a volumetric feed, such fuel density variations further cause variations in the mass flow rate of fuel feed via the fuel feed manifold into the furnace. More¬ over, the design of the pulverized fuel feed manifold often fails to distribute the pulverized fuel flow evenly between the burners connected to the mill section. Such variations in the fuel flow distribution are frequently dependent on the load imposed on the mills. Due to lacking methods suitable for continuous measurement of pulverized fuel mass feed rate in a reliable manner, generally the fuel flow is assumed to be uniformly distributed between the burners. Then, the set values of combustion air infeed dampers are typically computed on the basis of the rotational speed of the fuel feeder, and the combustion air flow is distributed uniformly between the burners connected to the mill section. Resultingly, some burners may operate with an air-deficient fuel mix¬ ture while the others are fed with excess air. Burners operated with too much excess air cause unnecessary NOx formation while the burners running air-deficient may pass unburnt fuel into the fly ash. Burners running air- deficient may also present stability problems.

The above-described type of problems associated with pulverized fuel firing are discussed in the following publications and others:

Lawn, CJ. (1987). Principles of Combustion Engi- neerinq for Boilers , Academic Press, pp. 50-58.

Hirvonen J. and Ikonen K. (1992): Image process¬ ing in Combustion Management, IFAC Symposium on Intelligent Components and Instruments for Con- trol Applications, May 20-22, 1992, Malaga,

Spain.

Lilja, R. (1993) . Control of combustion for jet flow burners based on both qualitative and quan- titative models, Technical Research Centre of

Finland, LIEKKI Combustion Research Program, Technical Review 1988-1992, Report L93-1, Abo Akademi University, 1993. Editors Mikko Hupa, Jukka Matinlinna, pp. 105-129.

Nihtinen J. (1993): Flame Image Monitoring and Analysis in Combustion Management. International Symposium on Improved Technology for Fossil Power Plants - New and Retrofit Applications. Electric Power Research Institute, March 1-3, 1993, Washington, D.C.

It is an object of the present invention to achieve a method of determining the amount of pulverized fuel in a flame.

The goal of the invention is accomplished by determining the amount of pulverized fuel fed in a pulverized fuel fired boiler by means of measuring the radiation emitted by the flame over a predefined effective area, simulta¬ neously measuring the amount of air fed into the flame and then determining the amount of pulverized fuel in the flame on the basis or a predetermined relationship.

More specifically, the method according to the invention for determining the amount of pulverized fuel is charac- terized by what is stated in the characterizing part of claim 1.

Furthermore, the method according to the invention for controlling the feed rate of combustion air is character- ized by what is stated in the characterizing part of claim 13.

The invention offers significant benefits.

Based on the use of a furnace camera and real-time image analysis computation, the control system according to the invention facilitates automatic optimization of the com- bustion process state under varying operating conditions. The method according to the invention facilitates effec¬ tive control of a pulverized fuel boiler so that the formation of nitrogen oxides can be minimized and yet achieving maximally complete combustion of the fuel.

Furthermore, burner operation will be more stable than in conventionally controlled firing. Because of the accurate control of the fuel-air ratio, the benefits of staged combustion techniques can be maximally utilized, whereby the formation of nitrogen oxides remains minor.

In the following the invention is described in greater detail with the help of the exemplifying embodiments illustrated in the appended diagrams in which

Figure 1 is a graph representing the irradiance of the burner flame during a change of fuel feed rate in a fuel burning system to which the invention can be applied;

Figure 2 is a block diagram of the burner-specific corrective feed rate control of combustion air in a fuel burning system to which the invention can be applied;

Figure 3 is a graph representing computed energy of radiation from the burner flame as a function of combus¬ tion air and fuel feed rate in a fuel burning system to which the invention can be applied;

Figure 4 is a graph representing the relationship between the burner flame irradiance and the behaviour of the flame ignition point during cyclic change of burner air feed rate in a fuel burning system to which the invention can be applied;

Figure 5 is a block diagram of the principles of burner- specific combustion air control system in a fuel burning system to which the invention can be applied; and Figure 6 is diagrammatic arrangement according to the invention for measuring burner flame irradiance in the method according to the invention.

The most up-to-date information on combustion process it¬ self in the furnace can be obtained by means of properly directed furnace cameras. Today, such furnace cameras are mainly used for monitoring purposes only. However, the flame image information in combination with modelling techniques can as well be extended to furnace burner con¬ trol with the help of modern technology. The basic goals of combustion control are to adjust burner air feed rate and staging so that the amount of unburnt fuel components in the flue gases remains as low as possible, the NOx emissions stay within permissible limits, heat released in the combustion process and heat radiation imposed onto the furnace walls are correctly distributed and local temperatures are correct for sulfur dioxide removal. Undeniably the greatest benefit of properly controlled combustion is the reduction of nitrogen oxides, which can be achieved by means correctly adjusted fuel-air ratio. A control scheme based on combustion process modelling offers a possibility of implementing these demands in an optimal manner within the constraints of each practical situation. The model gives the operator such novel information on the process that was not earlier available by conventional methods. This novel information makes it possible to design such a closed-loop control system that can automatically keep the operating parameters of the combustion process within correct limits. When the model is connected to a furnace camera, it will be possible to monitor any possible variations of the combustion process by means of the model.

Next, two exemplifying embodiments of the invention are described illustrating the utilization of image informa¬ tion in combination with the qualitative model for the control of the combustion process. The object of the first control scheme is to correct the burner-specific air-to-fuel ratios. The object of the second control scheme is to act as model-based fuzzy control.

To obtain real-time information on the combustion process itself, camera monitoring systems for furnaces have been developed of which the commercially available DIMAC system can be mentioned. This system typical of its kind consists of the following hardware elements:

1) The camera system itself comprises burner-specific, air-cooled, solid-state cameras which are mounted to monitor the burner perpendicularly from its side. Each burner to be monitored is provided with a dedicated camera.

2) The burner flame image analysis system comprises camera-specific image-processing cards. While these cards basically operate independently, they may also perform mutual exchange of information and their function is to analyze images delivered by the cameras, whereby each card analyses several images per second and produces real-time information on the combustion process.

3) The burner control system comprises a PC, monitors and a tracking ball. The control system performs retrieval and storage of data delivered by image processing cards. It also displays the results to the operating personnel and incorporates the operator interface and configuration tools.

Example l

This example is related to the hardware illustrated in Fig. 2. Because the fuel feed in pulverized fuel firing generally occurs as volumetric feed, the mass flow rate of the pulverized fuel in the fuel feed manifold 10 varies prior to entering the furnace 2. Moreover, the design of the fuel feed manifold 10 often causes non- uniform distribution of the pulverized fuel flow between the burners 11 connected to the mill section. Such variations in the fuel flow distribution between the burners 11 connected to the mill section are frequently dependent on the load imposed on the mills. Due to the lack of methods suitable for continuous measurement of pulverized fuel mass feed rate in a reliable manner, generally the fuel flow is assumed to be uniformly dis¬ tributed between the burners 11. Then, the set values of combustion air infeed dampers are typically computed on the basis of the rotational speed of the fuel feeder, and the combustion air flow is distributed uniformly between the burners 11 connected to the mill section. Resulting- ly, some of the burners 11 may operate with an air- deficient fuel mixture while the others are fed with excess air. Burners operated with too much excess air cause unnecessary NOx formation while the burners running air-deficiently may pass unburnt fuel into the fly ash. Burners running air-deficient may also present stability problems.

Referring to Fig. 1, the behaviour of the irradiance, or the heat radiation emitted to a unit solid angle of cone, of the burner flame as a result of a change in the fuel feed rate is shown therein. As the flame irradiance is a function of the fuel feed rate and burner air, it can be utilized for correcting the air-to-fuel ratio of the burner. The invention is based on the exploitation of this relationship.

Now referring to Fig. 2, an embodiment of a control sys- tern for correction of combustion air feed rate is shown therein. Individual correction of the air-to-fuel ratio of a burner 11 is made by means of a damper 6 of an air feed duct 9 when the irradiance 15 of the burner flame deviates from its set value. The set value for the burner flame is computed on the basis of burner air excess and fuel feed rate measured at point 5. Because the feed rate of pulverized fuel is generally not measured continuous¬ ly, the speed 7 of the fuel feeder is used as an estimate of the feed rate of fuel into the burner. The dependence of the set value of flame irradiance on the fuel feeder speed 7 and the burner air feed rate measured at point 5 is determined during commissioning at a sufficiently high number of operating points. During the procedure of determining the set value of the irradiance 15, the feed rate of pulverized fuel is measured on a discontinuous sampling basis. The control scheme proposed herein reduces the error of the burner air-to-fuel ratios. The achieved accuracy of the control system is essentially dependent on how precisely the set value of the burner flame irradiance can be determined.

Example 2

The above-described control arrangement of burner air feed rate can be essentially improved by implementing continuous burner-specific pulverized fuel feed rate measurement according to the invention. Again referring to Fig. 1, the system illustrated therein is modified by replacing the fuel feeder speed 7 as the input signal by the fuel mass feed rate signal computed by the furnace model. Computation of the fuel mass feed rate with the help of the model is based on the flame irradiance value obtained from the camera-based machine vision system and the measured burner air feed rate. As is known in the art, for a constant fuel quality, the amount of heat radiation from the flame is strongly dependent on the mutual ratio of fuel feed rate to burner air feed rate. Referring to Fig. 2, the graph shown therein illustrates such a relationship between the heat radiation computed from the furnace model, the fuel and the burner air feed rates. Furthermore, because the irradiance values comput¬ ed from the furnace camera image represent a portion of this heat radiation, the total radiation computed from the model and the irradiance value computed from the furnace camera image are connected to each other by a simple interrelationship, which is also verified in power plant tests. The relationship between the irradiance value obtained from the furnace camera image and the heat radiation computed from the model can be determined with the help of a few tests, after which the model is capable of computing the amount of fuel in actual combustion environment. In the determination of the flame irradiance it is essential that the values used in the model are always computed from a zone whose position stays constant with regard to the flame. Since the flame position rela¬ tive to the camera varies, the zone to be measured is advantageously selected to be an easily definable zone such as, e.g., a zone situated at a predefined distance from the ignition point of the flame. This arrangement eliminates the effect of the local variations in the flame position on the measured irradiance. In Fig. 4 are shown the variations of the irradiance values determined from the flame image and the flame ignition point in a test performed at a power utility plant, whereby the burner air flow rate was cyclically altered at 10 min intervals.

The combustion process is an example of a multiparameter, complex, nonlinear system, whose behaviour is dependent on the mutual relationships between the different param¬ eters. This system can be controlled by means of a method in which an estimator computes from the measurement data, flame image information and predetermined set values the required control signal values for the system controller. Based thereon, the controller retrieves the instantaneous reference value for each control variable, compares it to the set value corresponding to the instantaneous status and issues a recommended value of a possible change for the actuators. The above-disclosed principle can be applied at three levels: control of a single variable, burner-specific control and control of the entire status of the furnace. The control of a single variable in this context refers to a situation in which a single process variable is monitored in order to affect only a single control variable. An example of such a low-level control system is a case in which the total amount of combustion air is controlled on the basis of flue gas oxygen content.

The model-based control task of a high-level system can be formulated as follows: "Determine control variables u, (i=l, 2,..., m) so that state variables x^ (j=l ,2, ... ,n) stay within preset ranges." Referring to Fig. 5, the schematic principle of a burner-specific control system is shown therein. According to the diagram, the model 20 is formed on the basis of camera image information 21 acquired from a furnace 2, whereby the gradients 22 of process variables and the states 23 of each individual burner are computed from the image information. Using the set values 28 as reference, the deviations of the actual state from the target values are computed and the ob¬ tained results are taken to a fuzzy logic control circuit 25, whose output signal is used to form the control sig¬ nals 24. Finally, the combustion air distribution 27 of each individual burner is set on the basis of the control signals. In detail, the operating principle of the control circuit is comprised of the following steps:

1. Determine the state of the combustion process with the help of the furnace camera information 21, the process variable measurement results 22,

23, and the model 20. Let x^ be the state deter¬ mined by simulation (u₂ being given). 2. If the state variables x. (i=l,2, ... ,n) thus obtained fall within preset limits, go to end; if not, go to step 3.

3. With the help of the model 20, determine the state variables x. (i=l,2, ... ,n) using control variable values ά _{l +}, du₁₍_ and du_{i 0}» Here, the measured (actual) values are denoted by index "0", the upward deviated values by index "+" and the downward deviated values by index "-" , respectively. The deviated values are selected so that the control variables still remain within the preset permissible limits. The number of thus obtained state combinations is 3 .

4. Compute the differential changes dx./du,_^

(j=l,2, ... ,n; i=l,2,...,m) for all values of du₁₊ and dU_j. (1=1,2, ... ,m) as in step 3. Using the heuristic rules, select the values of control value derivatives dx./du_x for the next steps.

5. Compute the errors of the state variables x₃.

6. Convert the errors of state variables x. and the approximations of the derivatives dx-/du₁ into fuzzy logic form. Classify the values into distinct size classes.

7. Combine the membership functions of each control variable using the maximum value selection principle.

8. Compute the deterministic values u_x of the control variables from the combined membership functions using the centroid method. It must be noted that the above-described method does not produce a complete arrangement of all alternatives solu¬ tions, whereby the obtained solution may not necessarily be optimal. However, the possibly unfound solutions that computationally may present more optimal values are not generally feasible in practice due to their special character. For the details and computational steps of the above-described model, reference is made to publication:

Lilja R. (1993): Control of combustion for jet flow burners based on both qualitative and quantitative models. Technical Research Centre of Finland, LIEKKI Combustion Research Program, Technical Review 1988-1992, Report L91-1, Abo Akademi University, 1993, Editors Mikko Hupa,

Jukka Matinlinna, pp. 105-129.

For flame monitoring, 1 to 2 images/second are taken from the burner flame, and the image information thus obtained is averaged typically over several minutes. Thence, the measurement results of the amount of pulverized fuel are available by means of the present embodiment according to the invention at a delay of approx. 5 - 10 minutes. If the quality of fuel stays constant, this rate of informa- tion retrieval is sufficient, since burner-specific adjustment in concurrent power plants takes place at intervals of a few hours.

Referring to Fig. 6, a diagrammatic pattern is shown of an image, which is averaged using the above-described method, of a flame 32 in front of a burner 11. The image is divided into constant-width zones 30. The value of effective irradiance can be computed in a reliable manner by first determining the location of the ignition area 33, e.g., as its distance a from the burner 11. Next, from the zones 30 is selected a zone which is at a con¬ stant distance b from the ignition area 33 and, using image processing techniques, for this zone 31 is deter¬ mined its heat radiation intensity which is proportional to the radiation intensity emitted by the flame to a unit solid angle of cone, usually called the irradiance of the flame. Obviously, this zone must be selected to be within the flame. Also the burner air feed rate can be estimated by means of image processing techniques. The tangential angle α of the flame side contour is proportional to the combustion air feed rate so that the smaller the tangen- tial angle α the larger the air feed rate into the burner 11. As the air feed into the furnace takes place in a staged manner, the amount of air fed into any individual burner cannot be determined reliably in a simple manner by measuring the total air feed rate into the furnace and the primary air fed into the burner. Hence, the amount of combustion air should be determined from the flame itself, and here, the tangential angle measurement of the flame side contour gives a sufficiently accurate estimate of the combustion air feed rate.

The irradiance values of the flame can be determined for the entire flame area in the form of a matrix, whose row and column dimensions are selected according to the desired accuracy of computation. The reference value of irradiation can be determined for a selected point of the matrix, or alternatively, a greater number of points can be used for determining the reference value in cases where a single reference point does not give sufficient information on the flame behaviour. While the relation- ship between the pulverized fuel feed rate, burner air and flame irradiance can be determined by experimental methods instead of using the model, such an approach is extremely tedious. Due to the rapid changes of the flame contour, a single analyzed image as such does not usually give usable results, whereby suitable techniques of image averaging must be used for the processing of the measure¬ ment results. This can be implemented by computing the numerical values for each image matrix and then averaging these, or alternatively, adding a number of images into an averaged image for which the numerical values of the matrix are then computed.

The invention is best suited for use in tangentially fired furnaces, because in these furnace designs the cameras are easier to mount to the furnace wall. In wall- fired furnaces only the outermost burners of a burner row can be imaged easily. Usually, while only a single camera is dedicated for each burner, the use of multiple cameras per burner is also feasible.

Claims

Claims :

1. A method of measuring the amount of pulverized fuel in a pulverized fuel fired furnace (2), in which method a mixture of pulverized fuel and combustion air is fed into the furnace (2), where the mixture is burned and the flame of burning fuel-air mixture is monitored by means of at least one furnace camera, c h a r a c t e r ¬ i z e d in that

- the distribution pattern of the radiation emitted by the burning fuel-air mixture flame is measured from the furnace camera image,

- the irradiance value of at least one area of the image having a fixed position relative to the flame is determined from the imaged radiation distribution of the flame,

- the combustion air rate in the flame is determined simultaneously, and

- the feed rate of pulverized fuel into the flame is determined from the measured values of flame irradi- ance and combustion air feed rate.

2. A method as defined in claim 1, c h a r a c t e r ¬ i z e in that the radiation of the flame of a pulver¬ ized fuel burner (11) is measured at a constant distance (b) from the ignition area (33) of the flame (32).

3. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the combustion air feed rate is measured by means of determining the tangen- tial angle (o) of the side contour of the flame (32).

4. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the irradiance value is determined from a flame area which is located at a defined distance from the ignition point of the flame.

5. A method as defined in any of foregoing claims 1 - 4, c h a r a c t e r i z e d in that the irradiance value is determined from a single area of furnace camera image.

6. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the heat radiance intensity matrix is determined from the furnace camera image for the entire area of the flame.

7. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the irradiance value is determined as an average value of a plurality of meas¬ urements.

8. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the flame is imaged from its side.

9. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the relationship between combustion air feed rate, flame irradiance and pulverized fuel feed rate for a given burner and furnace is determined computationally using modelling techniques.

10. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the relationship between combustion air feed rate, flame irradiance and pulverized fuel feed rate for a given burner and furnace is determined using experimental techniques.

11. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the furnace camera is arranged to image the flame of a tangential burner.

12. A method as defined in any foregoing claim, c h a r a c t e r i z e d in that the furnace camera is arranged to image the flame of a wall-mounted burner.

13. A method of controlling the combustion process by means of pulverized fuel feed rate measurement in a pulverized fuel fired furnace (2), in which method a mixture of pulverized fuel and combustion air is fed into the furnace (2), where the mixture is burned and the flame of burning fuel-air mixture is monitored by means of at least one furnace camera, c h a r a c t e r ¬ i z e in that

- the distribution pattern of the heat radiation emitted by the burning fuel-air mixture flame is measured from the furnace camera image,

- the irradiance value of at least one area of the image having a fixed position relative to the flame is determined from the imaged heat radiation distrib- ution of the flame,

- after the determination of the first irradiance value, at least one second irradiance value of the same area of the image is determined and compared with the first irradiance value,

- simultaneously, the combustion air feed rate to the flame is determined,

- the feed rate of pulverized fuel into the flame is determined from the values of flame irradiance and combustion air feed rate determined is said first and second steps, and the feed rate values thus obtained are compared with each other to determine the change in fuel feed rate, and

- the feed rate of combustion air entering the flame is adjusted according to the changes in the pulverized fuel feed.

14. A method as defined in claim 13, c h a r a c - t e r i z e d in that a burner-specific computation (20) is performed on the basis of a first and a second value of irradiance and the value of combustion feed rate to determine the feed rate of pulverized fuel.