BOILER EFFICIENCY AND NO x OPTIMISATION THROUGH ADVANCED MONITORING AND CONTROL OF LOCAL COMBUSTION CONDITIONS

Transcription

1 BOILER EFFICIENCY AND NO x OPTIMISATION THROUGH ADVANCED MONITORING AND CONTROL OF LOCAL COMBUSTION CONDITIONS PAPER PRESENTED AT THE SIXTH INTERNATIONAL CONFERENCE ON TECHNOLOGIES AND COMBUSTION FOR A CLEAN ENVIRONMENT JULY 2001, OPORTO (PORTUGAL)

2 Boiler Efficiency and NO x Optimisation through Advanced Monitoring and Control of Local Combustion Conditions A. COPADO, F. RODRÍGUEZ INERCO Parque Tecnológico de la Cartuja, C/ Tomás Alba Edison, s/n, Sevilla, Spain acopado@inerco.es L. CAÑADAS, V. CORTÉS AICIA Camino de los Descubrimientos, s/n, Sevilla, Spain P. GÓMEZ ENDESA, Compostilla P.S. Ctra. Cubillos del Sil, s/n, Ponferrada León, Spain E. PÉREZ-SANTOS ENDESA, R&D Marqués de Villamagna, nº 3, Edif. Torre Serrano, Madrid, Spain ABSTRACT This paper presents the methodology, development and results of a power station optimisation programme performed by INERCO, AICIA and ENDESA. The overall aims of this programme have been the improvement of boiler efficiency and the reduction of the environmental impact associated to NO X emissions, by means of the following specifically developed tools: - Advanced automatic monitoring system of local combustion conditions (OPTICOM system). This system enables the collection and analysis of gas samples from the boiler furnace that provides on-line measurements of local furnace parameters as temperature and gas concentrations (O 2, CO, NO x, SO 2). - Signal quality assurance system. This software package detects anomalous values of measured variables, substituting them for the best allowable estimations. - On-line calculation and supervision system of unit energy efficiency. This system continuously displays the global boiler efficiency and the performance of significant parts of the installation, providing a reliable evaluation of the influence of any parameter adjustment in the overall facility performance. - An advisory and modelling software package for optimisation of boiler efficiency and NO X emissions (SIRE system). Key Words: combustion, emissions, optimisation, monitoring, software. INTRODUCTION The combustion process in large power plant boilers is almost a black box for the operators. It is significant that in an activity in which the largest production cost is the cost of fuel, the lack of information about the process involved is greatest precisely in the area of how this fuel is utilised. In this respect, the limited degree of boiler monitoring is largely related to combustion conditions inside the furnace, and to the adequate characterisation of fuel and air supplies (Scott (1995); Rodríguez (1998)). This limitation concerning monitoring of the boilers, which has a significant effect on their optimisation, is caused by the fact that the majority of these power plants were designed at a time when the plant s production (and not its emissions or efficiency) was the critical operating parameter. The inadequate monitoring described means that, in general, the operation of the boilers is based on the use of certain combinations of global or indirect variables, derived either from the recommendations of the boiler supplier or from the accumulated experience of the operators of each particular facility. These combinations frequently have more to do with operational stability and historical inertia, i.e. following customary practices, than with true optimum operating conditions. A clear example of this situation is the adjustment of combustion air. The minimisation of excess air is, in fact, one of the most direct and effective primary measures (combustion regulation adjustments) for optimising performance and NO x emissions in any type of boiler. However, boiler operators are extremely loath to use this type of adjustment, due to the possible creation of substoichiometric areas in the furnace which may cause high levels of unburnt fuel or even a plant shutdown. Therefore, relatively high base levels of excess air are habitually used, in spite of its negative effect on heat rate and on the generation of NO x, with priority being given to considerations of operational safety. However, this critical parameter is usually calculated as the average of measurements taken at only 4 or 6 points in the boiler outlet, whose representativeness, with regard to the combustion conditions inside the furnace near each burner, is very limited. A similar case could be made for the monitoring of air and fuel feed to the furnace, and the control of these as a means of optimising the combustion process. In this context, and taking into account the increasing competitiveness among power plants, INERCO, a Spanish engineering and consultancy firm, has been researching a new approach to optimisation of heat rate and NO x emissions in coalfired power plants. This research has been carried out with the collaboration of ENDESA, the leading electric company in Spain, and AICIA, a research centre linked to the University of Seville. It has also been partially financed by the ECSC. The work has focused on the evaluation and development of the most appropriate computer tools for the optimisation process and on the analysis and experimentation of new systems for characterising and monitoring the operation of boilers. A new philosophy is proposed that relies on furnace specific monitoring systems and the computer architecture shown in Figure 1. According to this structure the global system comprises the following components: - Monitoring systems and instrumentation specifically installed for the optimisation process. The use of these specific measurement systems is one of the novelties of the new optimisation approach proposed. Specially of note is the application of the OPTICOM system. - A Control System responsible for information exchanges. This module receives and stores data from: - operation, - coal analysis, - emission measurements and - mill classifier and combustion air damper positions, and includes an operating signal quality assurance system. - An on-line calculation and supervision system (CESAR system), which provides a reliable assessment of the unit s performance. This module includes the program libraries necessary for calculating the boiler local and global parameters. - SIRE system. This module advises plant operators on decision-making for optimised boiler settings. It allows joint on-line optimisation of the performance/no x equation

3 for any boiler operating condition and makes use of cutting-edge modelling and mathematical algorithms. III) IV) determined in power stations, these models allow the system to predict the airflows to the furnace depending on the damper position. In order to monitor the combustion flue-gas flow, meters were installed to determine the carbon in-ash and gas distribution at the boiler outlet (Otero et al. (1999)). A final element influencing performance was the imbalance level of gas distribution inside the boiler. Information on in-furnace imbalances was supplied by a novel proprietary monitoring system, the OPTICOM system, which allows the identification of local malfunctions at each burner. Figure 1 Optimisation system structure 5. Carbon in-ash and gas distribution at boiler exit 3. Coal flow measurement This article describes the main aspects of the research carried out to develop these systems and their characteristics, with particular emphasis on the following aspects: - New monitoring tools. Activities related to the characterisation and monitoring of the plant. Detailed description of the OPTICOM system. - Signal quality assurance system. - On-line calculation and supervision system (CESAR system). - Optimisation system of boiler efficiency and NO x emissions (SIRE system). Different philosophies behind the modelling and optimisation module. Mathematical and computer tools implemented. Finally, the article describes the results obtained by the application of this development at a full-scale Spanish power plant and author s future plans in the field of optimisation. MONITORING SYSTEMS One of the main problems faced by optimisation systems is the suitability to the specific characteristics of the plant. In fact, experiences (Cañadas et al. (1997); Rodríguez (1998)) in off-line optimisation programmes show the need to apply this type of software only after having a prior in-depth knowledge of the plant operation. To solve this problem, it was planned to develop additional online and off-line monitoring systems and instrumentation needed to improve boiler monitoring levels. Figure 2 shows several examples of advanced monitoring systems implemented in an arch-fired boiler. These systems were installed on the basis of recommendations obtained from an off-line diagnosis carried out on one of the boilers of Compostilla II Power Station, belonging to ENDESA. As it can be seen in Figure 2, the boiler was equipped with monitoring systems for the furnace, coal and air supplies and flue-gas at the boiler outlet. I) With respect to the furnace monitoring, the off-line combustion diagnosis showed that the flame height in this type of boiler had considerable influence on plant performance and NO x emissions (Cañadas et al. (1997)). In order to characterise the type of flame, two infrared pyrometers were installed, one in the ash hopper and the other at the end of the radiant area, to detect the effects derived from this situation. II) Imbalances in air and coal supplies also have an especially negative effect on boiler performance (Yan et al. (1998); Vierstra et al. (1998)). Information on coal supply imbalances was obtained by a novel proprietary semi-automated pulverised coal sampling system (EMIR II), which provides reliable information on the coal flow to each burner. To detect airflow imbalances, a windbox monitoring system based on fluid dynamic models was used. Taking into account that air supplies are not usually on-line 1. Temperature of gases (infrared pyrometers): Detection of height of flame Figure 2 Specific instrumentation for optimisation 2. Monitoring of imbalance levels inside the boiler 4. Wind-box monitoring system: Prediction of air flow supplies into the furnace OPTICOM System Description OPTICOM technology has been developed by AICIA and INERCO, with the aid of partial financing from the European Coal and Steel Community (ECSC) and the collaboration of ENDESA s Compostilla Thermal Power Station, in the industrial validation of the system. The OPTICOM technology consists essentially of a system which allows measurements to be taken in any area of the interior of the furnace of industrial boilers, especially near the burners. Some examples of this type of measurement are the evaluation of local levels of gas concentrations, temperatures, heat flows, and even image capture, in these areas of high temperature and very limited access in boilers of traditional design. The aim of these measurements is to identify the combustion conditions at any particular point inside the boilers, in order to be able to optimise heat rate, auxiliaries consumption, generation of pollutants and slagging (Soud (1999)). This local information makes it possible to consider the unit as a collection of small virtual units, each made up of a single burner. The targeted adjustment of each of these smaller units results in overall optimisation of the boiler. In order to meet these goals, OPTICOM allows these measurements to be made through small openings made in the fins which join the tubes making up the water walls of membrane tube boilers (this approach would not be directly applicable to boilers of tangent tube construction). These fins are typically between 14 and 22 mm for most boiler designs. Therefore, the width of these openings is limited by the width of the fins themselves, while the height is unlimited, due to the geometry of the fins (Figure 3). This new concept for measurement inside the furnace of industrial boilers, currently being patented, makes it possible to place the openings in any location required, without being limited to those inspection ports included in the original design of the boiler, and permits the direct measurement of combustion conditions inside the furnace. In this way, it is possible to take measurements at the level of each burner in the boiler, without any significant structural modifications to the unit. In order to access the inside of the furnace, a extremely narrow watercooled probe has been specially designed for insertion through the openings described above. This probe (10-12 mm width) is able to withstand, in addition, the high temperatures ( ºC) in these areas of the furnace.

4 II) Measurement of imbalances in the air and coal supplies (based on O 2 and CO mapping). III) Possibility of optimisation in scenarios with high variability of fuels (by associating SO 2 results to coal S content for each burner), including cases in which fuels of different characteristics are supplied to each burner. IV) Measurement of the type and stability of the flames (through NO x, temperature and signal variability evaluations). V) Correct determination of the real levels of excess air for the adjustment of combustion. Figure 3 Scheme of the OPTICOM technology A sketch of the automation of the OPTICOM system is given in Figure 4. This figure shows the narrow water-cooled probe is coupled to a pneumatic cylinder, controlled by a multi-way valve, to insert or retract the probe in or out of the boiler furnace. The sample collected by the probe passes through a heated filter to retain fly ash and then goes to a set of valves which either send the sample to the conditioning system, or send pressurised air towards the heated filter and the probe, giving rise to a reverse flow which cleans both of these items. Figure 5 OPTICOM measurement of existing combustion profiles in an arch-fired boiler before and after local flame tuning Figure 4 Scheme of the automated OPTICOM system A programmable logic controller (PLC) is used to collect the analysis results, monitor the entire process and report on possible incidents. This information is sent to the Control System, responsible for storing and transferring data, or to a control room computer with the necessary user interfaces. The OPTICOM measurements can be used as additional information for the plant operators and as inputs to the SIRE system. Figure 5 shows two OPTICOM monitoring screens, presenting the profiles of gases from each burner, before and after flame tuning. Using this automated OPTICOM system it is possible to obtain measurements of gases and temperatures in a boiler with 12 burners per wall in around 30 minutes. The advantage of automation is not only that no human resources are needed to take the measurements, but also that the periodically obtained results can be processed by the plant Supervision System, in order to provide for correct optimisation of the unit. OPTICOM technology has a number of advantages with respect to other traditional combustion monitoring systems. These advantages, common to any type of boiler design, may be summed up in the following five areas: I) Direct measurement in any part of the boiler furnace. SIGNAL QUALITY ASSURANCE SYSTEM All industrial facilities need to analyse the signals received and the plant instrumentation which generates them, and have a system for guaranteeing the quality of these signals, ensuring their reliability on-line. The signal quality assurance system developed by INERCO, allows for filtering of measured values for the fundamental boiler and cycle variables, replacing those which do not pass validation tests by the best available estimates. Its function is to detect erroneous measurements and eliminate sources of error associated with processes for determining performance, heat rate and operating parameters. Self-organising maps have been used as the ideal tool for signal filtering and quality assurance. A self-organising map is a type of neural network able to classify a set of patterns in various groups, each characterised by some property which can not be identified by simple statistical analysis. In this way, given a particular input pattern, the map is able to decide which class it belongs to and even filter the most probable values the pattern should have depending on the assigned group. The system uses different ways to validate the measured and calculated signals, replacing erroneous inputs with the expected correct values (Figure 6). Firstly, all the measured signals are fed

5 into a map, which checks if any of the values is out of the range determined by the rest of the inputs. Afterwards the signals are passed through a second statistical and coherence filter to assure their goodness. Finally, a third statistical and coherence filter is applied to the calculated values. Measured Signals N E U R A L N E T W O R K Expected Measured Values MEASUREMENT STATISTICS AND COHERENCE Expected Calculated Values Validated Measured Values CALCULATION SYSTEM CALCULATION STATISTICS AND COHERENCE Calculated Values Figure 6 Scheme of the signal quality assurance system Validated Results ON-LINE CALCULATION AND SUPERVISION SYSTEM (CESAR) The on-line calculation and supervision system (CESAR) allows to improve the heat rate of the power units, as a fundamental factor in the production cost reduction policy in power stations. The basis for improving heat rate is the availability of an on-line and reliable evaluation of operating results. The CESAR system provides a power station with the necessary tools to quantify individual efficiencies and overall heat rate of the units in real time, as well as making it possible to supply all the information necessary to generate Operating Reports. CESAR capabilities include: - Continuous evaluation and display of parameters and performance characteristic of the operation of the main pieces of equipment of the units. CESAR evaluates those boiler parameters needed by SIRE. - Calculation of deviations from the operating parameters with respect to pre-established reference conditions (Figure 7). OPTIMISATION SYSTEM OF BOILER EFFICIENCY AND NO X EMISSIONS (SIRE) The previously described monitoring and calculation systems supply a great deal of information on what is happening in the plant. However, these systems do not, in general, allow a direct optimisation of the used resources. An adequate use and treatment of the information by means of computer tools would lead to higher economic benefits for the companies and would favour the conservation of the environment. Additionally, the existing optimisation programs usually present the following limitations (Rodríguez (1998)): - In general, from the modelling point of view, normally there is neither a profound previous characterisation of the facilities nor an adequate instrumentation to monitor the behaviour inside the boiler. - With regard to the quality and response time of these systems, it should be noted that the optimisation and training process is not usually dynamic, and in addition there is no information on the quality and reliability of the recommendations obtained. In this context, INERCO has been researching a new approach to optimisation, and developing a software package, named SIRE, which can help plant operators to improve heat rates and limit NO x emissions. The final aim of this tool is to analyse all the generated information and process the measured data jointly with previously obtained knowledge. Depending on this analysis, the optimisation system can recommend, for instance, the most adequate air damper and mill classifier positions to maximise boiler efficiency. According to this objective and depending on the requirements and the available information at the facilities, three different SIRE systems have been developed with different scopes: - A first SIRE version (OPTINOX) for arch-fired units is currently installed at Compostilla Power Station, belonging to ENDESA (Figure 8). The optimisation system makes basically use of boiler average data, looking for a global operation improvement. Apart from calculated data such as efficiency, SIRE also receives measured information such as furnace temperature from infrared pyrometers, oxygen, load, carbon-in-ash, gas distribution and global imbalance level from OPTICOM. Figure 7 CESAR system: efficiency losses report - Measurement of the effect of any modifications carried out on unit variables, whether deliberate or accidental, on operating results. - Immediate evaluation of the results of parametric sensitivity tests in order to pinpoint optimum operating conditions. - Identification of items which penalise unit performance due to any type of malfunction. - Hierarchical arrangement of energy information in real time for use by different organisational levels. - Economic estimates of results for certain levels. Figure 8 SIRE system: OPTINOX version installed at Compostilla P.S. - A second SIRE version has been created that is mainly based on data from the OPTICOM system. Thanks to the determination of gas concentrations within the furnace, SIRE is able to optimise heat rate, auxiliaries consumption, generation of pollutants and slagging. In this case, the aim of SIRE is to process these local measurements and identify the combustion conditions at any particular point inside the boilers. The local adjustment of each single burner results in overall optimisation of the boiler.

6 - Finally, at present, a new SIRE version is being developed that will integrate both the local and the global optimisation processes of the boiler operation. This system will collect information from the OPTICOM technology and from the available information, namely: pyrometers, air and coal flow meters, carbon-in ash, flue gas distribution, efficiency, etc. Philosophy Behind the ling Process During the development of the project, one of the fundamental aims was to design a system that could be automatically adapted to any boiler operating condition. This objective has been attained thanks to a new modelling philosophy based on the use of different partial sub-models and the calculation of reliability indexes. Firstly it was necessary to classify the measured and calculated variables in three levels: - Collateral variables. They are not controllable by the operator but characterise the surrounding conditions of the plant, i.e. climate, market load demand, fuel characteristics, etc. - Operating variables. They can be managed by the operator in order to optimise the operation of the plant. Examples are excess air, air distribution, etc. - Target variables or functions. This group includes energy performance, NO x emissions and other target functions to be optimised: performance with a limit on emissions and the cost of the performance-ecotax equation due to emissions. Taking into account this classification, the system makes use of different models based on different collateral variable ranges. In other words, if we consider a bidimensional example (Figure 9) with the coal volatile content and the imbalance level as the normalised behaviour variables, there is no way we could find a single model to simulate all the possible boiler conditions with the same acceptable reliability level. To avoid this problem, the modelling approach divides the overall boundary into smaller sub-models just defined in those areas where the boiler has been operating, generating new sub-models as the boiler begins to operate in new regions of the control boundary. additional tests and criteria to select the ideal boundaries for testing. n INDEX : The model needs new tests within this area INDEX : The model does not need new tests within this area Figure 10 Reliability indexes. Bidimensional example In conclusion, thanks to these indexes and the mentioned novel modelling approach, SIRE has become an optimisation system with the following adaptability characteristics: - Any boiler operating condition can be simulated simply by adding new partial models. - The system is easily re-trainable simply by adding specific tests to the old or new models. This training process is controlled by on-line reliability indexes (Figure 11). - The system can directly be applied to other boiler designs. In order to be adapted to a new facility design, SIRE only requires an appropriate prior characterisation of the new plant Coal Volatile 0 33 Not Not Not 0 33 Not Imbalance level 0 66 n Figure 9 scheme. Bidimensional example with coal volatile and imbalance level as collateral variables 1 00 Another novel aspect within the new modelling approach is its ability to dynamically monitor the optimisation process reliability. If we consider one of the models formerly defined by the behaviour variables (Figure 10), there will probably be an irregular distribution of the testing values. The variation between an actual value and the predicted value will be different depending on its location on the model boundary. In order to evaluate this difference, the system is continuously computing an index that reports on the differences of the model used at any time. Areas without dots will have low index values and therefore low reliability predictions must be expected. This index also informs the training system about the need to make Figure 11 SIRE system: Training module Software Characteristics The selection and development of SIRE mathematical and computer tools are the result of an intensive study of the nature and limitations of the problem to be solved: the dynamic optimisation of a highly non-linear process, with numerous variables involved, whose values are not always exact, as indicated by Booth et al. (1997). Cutting-edge computer tools and mathematical techniques have been used with the aim of obtaining a product that is robust, flexible, reliable, easily adaptable to different operating conditions and which gives on-line results. Different artificial neural network (ANN) topologies, genetic algorithms (GA) and fuzzy logic (FL) based on expert systems were used (Figure 12). Other modelling and optimisation techniques were rejected for different reasons (Martin del Brio et at. (1997)): - Fluid dynamic and statistical methods because of their inability to operate in real time.

7 - Simplex because they are unable to adapt to changing situations, they are highly sensitive to noise and are incapable of modelling complex non-linear problems. OPTIMISATION CONTROL SIGNALS EXPERT SYSTEM TO CONTROL OPTIMISATION PROCESS GENETIC ALGORITHM DATA BASE AND INTERFACE DATA PROCESSING MODULE PREDICTION NEURAL NETWORK TARGET SIGNALS EXPERT SYSTEM TO CONTROL TRAINING TESTS Figure 12 SIRE System: General Configuration MODEL PARAMETERS EVALUATION MODEL Algorithms using ANN, GA and FL were designed to avoid the lack of modularity and versatility of existing commercial software. In addition, they were jointly used in order to improve their potential in those aspects in which commercial products have important deficits, such as the initialisation problem or the effective convergence of the algorithms. The following are the most remarkable aspects of the algorithms used for the design and development of the SIRE system: - For the boiler models of the optimisation system, a back propagation network was chosen. This type of ANN allows to simulate a process using a set of representative operation tests captured and pre-processed by the system. The learning algorithm generally associated with this type of network usually presents convergence control difficulties. This disadvantage has been solved by adding: - A new specific random number generator. - A new algorithm that searches for the best starting point to improve the process convergence. - Secondly, genetic algorithms (GA) have been chosen to look for boiler optimum operating conditions, using boiler models defined by neural networks. - Finally, expert systems based on Fuzzy Logic (FL) have been implemented and configured to control the optimisation process, evaluating and filtering those solutions out of the boiler critical operation boundaries. RESULTS AND CONCLUSIONS A boiler optimisation system has been developed based on novel monitoring technologies, cutting-edge mathematical and computer tools (CESAR and SIRE systems) and a Control Module, which includes a signal quality assurance system. Unlike other industrial processes, the monitoring of the combustion of fossil fuels in large boilers is, in general, markedly deficient. The application of novel advanced technology for local monitoring inside the furnace is a new approach, of proven effectiveness, for the performance optimisation of these processes. Specially of note is the application of OPTICOM technology. This monitoring approach has significant advantages with respect to other traditional combustion instrumentation, based on the direct characterisation of furnace conditions that OPTICOM provides. On the other hand, existing software for optimisation usually presents limitations related to response time, adaptability and reliability. The modelling and advisory system developed, SIRE, aims to optimise boiler performance by means of neural networks, genetic algorithms and fuzzy logic. SIRE presents the following advantages: - Any boiler operating condition can be simulated. - The system is easily re-trainable. This training process is automatically controlled by on-line reliability indexes. TRAINING SYSTEM - The system can be applied directly to other boiler designs. In order to be adapted to a new facility design, SIRE only requires an appropriate prior characterisation of the new plant. At present, the optimisation system is operating in an archfired boiler at Compostilla Power Station, belonging to ENDESA. The application of the optimisation system, in the scope of optimisation programmes based on the use of primary measures, has brought about substantial improvements, over 1%, in the overall plant performance, and NO x reductions of around 30%. All of this without having to modify boiler parts or use gas treatment equipment. Application to front-wall and tangentially fired units is currently being assessed, with preliminary results also showing promising prospects for this technology. In order to improve even further the capabilities of the optimisation system, several projects are at present been developed focussed on three specific areas: I) Integration with other complementary monitoring system technologies, such as wind-box pressure, air and coal flow measurements and artificial vision. II) Development of individual input regulation systems for coal and air managed from the Control Room. III) Monitoring software for the local control of combustion in a closed loop (with limited range), based on OPTICOM measurements. The final aim is to achieve the on-line adjustment of the parameters that are critical for the combustion process. In order to reach this objective, the first activity will be to identify the critical variables and the range of reliable safe values inside which the parameters will evolve automatically. ACKNOWLEDGEMENTS The authors gratefully acknowledge the collaboration of the other INERCO, AICIA and ENDESA personnel involved in activities resulting in this paper. This work has been carried out with a partial financial grant from the European Coal and Steel Community. REFERENCES Booth, R.C.; Kosvic, Th.C.; Parikh, N.J. (1997).The Emissions, Operational and Performance Issues of Neural Network Control Applications for Coal-Fired Electric Utility Boilers. EPRI/EPA Megasymposium. Cañadas, L.; Cortés, V.; Rodríguez, F.; Otero, P.; González, J. F. (1997) "NO x Reduction in Arch-Fired Boilers by Parametric Tuning of Operating Conditions". EPRI-EPA Megasymposium. Martín del Brío, B.; Sanz, A (1997). Redes neuronales y sistemas borrosos. Ra-Ma. Otero, P.; Gómez, P.; Albadalejo, J. L.; Rodríguez, F.; Cañadas, L. (1999). Efficiency and environmental improvement programme in Compostilla P.S.. POWERGEN 99-Europe Conference. Pennwell. Rodríguez, F. (1998). Aumento de la Competitividad Energética y Medioambiental de Grandes Instalaciones de Combustión. Química Hoy. Scott, D.H. (1995). Coal Pulverisers-Performance and Safety. IEACR/79, IEA Coal Research. Soud, H.N. (1999). Computers and Air Pollution Control. CCC/17, IEA Coal Research. Vierstra, S.A.; Earley, D. (1998). Balancing Low NO x Burner Airflow Monitors. POWER-GEN INTERNATIONAL 1998 Conference. Yan, Y.; Reed, A. R. (1998). On-Line Flow Measurement of Particulate Solids. A State of the Art Review. Proceedings of 1 st International Symposium on On-Line Flow Measurement of Particulate Solids.