To control the quality of coal being sent to the burners located on the furnace walls. The word quality here means the temperature and fineness of the PF. The set temperature values are dependent on the percentage of volatile matter that exists in the main fuel. The controlled temperature is important for many reasons such as stability of ignition, better grindability of solid fuels, better floating ability of suspended PF particles, etc. However, a temperature more than 65 to 70 is not recommended for various reasons.
Operating data from a coal mill is used to compare the fault detection observer-based method and PCA/PLS models based approach. There are 13 process measurements available representing different temperature, mass flows, pressures, speed etc in the coal mill.
The measurement is not updated, if the variation is less than 1%. The variations of T(t) is in the major part of the operational time inside this interval. Therefore, it is not suitable to be chosen as the predictor variable. However, the variations can be extracted from the TPA(t), which is used to control the temperature of the mill. Therefore, the PLS model is developed with the temperature of the mill as the dependent variable. In addition 6 of the other variables are chosen as regressors since there is barely information in the remainder.
A static PCA model is first developed, which captures around 99% of variations with 5 PCs (see Fig.5), which indicates strong collinearity among regressors. As shown in Fig.6, both Q and T2 statistics (with 95% confidence level) of the static PCA model are noisy, which potentially lead to false alarms. A static PLS model with 2 LVs achieves the minimal PRESS (see Fig.7), which is applied to the test dataset. Fig.8 shows the comparison between process measurement and the static PLS model prediction, together with the 95% confidence level. The process gradually drifts away form the NOC model, which eventually moves beyond the threshold around the sample 150. Due to the noise involved in the prediction signal, the estimation moves in and out the threshold from 110 till 200, when it is clearly out of the confidence level. Both Figs.6 and 8 reveal that static PCA and PLS models may lead to false alarms due to the noisy estimation. In addition, process measurements are commonly auto-correlated, this behavior is expected since the coal mill runs dynamical. Thus, dynamic models are developed by including time lagged process measurements, to address the issue of auto-correlations and reduce the possibility of false alarms due the none modeled dynamics.
Including time lagged terms enhance the NOC model by including historical data. However, time lagged terms also introduce additional noise into the modeling data block. For example, including n+1 time lagged terms might lead to poorer validation performance than the model with n terms due to measurement noise. Therefore, PRESS is used to choose an appropriate number of time lagged terms for a dynamic PLS model.
The predictive ability of the PLS model is improved with the inclusion of time lagged terms. The PRESS decreases from 1.645 to the minimal value of 1.142, which is obtained with a dynamic PLS of 3 LVs using 8 time lagged terms. The application of the dynamic PLS model to the test data reveals that the fault occurs in the process around sample 160. Fig.9 also shows a much smoother prediction such that the possibility of false alarms is significantly reduced. A dynamic PCA model is developed by the inclusion of 8 time lagged terms. The number of PCs is chosen as 2 through cross-validation, which explains 70.6% of process variations. The Q statistic of the dynamic PCA model is shown in Fig.10, the fault is detected around 160 samples, which is consistent with the dynamic PLS model.
The control loop for mill outlet temperature discussed here is mainly for TT boilers based on a CE design with bowl mill (refer to FigureVIII/5.1-2). A similar loop is valid for a ball-and-tube mill, which is discussed separately in the next section. In order to understand the loop in the figure, it is advisable to look at FigureVIII/5.0-1 and the associated PID figure (refer to FigureIII/9.2-4).
The outlet temperature of the coal mill is maintained at desired point so that the coal delivered from the mill is completely dry and achieves the desired temperature. Also, in case of high temperature at the mill outlet, cold air is blown in to reduce the risk of fire.
Normally, the entire requirement of PA flow necessary for a particular load at the mill is initially attempted through HAD so as to ensure complete drying of the coal (especially during rainy seasons) and to raise the mill temperature at a desired point. However, there may be times during hot dry summers when the mill outlet temperature shoots up. This is also never a desired situation because of fire hazard. In fact, to combat this fire hazard, arrangements for mill-inerting systems with inert gases (e.g.,N2 and CO2) need to be made (another purpose is to reduce air supply).
This is more important for ball-and-tube mills, especially when these are operated with one side only. Therefore, CAD comes into operation whenever there is need to bring down the mill temperature. Naturally when this damper operates (i.e., starts opening through process feedback), the hot air damper closes. Here also is a cross-operation of the two dampers but through process and not directly via the loop, so control loop disturbances are fewer than in the old days when cross-operations were implemented in the loop.
Mill outlet temperatures measured by redundant temperature elements and transmitters are put in an error generator. (Temperature element specialties were discussed earlier and so not repeated here.) The output of the error generator drives a PID controller. In general, since temperature is a sluggish parameter it is always advisable to use PID controllers for better results. To prevent controller saturation, controllers are put into service only when both the loops are in auto. The output of the controller through I/P converters normally drives pneumatic actuators meant for CAD.
As stated earlier, only when both HAD and CAD are in auto is the controller put into operation. Since FSSS operations depend on mill temperature conditions, with the help of the limit value monitor (LVM) necessary contacts statuses are shared with FSSS. The loop can be released to auto by an FSSS command. As a protection, both the full opening command and the >x% command for the mill CAD are issued from FSSS so sufficient cold air is circulated. If the auto release command from FSSS is missing or if HAD is in manual, it is necessary to inhibit auto operation so that the operator pay complete attention to the mill outlet temperature. That is the check back signal for FSSS from the loop for damper position.
How breakage energy and force are applied in the mill in order to achieve size reduction in an efficient and effective manner. This is a matter of design and performance of mills and the main subject of this section;
How the material being reduced in size behaves in terms of breakage characteristics such as strength and resulting broken size and shape. This relates to how the material responds to the application of breakage energy and force in terms of rate and orientation of application.
The analysis of individual mill design and operation is complex; so, for simplicity we will consider a typical mill layout for one mill type only. As VSMs have come to represent the bulk of the power station mill fleet, the explanation of mill operations will be based on this mill type. Figure13.2 illustrates the typical key components of a VSM.
In coal milling for power stations, a closed-loop process is used in which the rejects from the classifier are returned to the mill for regrinding. In VSMs, the re-circulation loop is within the mill, but some mill types would have an external loop. In fact, there are a number of re-circulation loops within a mill system. The situation is further complicated by the mill reject streams that reject undesirable material (tramp metal and non-coal bearing rock) from the mill. Generally, the following steps illustrate the path through a VSM:
Air entering through the Port Ring creates a fluidising zone in which heavy material (Mill Rejects) such as rock falls through the Port Ring into the Air Plenum below the Grinding Table and is ejected from the Mill through the Mill Reject System;
From the fluidising zone the ground coal is lifted up inside the Mill Body. Larger particles of coal reach a terminal velocity at which gravity will pull them back on to the Grinding Table for regrinding (Elutriation);
The fineness of the milling product and the capacity of the pulverizer are strictly connected. With increased fineness grows the overall circulation rate of coal in the mill, coal retention time and the flow resistance. As a result, the maximum mill capacity decreases and the rate of change of operational parameters of the furnace system deteriorates. In extreme cases, the performance of the boiler may be limited, and therefore improving the fineness of milling product must often include the modernization of the grinding system. The increase of the throughput of a pulverizer, which compensates the loss of capacity resulting from the increased fineness of coal dust, may be achieved through:
The analysis  proves that the maximum capacity of the ball-ring mill is obtained using 5 or 6 balls. Because earlier, as a rule, a greater number of balls was used, there is a possibility to increase the capacity by replacing the existing balls through a lower number of bigger balls. For example, in the EM-70 of FPM SA 9 balls of the diameter 530mm were replaced with 7 of 650mm. Such a modernized milling system can usually be set up on the existing gearbox. It should be noted that the costs associated with replacement of the classifier and the grinding elements are only slightly greater than the costs of the major repair of the mill. In the case where an existing mill has a grinding unit of the number of balls close to 6, the only way to increase performance is to increase the diameter of the balls, but this requires replacement of the mill body.
It has to be mentioned that the number of balls is increased during the mill operation. For example, the initial ten balls, after lowering the diameter below some value (due to wear), is complemented with the 11th additional ball.
If the existing pulverizer is equipped with 6 or 7 balls, increasing of its capacity is also possible by means of replacing the ball-ring system with the bowl and roller milling device. The milling costs per Mg of fuel in both systems are similar. However, with the same dimensions of the milling systems, the capacity of the roller system is about 15%20% higher. Another advantage is the shorter renovation time, which is about 714days for the ring-ball system, while for the roller mill, only 37days. In addition, hardfacing and re-profiling of grinding components are much easier for roller milling systems.
During the modernization of milling plant with compression mills, detailed analysis requires the selection of cross sections of nozzle-rings at the inlet of the drying agent to the mill, in order to minimize the amount of coal removed from the grinding chamber. The preferred solution is a rotating nozzle ring integrated with the bowl. This ring equalizes air distribution pattern at the periphery of the grinding chamber, which allows increasing the capacity of the grinding system without fear of excessive loss of fuel from the mill.
The rotational speed of the vertical spindle mill affects the operating conditions of the grinding unit. At high rotational speeds, the grinding unit operates at high flow of the material in the radial direction and low layers of the material under the grinding elements (balls, rollers). This causes the particles to be discharged without comminution and increases circulation in the mill. At the same time, the flow resistance and milling energy consumption (including erosive and abrasive wear) of the mill will increase.
If the rotational speed is too low, the material flow will decrease significantly. The thickness of the material layer under grinding elements will exceed the maximal height for which the particles are drawn under the grinding elements, causing excessive buildup of the material in front of the grinding elements. The material outflow from the bowl (or the bottom ring) is not supported by grinding elements movement, which results in higher flow resistance and uneven loading of the nozzle ring. These factors cause a significant decrease in mill efficiency.
Tests carried out for some industrial mills have proven that the change of grinding unit rotational speed strongly influences mill capacity. Therefore, by changing the gear ratio of the mill, both milling capacity and dynamic properties of the mill can be improved.
The fuel injector is designed to introduce the dispersed coal particles in a medium of air into the furnace. The mass ratio of air to coal is dependent on the coal mill manufacturer and usually ranges from 1.75 to 2.2 with a typical value of 2.0. An air to fuel mass ratio of 1.8 produces a primary stoichiometric ratio of approximately 0.16, or 16% of the air necessary for complete combustion of the coal. According to the previous discussion of NOx formation chemistry it is expected that lower NOx concentrations are achievable with lower primary gas/fuel ratios. The diameter of the coal transport line is constrained by the minimum velocity at which coal particles remain entrained in the carrier gas, or the coal layout velocity. This velocity is generally accepted to be 50 ft/s (Wall, 1987). The dimension of the fuel injector itself is selected by the burner manufacturer to provide the desired gas and particle velocity at the exit of the burner. The velocity here is anywhere from 50 to 115 ft/s and is chosen to provide the desired near flame aerodynamics impacting the mixing between the primary and secondary air. In many applications, there is an elbow, scroll or turning head in the coal pipe at the burner inlet. Such inlet devices result in roping, or an uneven distribution of coal within the fuel injector. Many manufacturers use components to redistribute the coal particles with an even density around the circumference of the fuel injector at its exit. A uniform distribution is typically desired to minimize NOx while maximizing combustion efficiency. The material of the fuel injector is chosen to be reliable under high temperatures and erosive conditions and is often a high grade of stainless steel. Another component of the fuel injector that is found on many commercial low NOx burners is a flame stabilizer. The function of this feature is to provide a stagnation zone at the fuel injector exit on the boundary between the primary and secondary air where small-scale mixing of coal and air occurs, providing ideal conditions for ignition and flame attachment.
Sulfur in coal can affect power plant performance in several ways. Sulfur in the form of pyrite (FeS2) can lead to spontaneous combustion and contributes to the abrasion in coal mills; therefore, if a lower quality coal containing pyrite is used in place of the design coal it can lead to problems. As the overall sulfur concentration increases, so do the emissions of sulfur dioxide (SO2) and sulfur trioxide (SO3). While the majority of the sulfur is converted to SO2 (about 12% of the sulfur converts to SO3), the increase in SO3 emissions increases the flue gas dew point temperature, which in turn can lead to corrosion issues. Most countries have legislation restricting SO2 emissions and utilizing higher sulfur coals will require additional SO2 controls (Miller, 2010). In some cases, the use of low quality fuels may impair the desulfurization equipment because of a greater quantity of flue gas to be treated (Carpenter, 1998).
All power stations require at least one CW pump and one 50% electric boiler feed pump available and running to start up a unit. In addition, fossil plant requires either coal mills or oil pumps and draught plant, e.g., FD and ID fans, PA fans, etc. Gas-cooled nuclear plant requires gas circulators running on main motors or pony motors at approximately 15% speed, whereas water reactors require reactor coolant pumps. Both nuclear types require various supporting auxiliaries to be available during the run-up stages, the poor quality steam being dumped until the correct quality is achieved.
When steam of correct quality is being produced, the turbine-generator will be run up to speed with all the unit supporting auxiliaries being powered from the station transformers via the unit/station interconnectors.
The Amer 9 plant utilizes both direct and indirect co-firing configurations. The plant co-fires biomass pellets up to a maximum of 1200ktyr1, generating 27% by heat through two modified coal mills. Only wood-based fuel has been used since 2006, due to reduced subsidies for agricultural by-products.
For the indirect co-firing option, low-quality demolition wood is gasified in a CFB gasifier at atmospheric pressure and a temperature of approximately 850C. The raw fuel gas is cleaned extensively and combusted in a coal boiler via specially designed low-CV gas burners. An advantage of this concept is that there is no contamination of the fuel gas as it enters the coal-fired boiler. This allows a wide range of fuels to be co-fired within existing emission constraints while avoiding problems with ash quality. The challenge, as always, is working within the relatively stringent fuel constraints while avoiding the inevitable high investment costs . Amer 8 also co-fires at high biomass feed levels but uses a standard hammer mill configuration.
Coastal power stations, due to their proximity to major urban areas, tend to be better managed in terms of production consistency and environmental standards. In China and India in particular, coastal power stations tend to mill coal more finely, use superior emissions to control technologies, and have a tendency to use higher-quality coal blends. The result is higher quality and greater consistency in fly ash chemical and physical properties, to the extent that the material is more desirable to local cement manufacturers and those in other domestic markets along the coast. This material is typically allocated in multiyear contracts.
Adding to this, coastal urban areas usually have high volume demand for construction materials. Coastal power stations are often fully contracted to supply cement-grade fly ash, as well as the run of station ash and bottom ash to serve this demand. This is particularly clear in China, where coastal cities such as Shanghai and Shenzhen have seen dramatic urban development over the last 20years. During this period, both cities have been net importers of fly ash, drawing from both inland and domestic coastal sources.
The cost of loading material onto vessels, whether in containers or bulk, is much lower at coastal power stations due to lower local land transport costs. As a result, coastal power stations have been logical first choices for exporters/importers, and many have already developed either domestic coastal markets for their ash or export markets.
At a time when combustion optimization seems to be such an important part of reducing emissions and improving efficiency, in part because of carbon constraints, we have to ask ourselves, how do we balance coal flow to the burners? For dozens of years utilities have spent countless dollars on a variety of techniques and services aimed at balancing their coal pipes. So why is combustion still so bad?
The reality is that manual coal sampling in real world power plant conditions is challenging. Call it dirty airflow or rotorprobing, ASME sampling or isokinetic sampling; it does not work for the purpose of accurately measuring coal flow. This should come as no surprisethe procedure is inherently flawed.
In a typical coal mill, primary air (PA) is used to both dry the coal and transport it to the burner(s). Figure 1 depicts a two-pipe coal milling system showing the PA that is used to transport pulverized coal to the burners through both outlet pipes.
Most original equipment manufacturers (OEMs) supply the user with a PA:feeder curve. One challenge in creating these curves is determining and maintaining optimum transport velocity throughout the entire range of mill loading/operation. The velocity is a function of coal piping geometry, coal type, moisture content, coal fineness and more. The value is important because at lower velocities, coal flow cannot be sustained and coal layout, pipe pluggage and fires can occur.
Through years of trial-and-error testing, these curves have often been changed in an effort to improve combustion and to optimize low NOX burner performance. Several factors are crucial to creating these curves.
Primary air is essential for proper drying of the coal. The mills outlet temperature is used to determine this and the hot and tempering air dampers are controlled to allow for desired mill outlet temperature, but it is the volume of the primary air that must deliver the coal properly to the burners. Too little and coal layout, mill spillage or pipe pluggage can occur. Too much, and coal pipe erosion, long flames/waterwall impingement, high loss on ignition (LOI), high NOX and more may result.
Another area of concern is with the air distribution to the pipes and whether or not the air (PA) is distributed equally to each of the pipes. Field testing the airflow in the coal pipes with no coal in thema clean air testis performed with Pitot tubes. The desired outcome is equal air velocities in the coal pipes. If clean air velocities are not equal, an orifice or restriction is sometimes located in the pipe(s) with the higher velocity to balance the velocities.
Testing is usually performed closer to the burners to accommodate for piping losses; longer pipes may need more air to achieve equal velocities near or at the burner entrance. But if the major job of the PA is to get the coal to the burners, why should any orificing or restricting be done prior to knowing how the coal is distributed among the pipes?
When coal mixes with PA in the milling system in addition to PA, it becomes a two-phase flow instead of single phase, with air as the gas phase and coal as the dense phase. Based on typical PA:feeder curves (approximately 2:1 mass basis), there is significantly more air in a coal pipe by volume than there is coal. If the coal enters the pipes such that the mass of coal to each of the pipes is equal and air velocities are equal as determined by clean air testing as described above, then all other things being equal (such as coal quality and distribution of the coal in the pipes) the pipes should again be in balance and equal.
However, since friction between the air and coal actually transports the coal to the burners, then the coal must travel more slowly than the air, so they are not equal. Because of this, measuring dirty air flow or the air in a coal pipe laden with coal is of no real use in sampling coal. If coal is to be extracted isokinetically from the coal pipe, it must be extracted at the rate at which the coal is traveling (not the rate that air is traveling) and then weighed.
Moving coal and transport air between the two pipes can be complex. Without the ability to accurately measure the coal mass and velocity, this blind movement with orifices or other adjustable valves often causes more imbalances and other combustion problems.
Although coal milling systems have not made many advancements in the past 50 years other than capacity and the use of wear-resistant alloys, coal measurement technologies have. We can now see the adverse conditions created through the use of old test methods.
A four-pipe mill delivering a greater amount of coal to one pipe than the others, with improper orificing caused by using manual sampling and other improper control strategies, could result in a lack of transport air in the heavily loaded pipe making it much more unstable.
The massflow measurements are made using a microwave mass measurement to measure coal density and an electrostatic cross-correlation to measure the velocity of the moving coal particles. (See Figure 3.) The massflow is the product of the two. The mill delivers a much higher loading (density) to pipe 2-1 LR. Lacking more transport air in this pipe, the coal particles move much more slowly. But previous dirty air tests show the air velocity is at or near that of the other pipes, so extractive sampling did not derive accurate results.
The velocity data is that of the moving coal particles. (See Figure 3.) Air velocity in these pipes ranges from 110 to 120 fps. The difference between the air and coal velocities is the coal slip. If isokinetic sampling were to be accurately performed, the coal would have to be extracted at the rate at which the coal is moving in the pipesthe coal velocity. Because the coal velocity cannot be measured with existing sampling methods, the industry practice is to measure the air velocity and then extract coal at that rate. As can be seen, the rates are not the same and any data generated with such methods will be incorrect.
Some field testing methods take into consideration that slip exists between the air and coal and adjust their sampling devices with this in mind. But the slip varies with coal loading, so correct slip factors cannot be determined without knowing the coal loading. Therefore, random slip values are used. The end result is that there is no way for these manual sampling devices to generate accurate coal flow measurements.
While air flow/velocity may be somewhat stable, the movement of the coal particles in coal pipes can be erratic. The coal particles have size and weight. The pressure in a coal pipe is low and the result is erratic movement of the coal. The higher the air concentration, the more stable the movement of the particles. But excessive air in the coal pipes will lead to pipe errosion, impingement on waterwalls, higher NOX and poor combustion related to throwing the coal through the flame (poor, incomplete combustion). Extracting coal at individual points in a coal pipe via a manual sampling procedure takes time. By the time the sample is obtained at one point, the coal flow at the next point has changed. Similarly, when the sampling of the last pipe of a mill is being completed, the massflow in the first one(s) is likely to have changed.
Coal flow is dynamic and constantly changing. Todays coal flow measurement technologies are real-time, generating coal flow data in each pipe as fast as every second. Only by using real time measurements can what the system is doing be seen and coal pipe balancing performed.
Adjusting coal between the coal pipes of a mill using orifices (fixed or adjustable) is complex because of the two-phase flow environment. A restriction in the pipes cross-section will affect both the air and the coal. Continuous coal velocity measurement is critical when using restrictive devices for balancing pipe coal flows.
Adjusting coal valves (at the mill outlet) can improve coal distribution while still maintaining proper coal velocity. The ability to continuously measure both the massflow and the velocity of the moving coal is critical to balancing coal pipes with valves and orifices.
Many operators choose to increase PA because of spillage or high mill differential due to wet coal, mill wear or other reasons. This higher PA results in high NOX, high LOI, high carbon moNOXide (CO), excessive slagging and more. Immediate identification of excessive coal velocity, through real time measurement, can lead to the reduction of PA. This in turn can improve coal distribution.
Why continue to perform or contract for manual sampling just because its been done that way for 50-plus years? The results will be the same as they were 50 years ago. Obtaining erroneous data results in changes that may hurt, not help, the combustion process.
Todays real-time measurement methods lead to correct data collection and then improved combustion. These instruments can be used in your distributed control systems (DCS) for on-line control of PA, coal feed, fuel:air ratio control using burner secondary air and more. All of this will result in reduced emissions, improved efficiency, fewer tube leaks and more.
While we stressed the main difficulty of manual samplingextraction of coal at erroneous ratesthere are several other challenges to manual coal sampling, including variations in coal distribution in each pipe and the inability of an extraction probe to detect this; effects of angular flow on Pitot tubes; difficulty with using manual probes such as holding the probe in proper position in each pipe; difficulty in reading dirty air flows (dP) with an anemometer and more. Some manual testing personnel do a more thorough job of obtaining data with sampling instruments than others. But true accuracy cannot be obtained with the flawed methods.
Authors: Dave Earley is with Combustion Technologies Corp. in Apex, N.C. He is a mechanical engineer with more than 20 years of industry experience, focused on combustion measurements and optimization for the past 15 years. Bill Kirkenir is the lead combustion engineer with Progress Energy Corp. He is a mechanical engineer with 30 years of experience in power plant operations and combustion optimization.