ABSTRACT

The optimum strategy for boiler control matches actual energy delivered against energy demand. Marry variables affect the traditional measurements of delivered energy such as variations in filet heating value, moisture content of the coal, pulverizer wear, pulverizer response characteristics under normal and abnormal operating conditions, and unmetered fuel such as ignitors or warm-up guns. For these reasons the traditional material metering methods of measurement do not reflect the actual energy or heat delivered to the furnace. This paper will discuss methods for obtaining a total fuel signal that reflects actual energy delivered.

PART 1: BACKGROUND

An ideal boiler control system matches the output energy of the boiler to the energy demand of its load. An ideal strategy is able to maintain this energy balance not only under steady state conditions but also under changing load requirements. To maintain the energy balance between boiler under changing load. accounting for the energy storage characteristics of the boiler is a necessity. Thus adequate provision must be made for over and under firing to compensate for the boiler energy storage characteristics.

Modem boiler control systems utilize the parallel regulation of boiler inputs to make actual energy delivered equal to the energy demand requirements. Effective parallel regulation of boiler inputs requires accurate measurement of the inputs to be controlled. Inaccurate metering of boiler inputs will result in improper control actions by the boiler input controllers. These improper actions then cause unnecessary upsets to the controlled variables of megawatts throttle pressure, steam temperatures. drum level and excess air. Instability of these key variables in mm will severely limit the units ability to respond to changing load requirements. Excessive deviations can result in the unnecessary and costly expenditure of equipment life.

Inaccuracies of metered boiler inputs not only cause problems for the controls. they also make the monitoring of critical relationships of boiler inputs impossible. For example boiler protection systems based on monitoring of Fuel/Air or Firing Raw/Feedwater ratios can only be as effective as the metered inputs. Similarly comparison of multiple measurements of boiler inputs such as multiple fuel input by different measurement techniques that is used to determine metering validity cannot function if the basic measurements do not reflect the actual process input.

You may ask - Why review the necessity of boiler input measurement accuracy in general, and the necessity of an accurate fuel input measurement in particular? The reason is there can neither be effective boiler control nor effective boiler monitoring and/or protection without the accurate measurement of the input variables to be controlled. This brings us to the problem of an accurate fuel measurement.

PART 2: THE FUEL MEASUREMENT PROBLEM

One of the most difficult problems that face the designer of control systems for fossil fuel fired boilers is the development of a fuel measurement that is adequate for meaningful fuel/air ratio control. And in the case of once-through units. an adequate fuel measurement is critical to control of firing rate/feedwater ratio as well. So, what are the problems of measuring fuel input?

1. Varying Heat Value Content

It used to be that utilities could depend on single sources of fuel supply fbr long periods of time. All that changed with the energy crisis of the 70's. High costs for energy as well as stringent emmission standards have forced utilities to source fuel supply from various suppliers depending on price and availability. Thus a constant heating value for any given fuel can no longer be assumed. This problem is further complicated by the industry shift back to coal firing which has always been subject to significant heating value variations. The heating value problem exists for all fuels, gas, oil or coal.

2. Measuring Fuel Delivery

Measuring fuel delivery has been convenient with gas, more of a problem with oil, and nearly impossible with coal . Why?

Gas and Oil Measurements

- Conventional orifice measurement techniques have limited turndown capability - typically 5:1. Wider range measurements require dual-range metering with the attendant complications of making a smooth transfer between low range and high range meters.

- In-line devices can offer wider turndowns - typically 10:1. but with the attendant problem of having to shut down the line should maintenance of the in-line device be required.

Coal Measurements

Coal delivery systems don't reflect the actual coal flow to the furnace due to two main considerations: mill storage capacity and mill response capability. Let's examine the effect of mill capacity and mill response as they relate to the method of control of mill output coal flow to the furnace.

- Control of Mill Output Coal Flow by Feeder Speed

When mill coal flow is regulated by adjusting feeder -speed. totalized tachometer output is used as a measure of fuel input. A typical example would be a CE-Raymond Bowl Mill with gravimetric feeder. The total feeder speed measurement must be compensated for the following factory in addition to variations in heating value.

- Mill Capacity - All mills have significant capacity with respect to both the grinding and also the storage of coal. Thus, during load changes the fuel flow into the mill does not necessarily represent the actual fuel going to the furnace. A typical time constant for a Mill is 100-160 seconds*. That means that it takes 100-160 seconds before a change of mill input will appear as a significant useful output. Equating air flow to measured fuel input under these conditions guarantees an erroneous fuel/air ratio. The classical solution to this problem is to emulate the mill dynamics with a "Mill Model" using feeder speed as its input. This approach is limited in that the actual time constant for a mill is quite complex and varies not only as a function of load but also with primary air flow and the way primary air flow is regulated.

- Loss of Coal - Coal hangups, a common occurrence, are not reflected as a change in the feeder speed signal even though there is no coal on the feeder belt. "No coal flow" contacts can help to correct this problem by removing the feeder speed measurement from the total. However, loss of coal tends to be a sporadic event which can result in frequent switching of the individual feeder signal out of the total which if not properly accounted for by the control system can cause significant upsets to the unit.

- Mill startup/shutdown procedures can affect the accuracy of the feeder speed based fuel measurement. A feeder may be shut down to relieve a mill overload or to sweep a mill clean - a condition where useful mill output exists even though the feeder speed is zero. Conversely, if the feeder is run to fill a mill, the result is a feeder speed signal that represents no fuel output.

- Control of Mill Output Coal Flow by Primary Air Flow Control

When mill output is controlled by regulating mill primary air flow and the coal feed to the mill is regulated as a function of mill level, the total of the individual primary air flows is used as a measure of total fuel input. A typical example is the B&W type EL pulverizer.

Limitations

Use of primary air flow as a measure of fuel input has severe limitations in that the amount of coal in a given primary air can vary significantly. This is caused by a number of factors.

- The amount of coal flow to the furnace depends on mill level. Some mill level controls are limited to three values of coal feed rate (high, low and off in the case of most feeder tables). This approach causes continuous oscillations in level, hence fuel to the furnace.

- There are inaccuracies associated with making the basic primary air flow measurement such as temperature variations. Even obtaining a reasonable mill air flow differential is difficult due to the short piping runs around the mill.

- Physical differences between mills (differences in grinding surfaces, classifier settings, etc.) create errors in the relationship between primary air flow and coal flow to the furnace.

- Coal flow versus primary air is only roughly linear over the mills operating range. During mill start-up and shut-down, a significant primary air flow is present. This flaw can have little or no fuel content as the mill is either filled up or swept empty.

- Ball Tube Mills

A special case is the ball tube mill. From the control point of view, a most important characteristic of this mill is its very large capacity. Up to 15-20 tons of coal may be stored in this mill as compared to the 3-5 tons storage in a more conventional pulverizer. Thus it is not unusual for this Mill to maintain full output for 10-20 minutes after its feeder has been shut down. Thus feeder speed is not a viable measurement of ball mill output.

The classical approach to measuring fuel output of a ball mill utilizes mill primary air flow as an index of coal that is swept up from the mill and delivered to the furnace. In addition to the limitations of primary air flow as a fuel measurement cited above, ball mills have a number of additional factors that affect the coal flow to mill air flow relationship.

* The status of ball mill charge is a major variant that effects the grinding efficiency of the mill, hence the reservoir of fines that will be transported by a given air flow.

* The measurement itself may be suspect since it may use as its primary element, the pressure drop across the classifier. There has been considerable debate as to whether coal flow is best represented by the raw differential pressure or by the air flow measurement.

* The non-linearities in the relationship of individual min coal flow to mill air flow result in differences in computed fuel input depending on the number of mills in service and the coal output supported by each mill.

3. Unmetered Fuel

Additional fuel inputs may be unaccounted for since these inputs are not metered. Examples of unmetered fuel inputs include ignitors and/or warm up guns.

The obvious conclusion from all of this is the desirability of developing an alternate method of measuring fuel input.

PART 3: DEVELOPMENT OF THE HEAT RELEASE CALCULATION (DRUM TYPE UNITS)

The difficulty of obtaining an accurate fuel measurement precipitated work in the early 60's to develop an approach to measuring fuel input that would reflect the actual BTU released in the furnace.

The results of this work led to the development of a method of calculated Heat Release which has been applied to over 180 drum type boilers, coal fired. oil fired, gas fired or multiple fuel fired units since its inception in 1962

Basically, Heat Release computes fuel input based on the following fundamental relationship:

Fuel Input QFI = Boiler Output QBO + Change in Stored Energy QSE (1)

Where Q = Heat Flow in BTU/minute or kilocalories/minute.

Steam flow* or first stage pressure is an excellent index of boiler output energy and can be measured simply and accurately.

Obtaining a measurement reflective of boiler stored energy changes is another matter. However, theoretical considerations supported by actual field experience show that the major part of boiler stored energy is found in the boiler metal temperature and the resultant fluid enthalpy.

Drum pressure, since it is essentially linear with fluid enthalpy, becomes a good index for energy storage.

For control purposes the actual level of stored energy is not important since fuel input is not needed to support a given level of boiler stored energy. Rather. it is the changes in stored energy that are significant to the controls since fuel input is required to establish a new level of energy storage.

Thus is it only the changes in drum pressure that are important. Since the basic units of the fuel energy balance relationship are the time rate of energy flow. the rate of change of drum pressure provides a representation of the rate at which stored energy is changing.

We can therefore rewrite equation (1) above as:

Fuel Input KIP, + K2 X dPD/dT

Where Pi = Turbine First State Pressure

PD = Drum Pressure

T =Time

Ki, K2 are constants

First stage pressure and drum pressure are easily measured variables. These variables can be combined to provide a total fuel input signal called Heat Release. This signal is independent of feeder speeds, differential pressures and the like. This signal also provides an accurate totalization of fuel input regardless of what combination of mills happen to be operating at the time, or the performance characteristics of each mill. In addition Heat Released inherently totalizes fuel from all sources. This means that it will recognize and correct for the insertion of oil torches as well as changes in fuel -quality.

FIELD APPLICATION

As indicated earlier. this technique has been in use for many years on fossil fired drum-type units of all varieties and particularly coal fired units. Since Heat Release is a measure of total heat input to the boiler, it is used as feedback in the fuel control system.

While all modem control systems employ 02 connection, it should be recognized that 02 control is not adequate to provide for changes in fuel heating value. In fact heating value upsets in the fuel system that remain uncorrected. result in undesirable excessive activity of the O,) control.

control is basically a calibrating type of control that depends on the response of an analytical measurement. After making a determination, the required correction is implemented via the integral action of the controller. All of this takes a considerable amount of time, since the controller's integral action is slow in order to maintain stability. Heat Release. on the other hand, takes its actions on a direct and calculated basis without use of integral action which is fundamentally destabilizing. 02 does have a value in making a final correction to resolve small errors in the Heat Release computation itself. With the use of Heat Release, the range of the 01 correction can be kept to approximately a 5 percent range ev7en though fuel quality itself may vary 10 percent.

While the Heat Release calculation has been most effective as a fuel feedback for drum type boilers. the same fuel measurement problem remains for the once-through units.

PART 4: BTU COMPENSATION OF TOTAL FUEL SIGNAL (ONCE THROUGH UNIT)

A partial solution to the problem of the total fuel measurement would be to at least compensate the total fuel measurement for variation in heating value.

We recognize from our original work on the Heat Release Calculation, that steam flow or first stage pressure is an excellent index of boiler output energy under steady state conditions. This assumes that ail boiler output steam flow is directed to the turbine. and that there are not significant changes to the unit which effect the cycle efficiency which as removing feedwater heaters from service.

Thus under steady state conditions for the once-through application we can write that

Fuel Input QFI = Boiler output QBO (3)

Where Q = Heat Flow in BTU/minute or kilocalories/minute.

Unfortunately, for the once-through unit. we have yet to Find a good index for scored energy. But let us consider what we can achieve based on use of the boiler output energy which is a readily measurable quantity.

With the once-through unit we must begin the solution of the fuel measurement problem by improving the classical approaches that are normally used.

For coal fired units the key lies in refining the "mill model." The first consideration is - What signals are to be included in the fuel totalization? A convenient rule of thumb would be that any fuel representing more than 5 % of total fuel input should be measured and totalized.

The second consideration is - Does the measured signal actually reflect fuel flow to the furnace? Signals that don't reflect a fuel flow contribution should be cut out. For example, feeder running with no coal on the feeder belt should be cut out. Accomplishing the switching of these signals ahead of the "mill model" should minimize fuel system transients as well as compensate for the 14ct that even though coal to the mill is cut off. the mill output. due to storage capacity of the mill. will actually decay over time. Similar strategies can be applied to the processing of mill air flow signals.

The final improvement is to provide automatic compensation of the total fuel signal for actual heat energy released. This is accomplished by computing the steady state ratio of boiler output energy to total fuel (P1/Total Fuel Input). By using this as a scaler on the total fuel measurement. the total fuel measurement for control will be equal to boiler output energy under steady SI.LIC conditions.

*Steam flow x pressure (no temperature compensation)

How can this be done without destabilizing the traditional controls for fuel feed, firing/rate feedwater ratio or Fuel/Air ratio? The key is that this correction must be made gradually - less than 1/10 of the reset rate of the fuel controller.

Thus the fuel flow feedback will be equal to the mill model output as calibrated for actual heat release under transient conditions. Under steady state conditions. fuel feedback will be equal to boiler output energy.

PART 5: BENEFM OF THE HEAT RELEASE BASED FUEL MEASUREMENT

The benefits of the Heat Release based fuel measurement must be considered separately for the drum type unit application as compared to the once-through unit.

FOR THE DRUM TYPE UNIT:

The Heat Release calculation will:

- Provide a simple direct method to measure ail fuel inputs to the boiler regardless of source or quantity.

- Recognize and correct for any changes in fuel heating value.

- Measure actual fuel burned independent of feeder - pulverizer response.

- Maintain uniform fuel-air ratio during load changes and at steady state.

- Stabilizes steam temperatures because of the uniform fuel-air ratio.

- Permit the monitoring of critical relationships between boiler inputs and the implementation of appropriate corrective action to maintain proper relationships between boiler inputs such as fuel/air ratio.

- Permit comparison of fuel measurements on basis of actual BTU released in the furnace.

FOR THE ONCE-THRU UNIT

The steady state Heat Release calculation will:

- Recognize and correct for long term changes in fuel heating value.

- Maintain critical ratios of Firing Rate/Feedwater and Fuel/Air at steady state, eliminating the offsets that can occur from varying heating values.

- Permit the closer monitoring of critical relationships between boiler inputs as well as institute appropriate corrective actions.

- Permit comparison of fuel measurements on basis of steady state heat release.

CONCLUSION

Use of the Heat Release based fuel measurement can significantly enhance the operation of boiler control systems for large utility boilers by maintaining the fuel energy balance relationship of the fuel controllers.

This approach assures that the full range of the oxygen control and temperature control (once-through only) will remain available for the control of these key boiler variables. Monitoring of critical ratios can be done to closer tolerance assuring better balance between boiler inputs.

While the Heat Release concept is well proven with many years of experience in large utility boilers, we anticipate that the extension of this concept to the once-through unit will make its mark on impr(yved operation of the large once-through units.

REFERENCES

I. T.W. Jenkins Jr, I Garber, FL. Stewart Heat Release Computer for Combustion Control Systems Proceedings of the American Power Conference. Vol. XXVI, pp. 333-343, 1964.

2. James H. Daniels, Boiler Turbine Control System U.S. Patent 3,247,671, April 1966.

3. T.W. Jenkins Jr, B. Littinan Response Capability in the Control of Large Generating Units IEEE Paper No. 71CP73 PWR presented at the IEEE Winter Power Meeting, New York, NY, January 31-February 5, 1971.

4. R P. Ryan Modernization - A Key to Improved Performance Instrumentation in the Power Industry, Volume 25. ISA Power Symposium, Phoenix Arizona, May 24-26, 1982.

5. Richard H. Morse, Boiler Control System U.S. Patent 4,213,304, July 22, 1980.

6. Richard H. Morse Balancing Boiler Inputs to Energy Demand - One of the Basic Considerations in the Design of Controls for Drum Type Boilers ISA Paper No. Cl.82-745, ISA Conference Exhibit Philadelphia PA, October 17-21. 1982.

7. Cyrus W. Taft Performance Evaluation of Boiler Control System Strategies Seminar: Control Systems for Fossil Fuel Power Plants EPRI Research Project MO Session 4, Atlanta Georgia, February 24-26, 1987.

8. H.C. Gery The Evolution of Coordinated Control Instrumentation in the Power Industry, Volume 31, ISA Power Symposium, St. Petersburg Florida, May 23-25, 1988.