Вы находитесь на странице: 1из 5

Babcock & Wilcox 1

Control System Retrofits to Improve Plant Efficiency


T. Leung
Babcock & Wilcox
Cambridge, Ontario, Canada
J. Peet
Babcock & Wilcox
Cambridge, Ontario, Canada
Presented to:
Canadian Electrical Association
Thermal Generating Station Construction
and Commissioning Session
March 1995
Vancouver, British Columbia, Canada
BR-1589
Summary
The availability of powerful first principles type
models designed for PC-based computer hardware pro-
vides a simple and effective tool for the study of tran-
sients within power plants. All of the essential compo-
nents of the plant including the controls can be modelled
and the behavior of the plant to various transients can
be evaluated. This paper will address the use of such a
dynamic model to design an improved control strategy
for pulverized coal fired drum type boilers connected to
single reheat type turbine generators. Comparisons with
the current techniques will be used to quantify the ben-
efits and demonstrate the simplicity of the control meth-
odology.
Introduction
Utilities in North America are faced with a declining
margin between installed capacity and electrical demand.
Consequently, a greater proportion of units are required
to participate in middle load and peaking duty. This ap-
plies to the larger, more efficient plant and older, smaller
units which previously were base loaded. A utilitys abil-
ity to meet the system demand in the most cost effective
manner will be, to some degree, dependent on the load
range and maneuverability of the individual units. The unit
efficiency under cycling duty will depend on the degree
to which boiler and turbine efficiencies can be maintained
over the operating load range. The parameters which can
affect the unit efficiency during cycling operation or limit
the units capability to cycle are:
1. Pressure and temperature control to the turbine
2. Extent of overfiring/underfiring in the boiler
3. Excess air and exhaust gas temperature leaving the
boiler
The paper will discuss new control techniques which
address the above parameters and thereby provide im-
provements in load cycling capability and efficiency of
coal fired thermal power plants.
The control system described is conventional in nature
and does not incorporate or involve the use of advanced
control concepts, artificial intelligence techniques, or
expert type systems.
Open Loop Response
To appreciate the approach taken in the development
of the control strategy, it is worthwhile to consider the
open loop characteristic of pulverized coal fired drum
boilers. A realistic open loop in which the boiler/turbine
unit operates safely and effectively requires all the lower
level controls within the hierarchy to be in automatic mode.
This includes drum level control, superheater attemperator
control, air/fuel ratio control, etc. Thus, the open loop
considered consists of the turbine control valve in manual
and the boiler firing rate control (boiler master) in manual.
All lower level control loops are in automatic.
The results of individual manipulation of the boiler fir-
ing rate and the turbine control valve are shown in Fig-
ures 1 and 2 respectively.
From the results shown in Figures 1 and 2, the follow-
ing conclusions can be drawn:
1. Main steam flow is primarily a function of the fir-
ing rate, not an independent variable.
2. Main steam temperature is a function of the main
steam flow for a given firing rate.
3. The response shown for a firing rate change (with a
fixed turbine valve) is representative of true vari-
able pressure operation (the only externally manipu-
lated variable available is firing rate).
The impact of the open loop test for boiler/turbine con-
trol is that the firing rate is the fundamental control ele-
ment from which main steam flow, and subsequently main
steam temperature, are derived. Improved control of the
firing rate, and the use of the firing rate as a feedforward
for turbine valve position and the attemperator control
valve, will result in improved coordination control of the
boiler/turbine unit for pressure and temperature.
2 Babcock & Wilcox
Figure 1 Fuel flow step change (100%- 95%-100%) open loop response.
Figure 2 Turbine valve step change (100%- 95%-100%) open loop response.
Firing Rate Control
The improvement of the firing rate control comprises
two separate aspects, namely:
1. The manipulation of the unit load demand signal into
the firing rate demand using feedforward and error
feedback techniques.
2. Modification to the basic pulverizer control loop to
improve the response characteristics of the delivery
of pulverized fuel to the burner when the firing rate
demand is changed.
The second aspect, item 2, will be addressed first since
it is a lower level control loop within the hierarchy; item
1 will be addressed when the configuration of the coordi-
nated control for firing rate, steam flow and steam tem-
perature is discussed.
Pulverizer Control
It has long been recognized that the large time con-
stants associated with modern vertical spindle, medium
to low speed type pulverizers create a significant time
lagged response of fuel to the burners from a change in
the demand signal. Solutions to this problem have been
proposed and utilized, such as kickers in the primary
air control and the feeder speed. While the fuel flow re-
sponse to the burners can be improved with an overshoot
of the primary air flow, the benefit is short lived as the
storage of ground coal within the mill is exhausted. Over-
feeding the mill with coal by a kicker on the feeder speed
is used to maintain the coal inventory within the mill and
thus sustain the positive response characteristic achieved
by the manipulation of primary air. The approach, how-
ever, is not fully successful, partly because the time re-
quired to replenish the ground coal in the grinding zone is
not satisfied. A typical pulverizer response similar to
Figure 3 usually results.
Simulations using the dynamic model (PC Trax) showed
similar results (see Figure 4) to Figure 3 when feeder speed
and primary air were increased simultaneously.
Babcock & Wilcox 3
The improved initial response of the fuel flow is over-
shadowed by the upset to the system caused by the failure
to sustain the fuel flow as the ground coal inventory is
lost in the mill. The overall effect is to undermine the
control of the plant rather than improve controllability.
Of note, the time delay shown in Figure 3 represents the
time constant of the firing system and the boiler while in
Figure 4, the time constant is the pulverizer only.
Figure 3 Pulverizer response measured as heat release.
The solution to this problem becomes simple once the
underlying principles are understood. Improved pulver-
izer response as shown in Figure 5 results from the incor-
poration of a time delay on the primary air flow as shown
in the pulverizer control logic, Figure 6.
Figure 4 Step change in PA and feeder speed response.
By allowing the real inventory in the pulverizer to
steadily increase, an increase in coal flow to the burners
can be sustained, resulting in a smooth output increase
which provides the fastest response without jeopardizing
controllability. This same thinking is applied to the co-
ordintion control discussed later and results in a similar
optimized response.
The smooth pulverizer response permits not only stable
control but permits tighter control of excess air without
unburned carbon, CO excursions or local reducing condi-
tions during load changes because the response is consis-
tent with a first order time lag.
Figure 5 Improved pulverizer response.
Figure 6 Improved pulverizer control logic.
Coordinated Control
The development of the unit coordinated control is
based on two relatively simple concepts, namely:
1. That a relationship exists which will remain essen-
tially constant in time among the MW output of the
turbine generator, the main steam flow to the tur-
bine and the boiler energy input.
2. That for a given energy input to the boiler, the out-
put energy (i.e. flow times enthalpy) will be fixed.
Therefore, the additional attemperator spray flow is
the difference between the actual boiler outlet steam
flow and the outlet steam flow which would corre-
spond to the given input (from the relationship 1
above).
From the above relationships, an effective control sys-
tem has been designed as shown in Figures 7 and 8.
The first concept [1 above] provides the following
feedforward signals:
Firing rate (fuel input) from total Unit Load Demand
(ULD)
Main steam flow from boiler heat release (fuel input)
The use of a feedforward for fuel permits a faster re-
sponse and a more stable control system. The boiler heat
4 Babcock & Wilcox
release is determined from a simple representation of the
pulverizer response characteristic shown previously (us-
ing a combination of time delays). The main steam flow
which corresponds to the heat release determined above
becomes the feedforward signal to the turbine control
valve. This signal is adjusted for pressure error and for
MW error before comparison with the actual main steam
flow. The resulting flow error is fed to the turbine valve
flow controller. Notably, main steam temperature error is
not included in the flow signal to avoid hunting between
the turbine valve and the attemperator spray valve. To
decouple the firing from the flow, a MW error is included
in the steam flow signal and a pressure error signal is added
to the firing rate demand. These changes produce the fol-
lowing results:
With an error in MW, the flow signal will be compen-
sated such that no change in pressure will occur.
With an error in pressure, the firing rate signal will be
compensated such that no change in MW will occur.
The control description above is based on the concept
1 stated previously, and provides a balance between fir-
ing rate and steam flow to minimize pressure error. Con-
cept 2 is used to develop the main steam temperature
control (Figure 8) which is functionally part of the coor-
dination control. This concept is demonstrated in the
attemperator spray valve, which provides a means to adjust
the main steam flow with little or no effect on pressure.
Consider an operating condition in which the steam
flow from the evaporator is less than the turbine flow re-
quired for the MW load and upon which the boiler input
is predicated. If there is no spray flow, the steam tem-
perature will be too high and the flow too low, although
the product of flow times enthalpy will be correct. With
the control described above, the flow signal will cause
the turbine valve to open to achieve more flow. Since the
flow is set by the evaporator absorption, there is no more
flow available and thus the boiler pressure will decrease.
Lowering drum pressure will increase the enthalpy of satu-
rated steam and thus steam flow will decrease further.
Meanwhile, steam temperature will continue to rise. Ob-
viously, increasing steam flow by spray would satisfy the
flow requirement without significantly affecting boiler
pressure and thus is fundamental to the firing rate/flow
balance.
The steam temperature control takes the feedforward
signal for steam flow generated in the coordinated con-
trol (with the MW error and pressure error included), ad-
justs it for any outlet steam temperature error and com-
pares this to the actual steam flow. The difference is the
incremental demand for spray flow, i.e. the spray flow
error. Since the basic signal goes to both turbine and
attemperator control valves, it becomes evident that tem-
perature error cannot be included in both signals. Any
phase or valve response characteristic differences would
result in oscillations between the two valves. Thus, the
incorporation of the spray attemperator control completes
the balance between firing rate and flow for which steam
temperature and pressure are held constant.
Simulations of this control in the dynamic model show
that the ability to respond to load changes is superior to
conventional boiler follow control in all aspects. Thus,
the ability to handle fast ramping for load response or step
Figure 7 Unit coordinated control.
Figure 8 Main steam temperature control.
Babcock & Wilcox 5
loading for frequency response is superior both in response
time as well as the critical parameter deviations (pressure,
temperature, overfiring, etc.). The ramping capability is
demonstrated in Figure 9, which simulates a 600 MW
coal-fired plant increasing load from 55% to 90% MCR
at 5% per minute in hybrid sliding pressure mode (as
shown in Figure 10). The small degree of overfiring re-
quired to achieve the ramp rate was achieved by limiting
the rate of pressure increase. The reserve capacity of the
turbine control valve in the hybrid mode (two valves open)
made this possible. The model also shows a noticeable
Figure 10 Turbine throttle pressure versus load for hybrid sliding
control.
time constant for the MW response from the IP and LP
cylinders. Consequently, the HP cylinder overshoots its
final equilibrium value while the IP and LP catch up. Thus,
in a pure sliding pressure mode, throttle pressure would
not equal generator output in a fast ramp. In addition to
the control methodology described above, an improved
drum level control was also developed to extend operat-
ing capability for those instances where drum level con-
trol may be the limiting condition.
The availability of detailed first principles mathemati-
cal models for thermal power plants provides a powerful
tool for understanding the transient behavior of the equip-
ment and thus allowing the development of improved con-
trol schemes. Applying this technique, the authors have
developed an improved control strategy for a coal-fired
utility plant which would permit greater flexibility of op-
eration within the permissible deviations of pressure, tem-
perature, oxygen, drum level, etc.
The use of this control could permit or extend the par-
ticipation of coal-fired units in primary regulation and
secondary load frequency control.
1. Langley, D.C. and Rosenberger, F.G., Coal-fired
Cycling Operations at Illinois Powers Havana Station,
presented to the American Power Conference, Chicago,
1980.
2. Peet, W.J. and Leung, T., Improved Steam Tempera-
ture Control for Drum Type Boilers in D.S.S. or Load
Cycling Operation, presented to the 9th CEPSI Confer-
ence, Hong Kong, 1992.
3. Peet, W.J. and Leung, T., Dynamic Simulation
Application in Modern Power Plant Control and Design,
presented to the Advances in Power System Control,
Operation & Management Conference, Hong Kong, 1993.
4. Peet, W.J. and Leung, T., Development and Appli-
cation of a Dynamic Simulation Model for a Drum Type
Boiler with Turbine Bypass System, presented to the
International Power Engineering Conference, Singapore,
1995.
References
Figure 9 Transient response for 5% load ramp from 55% to 90%
MCR (hybrid sliding pressure mode).

Вам также может понравиться