Beruflich Dokumente
Kultur Dokumente
May 2012
Index
1 Introduction .......................................................................................................... 1
3.5 Correlation of Power Output and of Live Steam Pressure Control Loop
Performances ................................................................................................. 50
6.1 Overview of Power Plant Units that Provided Measurement Data ................ 100
8 Control Loop Performance and Economic Efficiency of a Power Plant ............ 107
8.1 Economic Efficiency of a Power Plant Operation and Influencing Factors.... 107
1
Empty unit line means that the unit depends on the application
Rm/104/ϑ Mean creep rapture strength for 10000 hours at the temperature ϑ
Rm/105/ϑ Mean creep rapture strength for 100000 hours at the temperature ϑ
5
Rm/2·10 /ϑ Mean creep rapture strength for 200000 hours at the temperature ϑ
STDT standard deviation
T considered time interval [min]
TR Theoretical lifetime of a thick-walled component to its rupture [h]
u(t) manipulated variable
uFB(t) controller output
uFF(t) feedforward controller output
vA Efficiency coefficient for isolated branches or openings
vL Ligament efficiency for adjacent branches or openings
w(t) reference variable
wT(t) target value
x(t) control valve position [%]
x'(t) slope of control valve position [%]
y(t) controlled variable
z(t) disturbances
αm concentration factor of the stress due to compressive loading [-]
αt concentration factor of the stress due to thermal loading [-]
βLt coefficient of thermal expansion [1/K]
Δϑw temperature difference in the wall of a thick-walled component [K]
ΔDFi k fatigue of a material due to stress load cycles from the class i,k [%]
Δt sampling time [s]
ν Poisson’s ratio [-]
σm material stress [N/mm²]
σp compressive stress [N/mm²]
σt thermal stress [N/mm²]
Chapter 1 - Introduction 1
1 Introduction
The significance of the topic ‘control loop performance’ in process and control engineering gains more
and more in importance. Firstly, the improvement of the control loop performance in power plants
enhances the operational safety of the power plant. Secondly, a high control loop performance is
necessary for the compliance with the increasing requirements of the electrical grid, e.g. regarding the
control of such technical parameters as frequency and voltage to ensure a safe operation of
distribution and transmission systems. Furthermore, a high control loop performance is a necessary
basis for the fulfillment of statutory regulations concerning an environmentally friendly power
generation. However, most importantly from an economic point of view, a high control loop
performance is essential for a flexible power plant operation, e.g. increasing the unit efficiency and
flexibility, reducing the wear of the actuators or extending the lifetime of power plant components.
Control loop performance can be quantized by defining control loop performance indicators. A
consistently defined set of control loop performance indicators can be applied in various fields. Against
the background of power plants, it can give reference values for:
• Tender and purchase, e.g. for the economic assessment of desired/expected control loop
performances during the tendering phase of new power plants or retrofits,
• Verification, e.g. for the verification of the practically reached control loop performance on the
part of both manufacturers and customers,
• Monitoring, e.g. for the control loop performance monitoring during power plant operation
using control loop performance monitoring systems,
• Benchmarking, e.g. as a benchmark for the comparison of different existing power plants, for
the technical and economical assessment of retrofits as well as for the control of success
during the commissioning of new control concepts.
This VGB research project is aimed to define a practicable methodology for control loop performance
evaluation, to develop practically applicable control loop performance indicators and benchmarks, and
last but not least to support suppliers and customers according to definitions of the respective
requirements.
The developed control loop performance indicators, achievable control loop performances and
benchmarks are not only aimed at a secure plant operation and the fulfilment of the requirements
given by the electrical grid, as is the target of the majority of the few existing guidelines. They are also
to represent the state-of-the-art control loop performance regarding a modern and economically
efficient power plant operation.
Besides, the customer’s requirements often differ from the supplier’s declarations and from the
practically considered values. Particularly different operational modes (steady-state operation,
ramping, frequency control) are not considered in the statements of the required or achievable control
loop performance. This can lead to high requirements concerning the control loop performances of
new built power plant units or retrofits, which are not achievable in practice in all operational modes.
Existing guidelines about control loop performance, established indicators and achievable control loop
performances for thermal power plant units are currently rudimentarily included in the series of
guidelines VDI/VDE 35xx (VDI 3500, 1996 - VDI 3508, 2003). However, both operators and
manufacturers consider the information, which is given in these guidelines regarding control loop
performance as insufficient.
The VDE/GMA technical committee 7.12 ‘Leittechnik in konventionellen Dampfkraftwerken' revises
these guidelines every once in a while, in order to keep them on the current state of the art. Finally,
the guideline VDI/VDE 3507 was revised. This guideline deals with the technical approval of steam
generator controls. The official draft of this guideline has been published on November, 1st, 2010.
Today, a high control loop performance is not only a guarantee of secure system operation and the
fulfilment of system rules given by the transmission system operator. A high control loop performance
is moreover the necessary basis for an economical and flexible power plant operation. Therefore, a
good control loop performance today has a bigger economical relevance for the power plant operators
compared to several years ago. This is reflected by the fact that, many operators improve the digital
control system of their power plant units and invest in an even better control performance. Against this
background, existing control loop performance definitions and indicators are insufficient because of
today’s various requirements to the control behaviour of power plant units. In particular, there are no
investigations of the correlation between a high control loop performance and its economical impact
on power plant operation, e.g. by increasing unit flexibility and efficiency or reducing the wear of
actuators or increasing the lifetime of thick-walled components.
The stated issues have partly entailed negative experiences both on the operator and manufacturer
side concerning the compliance and verification of control loop performances in power plants, both
turnkey-projects and retrofits.
Different power plant types, control loops, operational modes and operating modes represent the
application framework for the systematically defined control loop performance indicators (see
Figure 1).
2 Theoretical Background
Chapter 2 covers the description of the considered process technology, the definition of control loop
variables and of the considered control loop performance indicators.
Boiler Types
When considering the Benson boiler, it is important to distinguish between once-through and low load
operation.
Unlike the drum boiler, the once-through boiler has no drum. This means that during the once through
operation the fluid passes through the economiser, evaporator and superheater without any
recirculation. Hence, the evaporation endpoint is not fixed locally. It moves depending on load as well
as depending on balance between the fuel and the feed water mass flow. The feed water mass flow
influences not only the evaporation endpoint, but also the steam mass flow, the steam temperature
and the steam pressure. This is why, it is very important for the feed water control to adjust the feed
water mass flow precisely to the fuel mass flow. Besides, the feed water control, the fuel mass flow
control, the steam temperature control and other Benson boiler controls must be coordinated [17].
For the rapid power output changes during the frequency control operation, the boiler must be able to
provide a certain amount of the steam at a specific point of time. The Benson boiler is able to store
less steam than the drum boiler. If the Benson boiler provides the necessary amount of steam at a
specific point of time, it will lead to large steam pressure deviations. For this reason, higher
requirements are made on the control of the steam pressure [17].
During low load operation the recirculation will take place in a Benson boiler. The recirculation is
obtained by means of the separator vessel, which has the same function as the drum in a drum boiler.
The aim of the recirculation is to avoid the flow of the water instead of the steam to the superheaters
and to the turbine in order to prevent these components from damage [17].
In contrast to the Benson boiler, the Sulzer boiler operates by means of the separator vessel during
low load operation as well as during the once-through operation. Due to the separator vessel, the
evaporation endpoint is fixed locally. The water level in the separator vessel represents the balance
between the feed water and the fuel mass flow. Depending on the water level in the separator vessel,
the feed water mass flow has to be changed. The storage capacity of the Sulzer boiler is slightly larger
than the one of the Benson boiler [17].
type and natural circulation type. The most common type of the HRSG in large combined cycle power
plants is the forced circulation type [3]. Accordingly, the control system of a HRSG corresponds to its
type.
The power output changes of a combined cycle power plant without supplementary firing are
controlled by means of the gas turbine. This is due to the fact that the gas turbine is able to react very
quickly to load variations. After a load change the steam turbine follows the gas turbine with a few
minutes delay depending on the response time of the HRSG [15].
In supplementary fired HRSGs an additional heat is produced by the supplementary firing in order to
increase the steam production. Therefore, the power output changes of the combined cycle power
plants with supplementary firing can be controlled either by means of the gas turbine only or by means
of the gas turbine in combination with the steam turbine.
Feedforward
Controller uFF(t) Disturbances z(t)
The desired output of a system is called the target value wT(t). The Pre-Filter accounts for soft
changes of the reference variable wT(t). The filtered target value is the reference variable w(t). The
output of the system is the controlled variable y(t). The process and accordingly the controlled
variable y(t) are influenced by disturbances z(t). The difference between the reference value w(t) and
the controlled variable y(t) is called control deviation e(t). The feedback controller uses the control
deviation e(t) to generate the feedback controller output uFB(t). The sum of the uFB(t) and the
feedforward controller output uFF(t) is the input of the process, which is called manipulated
variable u(t).
a)
w(t)
y(t),w(t)
O-T
y(t),
b)
-T
O-T
O
0
t1 t1+T t
In the first step the indicator negative overshoot O-T is applied to the measurement data over the time
interval [t1, t1+T] (see Figure 5-a). In doing so the first value of the negative overshoot O-T is
determined for the time interval [t1, t1+T] as shown in Figure 5-b. Afterwards, the time window of length
T is moved by one time sample and the measurement data over the time interval [t2, t2+T] are
considered. The second value of O-T is determined by evaluating the data in the second time window
(see Figure 6). In the next step the O-T is determined for the data from the third time window with the
time interval [t3, t3+T] (see Figure 7). In this way the time window is moved along the entire curve until
the last time window with the time interval [tend-T, tend-T+T] is reached. After that, the entire
measurement data curve is evaluated. The achieved values of the negative overshoot O-T for all
considered time windows are shown in Figure 8-b.
Chapter 2 - Theoretical Background 8
a)
w(t)
y(t),w(t)
O-T
y(t),
y(t) Actual value
w(t) Set point
t2 t2+T t
b)
-T
O-T
O
0
t2 t2+T t
a)
w(t)
y(t),w(t)
O-T
y(t),
b)
-T
O-T
O
0
t3 t3+T t
The results of the measurement data evaluation are represented in form of a histogram in Figure 8-c.
This histogram demonstrates the percentage distribution of the achieved values of the negative
overshoot O-T. To create this histogram each determined value of the negative overshoot O-T is plotted
on the x axis and its percentage distribution on the y axis. The sum of the percentage distributions on
the y axis amounts to 100%.
Chapter 2 - Theoretical Background 9
a)
y(t), w(t)
y(t), w(t)
0 T t
b)
O -T
O-T
0
0 T t
[%]
O-T-T[%]
c) 100
of O
Distribution of
50
Distribution
0
0
O
O-T-T
2.3.2 Overshoot
The overshoot is the maximum deviation of the controlled variable y(t) from the reference
variable w(t).
The positive overshoot O+T in the time interval [t, t+T] is the difference between the positive peak
value of the controlled variable M+T in the time interval [t, t+T] and the set point value w.
O+ T = M + T − w (2.1)
The negative overshoot O-T in the time interval [t, t+T] is the difference between the set point value
w and the negative peak value of the controlled variable M-T in the time interval [t, t+T].
O−T = w − M −T (2.2)
The peak-to-peak value PTPT in the time interval [t, t+T] is the measure between the maximum
positive and the maximum negative deviations of the controlled variable from the set point in the time
interval [t, t+T]. Peak-to-peak value PTPT in the time interval [t, t+T] equals the sum of the positive and
the negative overshoot in the time interval [t, t+T] (see Figure 9).
The control loop performance indicators O+T, O-T and PTPT are to be applied to the measurement data
with a constant set point. Figure 8, Figure 9 and Figure 10 represent determined O+T, O-T and PTPT of
exemplary measurement data as well as their percentage distribution in form of histograms. The
smaller the values of O+T, O-T and PTPT are, the better is the control loop performance.
Chapter 2 - Theoretical Background 10
a)
y(t), w(t)
y(t), w(t)
0 T t
b)
O +T
O+T 0
0 T t
+T [%]
[%]
c) 100
O+T
of O
Distribution of
50
Distribution
0
0
O+T
O+T
a)
y(t), w(t)
y(t), w(t)
0 T t
b)
PTPT
PTPT
0
0 T t
TT [%]
PTP [%]
c) 100
of PTP
Distribution of
50
Distribution
0
0
PTPT
PTPT
The determined MVDT-values of exemplary measurement data and their percentage distribution are
shown in Figure 11. The closer to zero the values of MVDT are, the better is the control loop
performance.
a)
0
-e(t)
-e(t)
0 T t
b) 0
MVD T T
MVD
0 T t
T [%]
MVD T [%]
c) 100
of MVD
Distribution of
50
Distribution
0
0
MVDT
MVDT
1 t +T 1 t +T
IAE T = ∫ e(τ )dτ = ∫ w (τ ) − y (τ )dτ (2.5)
T t T t
Chapter 2 - Theoretical Background 12
The integration of the absolute value of the control deviation over a predefined time interval [t, t+T]
yields a finite value for any signal, even if it never reaches its steady-state. Division of the result by the
time span T normalizes the IAET and makes it comparable.
Figure 12 represents the determined IAET-values of exemplary measurement data and their
percentage distribution.
a)
|e(t)|
|e(t)|
0
0 T t
b)
IAETT
IAE
0
0 T t
[%]
T
IAETT[%]
c) 100
of IAE
Distribution of
50
Distribution
0
0
IAE
IAETT
1 t +T 1 t +T
T ∫t T ∫t
ISE T = e ²(τ )d τ = (w (τ ) − y (τ ))²dτ (2.6)
Figure 13 shows the determined ISET-values of exemplary measurement data and their percentage
distribution.
Chapter 2 - Theoretical Background 13
a)
e²(t)
e²(t)
0
0 T t
b)
ISET
ISET
0
0 T t
[%]
ISETT[%]
c) 100
of ISE
Distribution of
50
Distribution
0
0
ISE
ISETT
The computation of the standard deviation in the time frame [t, t+T] is described by the following
formula:
N
1
STDT =
N
∑ (en − eT )2 (2.7)
n =1
where N is a number of samples in the time frame [t, t+T], en represents the sample number n of the
control deviation e(t) in the time frame [t, t+T] and eT is the arithmetic mean value of the control
deviation in the time frame [t, t+T] defined as:
N
1
eT =
N
∑ en (2.8)
n =1
Measurement data with a bell shaped probability density function can be approximated by a normal
distribution or Gaussian distribution. When the data samples are tightly bunched together and the
bell-shaped curve is steep, the standard deviation is small. When the samples are spread apart and
the bell curve is flat, the standard deviation is large [23].
The following conclusions are allowed for normally distributed measurement data (see Figure 14) [23]:
• 68.27 % of the measurement data are within the range of one standard deviation eT ±STDT,
Chapter 2 - Theoretical Background 14
• 95.45 % of the measurement data are within the range ofeT ±2STDT,
• and 99.73 % of the measurement data are within the range of eT ±3STDT.
Nk
PD kT = ⋅ 100% (2.9)
NT
where NT is number of e(t)-samples within the time frame [t, t+T] and Nk is the number of e(t)-samples
within the range k in the time interval [t, t+T].
For the demonstration of the control loop performance indicator PDkT five following ranges are defined
(see Figure 15):
• Range [-A;A];
• Range [-B;-A) & (A;B];
• Range [-C;-B) & (B;C];
• Range [-D;-C) & (C;D];
• Range [-∞;-D) & (D;+∞];
The range [-A;A] is always close to zero and is the range of desired deviations. The larger the number
of samples in the range [-A; A] is, the larger is the PD[-A; A] T and the better is the control loop
performance. The larger are the control deviations, the larger are the PD[-B;-A)&(A;B] T, PD[-C;-B)&(B;C] T,
PD[-D;-C)&(C;D] T and PD[-E;-D)&(D;E] T.
E
D
C
B
A
-e(t)
-e(t)
0
-A
-B
-C
-D
-E
0 T
t
Figure 15: Negative control deviation of exemplary measurement data and defined ranges
Chapter 2 - Theoretical Background 15
50 50 50
0 0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
PD[-A;A] T [%] PD[-B;-A)&(A;B] T
[%] PD[-C;-B)&(B;C] T [%]
Distribution of PDT [%]
50 50
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
PD[-D;-C)&(C;D] T [%] PD(-∞ ;-D)&(D;+∞) T [%]
Figure 16 represents the percentage distribution of the samples in all defined ranges. The first
histogram, which includes the percentage distribution of the range [-A; A] shows that:
• 95 to 100 % of all measurement data samples are concentrated in the range [-A; A] in 90 % of
all considered time windows and
• 90 to 95 % of all measurement data samples are concentrated in the range [-A; A] in 5 % of all
considered time windows.
Measurement data samples, which have exceeded the range [-A; A], are distributed in the following
ranges [-B;-A)&(A;B], [-C;-B)&(B;C] and [-D;-C)&(C;D]. In these ranges are located from 0 to 5 % of all
measurement data samples in all considered time windows.
In the histogram including the percentage distribution of the range [-∞;-D)&(D;+∞] are no data,
because there are no measurement data samples in this range.
∑ Dn ∑ x(t n ) − x(t n − ∆t )
D= n =2
= n =2
(2.10)
T T
where N is number of samples within the considered time frame T, Δt is the sampling time, Dn
represents the regulating distance between two neighbouring samples and x(tn) is an actuator position
at the time tn:
Chapter 2 - Theoretical Background 16
x(t)
N-1
N
x(tn)
Dn
x(tn-Δt)
tn-Δt tn t
T
x (t n ) − x (t n − ∆t )
x ' (t n ) = (2.11)
∆t
Figure 18 shows an example for the actuator position x(t) and its slope x'(t). If the actuator position x(t)
increases (see Figure 18-a), then its slope x'(t) will be positive (see Figure 18-b). If the actuator
position x(t) decreases, then its slope x'(t) will be negative (see Figure 18). If the actuator position x(t)
doesn’t change, then its slope x'(t) will equal to zero. A change of the sign of the x'(t) represents a
shift in direction of the actuator position. A zero slope means no shift in direction of an actuator.
a) x(t)
+ _ 0
+ _
+ _
x(tn) +
+
x(tn-Δt)
tn-Δt tn t
T
b) x'(t)
+
+ +
+ +
0
0
_ _ _
t
T
Figure 18: Actuator position x(t) and slope of actuator position x'(t)
The number of shifts in direction of actuator position x(t) - NSD - equals the number of changes
of sign of x'(t) within the time frame T.
The calculation of NSD is carried out in two steps (see Figure 19):
Step 1: Removing of 0-values from x'(t) → yields to x'n
Step 2: Calculation of NSD by following formula:
Chapter 2 - Theoretical Background 17
1 N
∑ sgn(x 'n ) − sgn(x 'n−1 )
2 n =2
NSD = (2.12)
T
in which:
+ 1 if x ' n > 0
sgn( x ' n ) := (2.13)
− 1 if x ' n < 0
Step 1 x'(t)
+
+ + +
+
0
0
_ _
_
t
T
Step 2 x'n
+
+ + + +
0
_ _
_
Positive Overshoot O+T Indicates the maximum positive deviation of the controlled variable
from the set point
Negative Overshoot O-T Indicates the maximum negative deviation of the controlled variable
from the set point
Peak-to-Peak Value PTPT Indicates the distance between the maximum positive and the
maximum negative deviations of the controlled variable from the set
point
Mean Value Deviation MVDT Indicates the mean value deviation of the controlled variable from the
set point
Integral of Absolute Error IAET Indicates the positive and negative deviations of the controlled
variable from the set point and penalizes large deviations as strong as
small ones.
Integral of Squared Error ISET Indicates the positive and negative deviations of the controlled
variable from the set point and penalizes large deviations stronger
than small ones by squaring of the control deviation
Standard Deviation STDT Indicates how tightly the measurement data are clustered around the
mean value of the controlled variable
Percentage Distribution PDk Indicates the percentage duration of deviations within defined ranges
Number of Shifts in Direction NSD Indicates the number of shifts in direction of an actuator
Figure 20: Block diagram of the steam temperature control system [37]
Steam
mass flow ϑI ϑO
Superheater
Attemperation PG Generator
spray water I Inlet
O Outlet
SP Set Point
The superheater inlet temperature ϑI is the auxiliary controlled variable. The task of the subordinated
control loop (the follow-up controller) is to maintain the temperature ϑI at its set point ϑI,SP. The
position of the attemperation control valve is the manipulated variable. The subordinated control loop
must be able to deal with the nonlinearity and possible wear of the control valve. Moreover, the
subordinated control loop has to be faster than the superposed control loop.
Most of the current live steam temperature control concepts are designed on the basis of such a
cascade structure. However, these concepts differ essentially by the used master controller. In the
simplest case it can be a PI controller. Nevertheless, the controller shouldn’t be aggressive to avoid
oscillations interacting with the slow controlled system. Hence, the control loop performance that can
be achieved by means of PI controller is often not sufficient [30]. To reach a better control loop
performance more sophisticated controllers are often used. This can be, for example, control
structures with declining feedback [31] or state controllers [5], [13], [30]. In any case, the
parameterization of the controller must be adapted very well on the actual superheater and requires a
high optimization effort.
Chapter 3 - Control Loops Considered 21
colour. The corresponding negative control deviation is shown in Figure 23-b. The considered time
period in this example is 420 minutes (7 hours). The data are evaluated by means of time windows
with the time interval T=60 min.
540
530
0 50 100 150 200 250 300 350 400 450
t [min]
b) 10
LS Temperature CD [K]
-10
-20
| 0 50 100 150 200 250 300 350 400 450
t [min]
Figure 23: Exemplary live steam temperature measurement data and their negative control deviation
3.1.5.1 Overshoot
The only indicator for the evaluation of the live steam temperature control loop performance, which is
included in actual guidelines, is the overshoot. In this research project it is distinguished between the
positive overshoot O+, the negative overshoot O- and the peak-to-peak value PTP. These indicators
indicate the maximum deviations of the live steam temperature from the set point. If the live steam
temperature control is able to keep the live steam temperature deviations and correspondingly the
values of O+T, O-T and PTPT as small as possible, then it will be possible to increase the live steam
temperature set point in order to achieve a higher efficiency of a power plant unit.
The results of the application of the indicators O+T, O-T and PTPT to the exemplary measurement data
are shown in Figure 24. Figure 24-a shows that the maximum positive deviations of the live steam
temperature O+T are concentrated in the range from 2 to 3 K in about 70 % of all evaluated time
windows. However, there are some large maximum positive temperature deviations in the range from
8 to 8.5 K. Since the percentage distribution of these large temperature deviations is lower than 20 %,
it can be concluded that they do not occur often.
Figure 24-b shows that the maximum negative temperature deviations O-T are larger than the
maximum positive temperature deviations O+T. Most values of O-T are concentrated in the range from
2.5 to 6 K. However, there are some large maximum negative temperature deviations in the range
from 14 to 14.5 K. The percentage distribution of these large O-T-values is lower than 20 %. Therefore, it
can be also concluded that these large O-T-values do not occur often.
Chapter 3 - Control Loops Considered 23
In Figure 24-c the percentage distribution of PTPT-values is shown. The PTP-T-values are
concentrated in the range from 5.5 to 8 K in about 80 % of all evaluated time windows. The large
PTPT-values in range from 14 to 14.5 K are determined only in 15 % of all evaluated time windows.
a)
Distribution of O+T [%]
100
Discretization 0.5 K
80
60
40
20
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
O [K]
+T
b) c)
are concentrated in the first two ranges (range [-3; 3] and range [-5;-3)&(3;5]) and that there are a few
steam temperature samples in further three bands
Distribution of PD [%]
Distribution of PD [%]
Distribution of PD [%]
100 100 100
Discretization 5 % Discretization 5 % Discretization 5 %
T
T
80 80 80
60 60 60
40 40 40
20 20 20
0 0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
PD[-3;3] T [% ] PD[-5;-3)&(3;5] T
[% ] PD[-7;-5)&(5;7] T
[% ]
Distribution of PD [%]
Distribution of PD [%]
100 100
Discretization 5 % Discretization 5 %
T
T
80 80
60 60
40 40
20 20
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
PD[-9;-7)&(7;9] T
[% ] PD(-∞ ;-9)&(9;+∞ ) T
[% ]
curve doesn’t cause any additional lifetime consumption of thick-walled components due to creep
damage and leads to a slight efficiency loss.
b) c)
[%]
80 80
Distribution of IAE
60 60
40 40
20 20
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 2 4 6 8 10 12 14 16 18 20 22 24
IAE [K] ISE [K²]
T T
A high performance of a live steam temperature control loop, which implies small deviation of the live
steam temperature from its set point, requires high manipulating effort. If the manipulating effort
increases, the wear of attemperator control valves will be higher. In order to represent the complete
performance of a live steam temperature control loop, it is necessary to evaluate the wear of
attemperator control valves.
Factors, which cause the wear of attemperator control valves, are, among others, a wide range of
regulating amplitude of a control valve and rapid changes of the control valve position. The regulating
amplitude of a control valve is evaluated by means of the control loop performance indicator ‘covered
regulating distance D’. Evaluating changes of the control valve position, the control loop performance
indicator ‘number of shifts in direction NSD’ is used. The following example shows the application of
both control loop performance indicators to the measurement data of six attemperator control valves.
The measurement data are taken from a different power plant than in the previous example. This
power plant has two tracks each with three attemperation stages ATT1, ATT2, ATT3 (see Figure 27).
Figure 28 shows two charts presenting the results of the evaluation of attemperator control valves
measurement data. The first chart (see Figure 28-a) represents the results for the covered regulating
distance D and the second chart (see Figure 28-b) for the number of shifts in direction NSD.
Figure 28-a shows that, the covered regulating distance of the considered control valves increases
from stage ATT1 to stage ATT3. Besides, the covered regulating distance in both tracks of each
attemperation stage is quite similar.
Chapter 3 - Control Loops Considered 26
ϑ31 ϑ32
ϑC31 ϑC32
ATT31 ATT32
ϑ21 ϑ22
ϑC21 ϑC22
ATT21 ATT22
ϑ11 ϑ12
ϑC11 ϑC12
ATT11 ATT12
Track 1 Track 2
Figure 27: Power plant tracks each with three attemperation stages ATT1, ATT2, ATT3
Figure 28-b shows that, the number of shifts in direction also increases from the stage ATT1 to stage
ATT3. However, there is a big difference between the tracks in the stage ATT3. The number of shifts in
direction of the control valve from track 1 at stage ATT3 is 1.75 times larger than the number of shifts
in direction of the control valve from track 2 at the stage ATT3, even though the corresponding covered
regulating distance of these attemperator control valves is quite similar in both tracks.
a) b)
10 4
8
3
6
2
4
1
2
0 0
1 2 3 1 2 3
ATT1 ATT2 ATT3 ATT 1 ATT2 ATT3
Track 1
Track 2
Figure 28: Calculated covered regulating distance and number of shifts in direction
The increase of D and NSD from the stage ATT1 to the stage ATT3 can be explained by the fact that,
the settings of the follow-up controllers in the first attemperation stage are, as a rule, intentionally
slower and the higher the attemperation stage, the more sensitive are the settings of controllers.
Possible reasons for different NSD-values in the stage ATT3 could be among others:
• different dimensions of attemperator control valves
• different characteristic diagrams of attemperator control valves
• asymmetry of fire in the boiler
• inappropriate action of follow-up controllers
NSDϑI
NSDINFLOW
NSDCV
NSDCD NSDϑI,SP
Figure 29: Application of NSD to the measurement data of steam temperature control loop
At first, the reaction of a control valve to possible disturbances of the inflow steam temperature
fluctuations are considered (see Figure 30).
Chapter 3 - Control Loops Considered 28
NSDϑI
NSDINFLOW
NSDCV
NSDCD NSDϑI,SP
The disturbances due to the inflow steam temperature fluctuations are to be corrected by means of the
control valve attemperation. If there are a lot of fluctuations in the inflow steam temperature, then the
number of shifts in direction of the inflow steam NSDINFLOW is large. If the number of shifts in direction
of the attemperator control valve NSDCV is also large, then the action of the follow-up controller is
suitable, because the control valve tries to compensate the steam disturbances by means of
attemperation.
If NSDINFLOW is small and NSDCV is also small, then the follow-up controller action is suitable too,
because there are not a lot of steam temperature fluctuations and the control valve shouldn’t move.
If NSDINFLOW and NSDCV differ strongly from each other, then the action of the follow-up controller is
questionable.
Figure 31 shows an example, which demonstrates the results of the application of the control loop
performance indicator NSD to the measurement data of six attemperator control valves (see
Figure 31-a) and the corresponding inflow steam temperatures (see Figure 31-b). In this example the
measurement data from the same power plant unit are used as in the example of chapter 3.1.5.6.
However, in the present example, a longer period of time is taken into account.
Accordingly, the NSDCV-behaviour shown in Figure 31-a is similar to the NSDCV-behaviour shown in
Figure 28-b: NSDCV increases from the stage ATT1 to the stage ATT3 and there is a big discrepancy of
the NSDCV between the tracks in the stage ATT3.
Figure 31-b shows that, the NSDINFLOW is relatively small and quite similar within all attemperation
stages. This means that, the signals of the inflow steam temperature are relatively calm in different
tracks of all attemperation stages.
a) b)
Number of Shifts in Direction [1/min]
4 4
3 3
2 2
1 1
0 0
ATT1 ATT2 ATT3 ATT1 ATT2 ATT3
ATT1 ATT2 ATT3 ATT1 ATT2 ATT3
Track 1
Track 2
Figure 31: NSD of attemperator control valve position and of steam temperature before attemperation
Chapter 3 - Control Loops Considered 29
By comparison of both charts in Figure 31 it can be suggested that, the big difference of NSDCV in the
stage ATT3 is not caused by the disturbances due to the inlet steam temperature fluctuations.
Control valve reaction to further disturbances is described below. These disturbances occur due to
possible superheater inlet steam temperature fluctuations, flue gas heat fluctuations or steam mass
flow fluctuations (see Figure 32).
Fluctuations of the superheater inlet steam temperature ϑI lead to the deviations of the superheater
outlet temperature ϑO from its set point ϑO,SP (see Figure 32). Moreover, fluctuations of the flue gas
heat and of the steam mass flow lead to the deviations of the outlet temperature ϑO from its set point
ϑO,SP too. Besides, the superheater outlet temperature deviation from its set point ϑO,SP corresponds to
the input signal of the master controller. In order to correct these deviations, the master controller
changes his output signal, which represents the inlet steam temperature set point ϑI,SP. Fluctuations of
the ϑI,SP and ϑI affect the control deviation signal located in the input of the follow-up controller (see
Figure 32). According to the fluctuations of the follow-up controller input, the follow-up controller output
will also fluctuate. Thereby, the follow-up controller output signal controls the attemperator control
valve.
The more the signals fluctuate, the larger is their NSD. If the values of NSDϑI, NSDϑI,SP, NSDCD and
NSDCV don't differ strongly from each other, the action of the follow-up controller will be suitable. If
there is a big difference between the values of NSDϑI, NSDϑI,SP, NSDCD and NSDCV, the action of the
follow-up controller will be questionable.
NSDϑI
NSDINFLOW
NSDCV
NSDCD NSDϑI,SP
Figure 32: NSD of attemperator control valve position and of control deviation
Figure 33 shows an example, which demonstrates the calculated values of NSDCV, NSDCD, NSDϑI and
NSDϑI,SP.
Figure 33-b shows a small increase of the NSDCD from the stage ATT1 to the stage ATT3 and a small
difference between the NSDCD from different tracks in the stages ATT2 and ATT3. The values of NSDϑI
and NSDϑI,SP are relatively small and quite similar within all attemperation stages (see Figure 33-c and
Figure 33-d). This means that, all considered temperature signals have a relatively small NSD. By
contrast the NSD of considered control valve signals is much bigger. Besides, the big difference of
NSDCV in the stage ATT3 leads to the assumption that, the follow-up controller from track 1 in the third
attemperation stage has too sensitive settings, which are to be checked.
Chapter 3 - Control Loops Considered 30
a) b)
3 3
2 2
1 1
0 0
ATT1 ATT2 ATT3 ATT1 ATT2 ATT3
ATT1 ATT2 ATT3 ATT1 ATT2 ATT3
c) d)
4 4
3 3
2 2
1 1
0 0
ATT1 ATT2 ATT3 ATT1 ATT2 ATT3
ATT 1 ATT2 ATT3 ATT1 ATT2 ATT3
Track 1
Track 2
3.1.6 Indicators for the Evaluation of Steam Temperature Control Loop Performance
Importance Actuators
In●●●●●
the previous chapters different control loop performance indicators were applied to an exemplary
Very high
Control Loop Performance Indicator Considered Effect
●●●●temperature
steam High measurement data. The control loop performance indicator mean value deviation
Covered regulating distance D
●●●T is useful
MVD Middlefor the evaluation of steam temperature
Number ofinfluence
shifts in direction
on NSD
Wear of actuators
the power plant efficiency and on
●● Low
the lifetime consumption of thick-walled components due to creep damage (see
● Very low
Table 2). In order to get additional information about the influence of steam temperature deviations on
the creep damage of thick-walled components, the indicator distribution of deviations PDk is to be
used.
Control loop performance indicators positive overshoot O+T, negative overshoot O-T, peak-to-peak
value PTPT, integral of absolute error IAET and integral of squared error ISET are general control loop
performance indicators.
According to subchapter 4.1 a permanent steam temperature deviation influences stronger the thermal
efficiency of a power plant unit and the lifetime of thick-walled components than the steam
temperature fluctuations. Therefore, the control loop performance indicator MVDT is the most
important indicator for the live steam temperature control loop performance. Hence, the importance of
this control performance indicator is always higher than the importance of such control loop
performance indicators as O+T, O-T, PTPT, IAET and ISET.
Moreover, the importance of control performance indicators depends on the operational mode and the
operating mode of a power plant unit. This is due to the fact that the live steam temperature influences
Chapter 3 - Control Loops Considered 31
differently the thermal efficiency and the lifetime of thick-walled components during different
operational modes and at different operating conditions (see subchapter 4.1). The importance of pre-
established control loop performance indicators during different operational modes and at different
operating conditions is described in the following subchapters.
3.1.6.1 Importance of Control Loop Performance Indicators during the Operating Mode
'Modified Sliding Pressure Operation' and 'Natural Sliding Pressure Operation'
If a power plant unit is operated in the "modified sliding pressure operation" or "natural sliding pressure
operation", the live steam pressure set point pSt set changes depending on load and therefore the
importance of control loop performance indicators is affected as follows:
• Full load operation. For economic reasons, the fossil-fired power plants are operated at the
highest possible live steam pressure (near the upper limit of the construction material) during
full load operation. Therefore, the corresponding compressive stress is relatively high. Since the
average steam temperature influences strongly the power plant unit efficiency and the creep
damage of thick-walled components, it is important to keep the steam temperature mean value
as good as possible at its set point during full load operation. Therefore, the importance of the
control loop performance indicator MVDT during full load operation is very high (see Table 2).
Besides, the steam temperature fluctuations should be kept as low as possible. Therefore, such
control loop performance indicators as PDkT, O+T, O-T, PTPT, IAET and ISET are very important
for full load operation (see Table 2).
• Part load operation and low load operation. Due to the decrease of the live steam pressure
during part load operation, the thermal efficiency and the compressive stress in the material
thick-walled components decrease too. Thus, both the steam temperature deviations and the
steam temperature mean value deviation have lower influence on the thermal efficiency and on
the lifetime of thick-walled components during part load operation. Therefore, the steam
temperature deviations and the steam temperature mean value deviation during part load
operation can be allowed to be bigger than during full load operation. Therefore, the importance
of control loop performance indicators such as MVDT, O+T, O-T, PTPT, IAET, ISET or PDkT during
part load operation is slightly lower than during full load operation. For the same reason, the
importance of these indicators during low load operation is slightly lower than during part load
operation (see Table 2).
• Positive and negative load changes require big live steam pressure changes, which cause
big compressive stress load cycles. When additionally large temperature changes occur, big
thermal stress load cycles arise in thick-walled components. The sum of the compressive and of
the thermal stress load cycles results in the total stress load cycles, which can exceed the
allowable design stress limit and consequently lead to the fatigue damage of thick-walled
components. In order to keep the total stress load cycles under the upper limit allowed, the
thermal stress load cycles should be reduced. For this aim, it is necessary to keep the steam
temperature peak-to-peak value during load changes under the limit of allowable steam
temperature peak-to-peak value PTPϑ,fatigue (see Table 2).
Chapter 3 - Control Loops Considered 32
Importance Actuators
●●●●● Very high
Control Loop Performance Indicator Considered Effect
●●●● High
Covered regulating distance D
●●● Middle Number of shifts in direction NSD
Wear of actuators
●● Low
● Very low
Table 2: Indicators for the evaluation of the steam temperature control loop performance, pSt set ≠ const.
In order to evaluate the influence of the steam temperature control on the actuator wear, the control
loop performance indicators covered regulating distance D and number of shifts in direction NSD are
used.
3.1.6.2 Importance of Control Loop Performance Indicators during the Operating Mode 'Fixed
Pressure Operation' and 'Initial Pressure Operation'
If a power plant is operated in the operating mode "fixed pressure operation" or "initial pressure
operation", the live steam pressure set point pSt set is constant [40]. As a rule, this set point doesn't
change during full load operation and during part load operation. Accordingly, there is no change in
this set point during a load change from full load to part load and vice versa. Therefore, the control
loop performance indicators MVDT, PDkT, O+T, O-T, PTPT, IAET and ISET have the same importance
during these operational modes (see Table 3).
Generally, the live steam pressure set point is reduced during low load operation. Therefore, the
importance of defined control performance indicators is lower during low load operation than during
part load operation (see Table 3). During a load change from full load to low load and vice versa as
well as from part load to low load and vice versa there is a change of the live steam pressure set point.
Due to this fact, the indicator PTPT is important during this operational mode (see Table 3). If the live
steam pressure set point isn't reduced during low load operation, the importance of control loop
performance indicators will be the same during all operation.
Chapter 3 - Control Loops Considered 33
Table 3: Indicators for the evaluation of the steam temperature control loop performance, pSt set = const.
Referring to [40], the dynamic behaviour of a power plant unit can be divided into three subprocesses:
Chapter 3 - Control Loops Considered 34
• Fuel supply and steam generation. The input variable of this subprocess is the thermal output
Q F . The output variable is the generated steam mass flow ṁStG. The thermal output, which is
necessary for the evaporation of the feed water and for the generation of the steam, is generated
by the combustion of fossil fuels with the necessary combustion air. This subprocess includes
further subprocesses such as fuel supply (including the mills), heat release, heat transfer,
evaporation of the feed water and the reheating of the steam generated. The changes in the
supply of the fuel mass flow affect the process of the live steam generation with a time delay.
Therefore, the transfer function of the fuel supply and steam generation is represented as a high
order time-delay element in Figure 34. This transfer function is characterized using the equivalent
dead time Tu and the balancing time Tg. The values of Tu and of Tg depend on the steam
generation system and on the type of fuel used.
• Steam storage. The input variables of this subprocess are the steam mass flow generated ṁStG
and the turbine valve opening yT. The output variables are the turbine mass flow ṁT and the live
steam pressure pSt. This subprocess represents the steam mass storage of the steam generator.
The steam mass storage is modelled as an integrator and its volume is characterized by the
storage time constant TS. The difference between the steam mass flow generated ṁStG and the
turbine mass flow ṁT determines the steam pressure pSt. At the same time, the steam pressure pSt
and the turbine valve opening yT determine the steam mass flow ṁT supplied to the turbine.
• Steam expansion, generation of electrical energy. During the process of steam expansion and
generation of electrical energy the thermal power is converted into electrical power. This process
includes components such as the high-pressure section of the turbine, the reheater, the
intermediate-pressure and low-pressure sections of the turbine and the generator. The time
response of the turbine and of the generator is negligible compared to the time response of the
steam generator. The reheater and the connecting pipelines are an additional steam mass
storage, which has the first-order delay effect, as shown in Figure 34.
Figure 34: Representation of the dynamic relationship in a steam power plant unit [40]
The target value of the generator output PGTarg is provided by the grid control. If a power plant unit
takes part in frequency control, an additional signal ΔPG that is proportional to the frequency control
deviation will be added to PGTarg. The sum of these two signals is checked by the target output limiter,
in order to ensure that this sum doesn't exceed the maximum allowable power output and maximum
allowable rate of change. The output signal of the target output limiter is the generator output set point
PG set [40].
During the fixed-pressure operation, the power output PG is controlled by the turbine valve opening yT.
Depending on the magnitude of the power output control deviation, the output controller changes the
turbine valve opening yT. Thereby, the position of the turbine control valve yT is controlled as a function
of power output (see Figure 35). In order to control the rapid load changes, the turbine valve opening
changes rapidly and the steam energy stored in the boiler is used. Due to rapid changes of the turbine
valve opening, the turbine steam mass flow follows the power output set point very quickly and the
power output actual value is controlled very precisely during this operating mode [40].
Figure 35: Steady-state characteristic of the fixed-pressure and of the initial-pressure operation [40]
Chapter 3 - Control Loops Considered 36
Besides, in this guideline the information is given that the verification of a control performance for the
unit control is possible only for the precisely described operation state with predefined test stimuli
under well defined operating conditions. In order to determine the control performance of the unit
control, the following points are to be taken into account:
• The response of the unit control to disturbances. Referring to [40], there are three main
disturbances, which influence the unit control performance. These disturbances are:
o Firing disturbances due to changes in the calorific value of the fuel,
o Disturbances in the discharge of coal dust from the mill,
o Disturbances due to changes of the feed water inlet temperature as a consequence of
disturbances in high-pressure preheaters.
Besides, [40] gives the information that, ‘Proper functioning of the unit control has been
verified if the controlled variable returns to a steady state at its set point after one, at the most
two, overshoots. These controlled variables are:
o the live steam pressure during the operating mode ‘turbine in control’
o the generator output during the operating mode ‘steam generator in control’'
• Response of the unit control to set point changes concerning provision of the
immediate–reserve capacity and the grid primary control. Requirements of the
‘TransmissionCode 2007’ are considered in the guideline VDI/VDE 3508. With respect to
these requirements step-like reduction of frequency from the set point by 200 mHz is to be
compensated by the power plant units supplying the grid. For this aim these power plant units
are to be able to increase their power output linearly by up to 2% of the nominal power output
within 30 seconds. The achievable rates of power output changes in the case of activated
immediate-reserve capacity are shown in Table 5.
• Response of the unit control to ramp-like output changes concerning the grid
secondary control. In the guideline VDI/VDE 3508 the information is given that ‘the response
of the unit output control to influence of the secondary control is verified by performing an
appropriate change of the generator output target.’ After a ramp-like change of the power
output set point, the power output actual value will follow it with the unit delay TUD.
Afterwards, the power output actual value will increase parallel with the ramp of the power
output set point. The values for achievable rates of power output changes in the case of
activated secondary control are shown in Table 5.
Chapter 3 - Control Loops Considered 38
Test stimulus :
output change, ±(2 to 5) ±(2 to 5) 20 to 30 20 to 30
% Pnorm
Test conducted
within percentage of
50 to 90 50 to 90 40 to 90 40 to 90
output range,
% Pnorm
Reference values Fixed Sliding Fixed Sliding Fixed Sliding Fixed Sliding
for achievable rates pressure pressure pressure pressure pressure pressure pressure pressure
of output change, 6 to 12
% Pnorm /min 20 20 20 20 7 to 20 2 4 to 8 3 to 6 2)
)
Table 5: Test stimuli for the verification of the unit control performance and reference values for rates of
power output change [40]
2
These values are valid for power plant units with once-through boiler (circulation steam generators
are considerably slower during the sliding-pressure operation)
Chapter 3 - Control Loops Considered 39
Influence of the power output control on the economic efficiency of the power plant unit is described in
chapter 8.
The power output measurement data in both data sets are normalized and indicated in %. These data
were evaluated by means of time windows with the time interval T=60 min.
a) 100
Set Point
Power Output [%]
99 Actual Value
98
97
96
0 50 100 150 200 250 300 350 400
t [min]
b) 2
Power Output CD [%]
-1
| -2
0 50 100 150 200 250 300 350 400
t [min]
Figure 37: Exemplary power output measurement data and their negative control deviation – data set 1
Chapter 3 - Control Loops Considered 40
a) 110
90
80 Set Point
Actual Value
70
0 50 100 150 200 250 300 350
t [min]
b) 10
Power Output CD [%]
-10
|
-20
0 50 100 150 200 250 300 350
t [min]
Figure 38: Exemplary power output measurement data and their negative control deviation – data set 2
3.3.5.1 Overshoot
The control loop performance indicators O+T, O-T and PTPT represent the maximum deviations of the
power output actual value from its set point. These indicators are to be applied to the power output
measurement data with constant set point. Therefore, these indicators are applied only to the
measurement data that are shown in Figure 37-a. The results of this application are represented in
Figure 39.
a)
[%]
100
Discretization 0.5 %
+T
80
Distribution of O
60
40
20
0
0 1 2 3 4 5 6 7 8 9 10
O+T [% ]
Distribution of PTPT [%]
b) 100 c) 100
Distribution of O-T [%]
Figure 39-a and Figure 39-b show that the maximum positive and negative power output deviations
are concentrated in the range up to 1.5 %. Figure 39-c shows that PTPT-values are concentrated in
Chapter 3 - Control Loops Considered 41
the range from 1 to 2 % in 87 % of all evaluated time windows. In 13 % of all evaluated time windows,
the PTPT-values are concentrated in the range from 2 to 3 %.
a) c)
b) d)
Distribution of IAE T [%]
100 100
Discretization 0.5 % Discretization 0.5 %
80 80
60 60
40 40
20 20
0 0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
MVDT [% ] STDT [% ]
b) d)
Distribution of IAE T [%]
100 100
Discretization 0.5 % Discretization 0.5 %²
80 80
60 60
40 40
20 20
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
IAET [% ] ISET [% ²]
The determined STDT-values of the second data set are concentrated in the range from 2.5 to 3.5 % in
80 % of all evaluated time windows (see Figure 41-c). In the remaining time windows, the values of
STDT are in the range from 0.5 to 2.5 %.
3.3.6 Indicators for the Evaluation of Power Output Control Loop Performance
All control loop performance indicators that are suitable for the evaluation of the power output control
loop performance are summarized in Table 6. The control loop performance indicator mean value
deviation MVDT is useful for the assessment of the power output control influence on the economic
efficiency of a power plant unit. Control loop performance indicators positive overshoot O+T, negative
overshoot O-T, peak-to-peak value PTPT, integral of absolute error IAET and integral of squared error
ISET are general control loop performance indicators, which indicate the power output deviations.
Table 6: Indicators for the evaluation of the power output control loop performance
The operating modes differ from each other in the assignment of the manipulated variables (turbine
valve opening yT and fuel mass flow ṁF) to the controlled variable (live steam pressure pSt) (see
Table 7).
In the operating mode ‘Turbine in control’ it is distinguished between ‘the fixed pressure operation’ and
‘the modified sliding-pressure operation’, which are described below.
In the operating mode ‘steam generator in control’, it is distinguished between ‘the initial pressure
operation’ and ‘the natural sliding-pressure operation’, which are described below.
• Efficiency. Live steam pressure has a significant impact on the efficiency of a power plant
unit. In the pressure range up to 250 bars, a rise of the live steam pressure results in an
efficiency improvement of 0.01% per bar [33]. However, referring to [16], a further increase of
the live steam pressure above 250 bar accounts for a very small portion of the additional
power plant efficiency improvement. For example, in the pressure range up to 300 bar, a rise
of the live steam pressure results in an efficiency improvement of 0.008% per bar [33].
Moreover, a higher pressure is leading to interference of the thermal flexibility and has also a
particularly high effect on costs [16].
• Plant lifetime. The highest possible live steam pressure is limited by available construction
material. This pressure shouldn't be exceeded in order to protect the construction material
from to high compressive stress.
• Load-following capability. A good pressure control is essential for high load gradients.
• Availability. A good steam pressure control reduces the probability of a forced power plant
outage due to the excess of security limits.
constant live steam pressure set point are considered (see Figure 43-a). The corresponding negative
control deviation is shown in Figure 43-b.
In the second data set the live steam pressure measurement data in the modified-sliding pressure
operating mode are considered (see Figure 44-a). The live steam pressure set point is not constant.
The corresponding negative control deviation is shown in Figure 44-b.
The live steam pressure measurement data are normalized and indicated in % in both data sets.
These data were evaluated by means of time windows with the time interval T=60 min.
a) 98
LS Pressure [%]
97
96
95 Set Point
Actual Value
94
0 50 100 150 200 250 300 350 400
t [min]
b) 3
LS Pressure CD [%]
| -1
0 50 100 150 200 250 300 350 400
t [min]
Figure 43: Exemplary LS 3 pressure measurement data and their negative control deviation – data set 1
a) 100
LS Pressure [%]
90
80
Set Point
Actual Value
70
0 50 100 150 200 250 300 350
t [min]
b) 4
LS Pressure CD [%]
-2
| -4
0 50 100 150 200 250 300 350
t [min]
Figure 44: Exemplary LS pressure measurement data and their negative control deviation – data set 2
3
LS means Live Steam
Chapter 3 - Control Loops Considered 47
3.4.5.1 Overshoot
In order to find out, the maximum deviations of the live steam pressure from the constant set point, the
control loop performance indicators positive overshoot O+T, negative overshoot O-T and peak-to-peak
value PTPT are applied to the live steam pressure measurements shown in Figure 45. The results of
this application are represented in Figure 46.
a)
[%]
100
Discretization 0.5 %
+T
80
Distribution of O
60
40
20
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
O+T [% ]
T
-T
80 80
60 60
40 40
20 20
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
O [% ] PTPT [% ]
-T
Figure 46-a shows that the maximum positive live steam pressure deviations O+T are concentrated in
the range from 0.5 to 2.5 %. Figure 46-b shows that all values of O-T are concentrated in the range up
to 1 % and are considerably lower than the values of O+T. Most PTPT-values are concentrated in the
range from 1 to 2.5 % (see Figure 46-c).
In Figure 48-c the ISET-values are concentrated in the range from 0.5 to 1.5 %² in all considered time
window. This means that there are no large live steam pressure deviations from the set point.
a) 100
Discretization 0.5 %
80
60
40
20
0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
MVDT [% ]
Distribution of ISE T [%]
b) c)
Distribution of IAE T [%]
100 100
Discretization 0.5 % Discretization 0.5 %²
80 80
60 60
40 40
20 20
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
IAE [% ] ISE [% ²]
T T
a)
b) c)
100 100
Discretization 0.5 % Discretization 0.5 %²
80 80
60 60
40 40
20 20
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
IAE [% ] ISET [% ²]
T
3.4.6 Indicators for the Evaluation of Live Steam Pressure Control Loop Performance
In the previous chapters different control loop performance indicators were applied to an exemplary
live steam pressure measurement data. Indicators, which are selected for the evaluation of the live
steam pressure control loop performance, are shown in Figure 49. These indicators give the
information about the influence of a live steam pressure curve considered on the efficiency of a power
plant unit, as well as on creep damage of thick-walled components. Furthermore, these indicators
show the ability of the live steam pressure control to compensate disturbances.
Importance
●●●●● Very high
●●●● High
●●● Middle
●● Low
● Very low
Figure 49: Indicators for the evaluation of the live steam pressure control loop performance
Chapter 3 - Control Loops Considered 50
3.5 Correlation of Power Output and of Live Steam Pressure Control Loop
Performances
During the operating mode ‘Turbine in control’, the power output PG is controlled by the turbine valve
opening yT and the live steam pressure pSt is controlled by the fuel mass flow ṁF. The turbine control
valve is a very fast actuator compared to the steam generator. Due to this fact, the power output is
regulated very precisely during the ‘Turbine in control’ operating mode. At the same time, the changes
of the turbine valve position are the disturbances of the live steam pressure control loop. For example,
the turbine control valve opens with an increasing power output set point (see Figure 50). The opening
of the turbine control valve leads to the decrease of the live steam pressure (see Figure 50 and Figure
51). Afterwards, the fuel mass flow increases with a time delay in order to keep the live steam
pressure at its set point. This means, when regulating power output as good as possible, it is
impossible to keep the live steam pressure at its set point. Accordingly, the live steam pressure control
loop performance is worse than the power output control loop performance.
Power Output [%]
100
98
96
0 50 100 150 200 250 300 350 400 450
t [min]
Actual Value
Set Point
LS Pressure [%]
100
95
90
0 50 100 150 200 250 300 350 400 450
t [min]
Actual Value1
Actual Value2
Turbine Valve [%]
200
100
0
0 50 100 150 200 250 300 350 400 450
t [min]
Turbine Valve 1
Turbine Valve 2
Thermal Output [%]
110
100
90
0 50 100 150 200 250 300 350 400 450
t [min]
Figure 51: Impact of the turbine valve opening on the live steam pressure
In order to compare the power output control loop performance with the live steam pressure control
loop performance, the control deviations of the power output and of the live steam pressure are shown
in Figure 52. These control deviations are represented in form of a scatter, in which every power
output sample is represented with its corresponding live steam pressure sample. Thereby, four
quadrants of the scatter can be distinguished:
• Quadrant I, in which the power output actual value is too large and the live steam pressure
actual value is too low
• Quadrant II, in which the power output actual value is too low and the live steam pressure
actual value is too low
• Quadrant III, in which the power output actual value is too low and the live steam pressure
actual value is too large
• Quadrant IV, in which the power output actual value is too large and the live steam pressure
actual value is too large
As shown in Figure 52, the number of samples in the quadrants II and III is considerably larger than
the number of samples in quadrants I and IV. This is due to the fact that, the power output actual value
considered is quite often below its set point (see Figure 37 and Figure 50).
In order to find out the ranges, in which most of the power output samples close to zero and most of
the live steam pressure samples close to zero are concentrated,
• a scatter part with 70 % of the power output samples close to zero and
• a scatter part with 70 % of the live steam pressure samples close to zero
are taken into account, as shown in Figure 53. The crossover of both of these parts defines the
required range. Such ranges are defined for all quadrants of the scatter considered (see Figure 54).
The results of this definition are shown in Table 8.
Chapter 3 - Control Loops Considered 52
Figure 52: Control deviations of the power output and of the live steam pressure in form of a scatter
Taking into account the quadrant II of the scatter considered, it can be seen that if 70 % of power
output control deviation samples are in the range from 0 to 0.7 %, the corresponding live steam
pressure control deviation samples are in the range from 0 to 1.5 %. Table 8 shows that the power
output range in each quadrant is at least twice as small as the live steam pressure range. This leads
to the suggestion that the power output control loop performance is better than the live steam pressure
performance in the example considered.
70%
Figure 53: 70 % of the power output samples close to zero and 70 % of the live steam pressure samples
close to zero
Chapter 3 - Control Loops Considered 53
Figure 54: Ranges of the power output and live steam pressure deviations
Changes of the power output lead to changes of the steam pressure, which cause transient changes
of the drum level. A steam pressure drop leads to drum level rise and vice versa.
Changes of the furnace output cause an increase/decrease of the drum pressure, which results in an
increase/decrease of the steam flow and accordingly in a drop/rise of the drum level.
It can be seen that the significant disturbance variables of the drum level control are:
• Changes of the steam flow ṁD,
• Changes of the steam pressure pSt,
• Changes of the furnace output ṁB.
The single-component control is the simplest form of the drum level control. The only controlled
variable of this control loop is the drum level. The drum level control deviation is fed to the PI controller
as shown in Figure 56 (a). The output of the PI controller is used to position the feed water control
valve. The single-component control is useful during steady-state operation, but it is not useful during
large and rapid load changes since it can lead to large and possibly inadmissible drum level control
deviations [36].
In the case of the triple-component control, not only the drum level control deviation signal, but also
the feed water mass flow signal and the steam mass flow signal are fed to the PI controller as shown
in Figure 56 (b). By means of this control type the feed water mass flow is adjusted to the steam mass
flow. At the same time, the adjustment of the feed water mass flow to the steam mass flow is
proportionally corrected depending on the drum level control deviation. In current practice the triple-
component drum-level controllers are predominantly used [36]. However, the triple-component drum-
level controllers are not suitable at very low loads, during which it is common to switch to the single-
element control [32].
In the case of the triple-component control with cascaded structure, the feed water mass flow is
regulated as a function of the steam mass flow. The adjustment of the feed water mass flow to the
steam mass flow is affected by the superposed PI drum level controller in such a way, that no steady-
state drum level control deviations occur (see Figure 56 (c)) [36].
Chapter 3 - Control Loops Considered 55
Figure 43 represents the measurement data of the drum level actual value and its set point during
steady-state operation. The considered timeframe amounts to 1000 minutes (16.7 hours).
485
Water Level in HP-Drum [mm]
480
475
470
465
460
0 200 400 600 800 1000
t [min]
Actual Value
Set Point
3.7.5.1 Overshoot
The dimensions of drums used in different power plant units typically differ from each other.
Furthermore, the maximum and minimum permissible drum level control deviations can be also
different from each other. In order to be able to compare the drum level control loop performances of
different drums, it is reasonable to consider the drum level control deviations in percent. In doing so it
is to be distinguished between the positive and negative control deviation ranges. The range from the
set point to the maximum allowable water level in the drum is the range of positive deviations. The
range from the set point to the minimum allowable water level in the drum is the range of negative
deviations. Both ranges are represented as the ranges from 0 % to 100 % (see Figure 58). At the
same time the maximum permissible drum level steady-state control deviations are to be in the range
from 0 % to x % and the maximum permissible drum level transitory control deviations are to be in the
range from x % to y % as shown in Figure 58. The closer to zero the drum level control deviation is,
the better is the drum level control loop performance.
Chapter 3 - Control Loops Considered 57
Figure 58: Schematically representation of a drum with minimum and maximum allowable water level
Figure 59-a represents the application of indicators O+ and O- to exemplary drum level measurement
data, which are shown in percent. It can be seen that the maximum drum level control deviation
amounts to 3 % and the minimum drum level control deviation amounts to 3.7 %. Thus, the drum level
control deviations are very small and are far away from maximum/minimum allowable drum level
control deviation limits.
a) 4
Water Level [%]
2
O+=3%
0
-2
O-=3.7%
-4
100 200 300 400 500 600 700 800 900 1000
t [min]
b) 100
Water Level [%]
50
-50
-100
0 100 200 300 400 500 600 700 800 900 1000
t [min]
Figure 59: Application of indicators O+ and O- to exemplary drum level measurement data
3.7.6 Indicators for the Evaluation of Drum Level Control Loop Performance
Reasonable indicators for the evaluation of drum level control loop performance and their relevance
are shown in Figure 60.
Chapter 3 - Control Loops Considered 58
0…x % Maximum permissible range for drum level steady-state control deviation
x…y % Maximum permissible range for drum level transitory control deviation
1* The steady-state drum level control deviations are to be kept within the range 0…x % and
short drum level transitory control deviations are to be permitted within the range x…y %
Figure 60: Indicators for the evaluation of the drum level control loop performance
Significant disturbances of the feed water control in once-through boilers are load changes, changes
in the furnace output, dislocation of the heating due to a change of the heat caloric value, changes in
the attemperator water mass flow, changes of the steam mass flow, changes of the feed water
temperature and of the feed water pressure as well as changes of the attemperation water
temperature and attemperation water pressure [38].
3.8.2.1.1 Enthalpy
The enthalpy at the end of the evaporator h is calculated on the basis of the temperature and pressure
of the slightly superheated steam. The enthalpy actual value h is compared with its set point hset.
Depending on the enthalpy control deviation, the enthalpy controller determines the reference value of
the feed water mass flow. This reference value is additionally influenced by a signal derived from the
power output set point. At the same time, the evaporator mass flow is compared with the minimum
permissible evaporator mass flow value in order to ensure the minimum mass flow through the
evaporator. The comparison of these values is ineffective during once-through operation and takes
effect during low load operation or when the enthalpy controller supplies less feed water due to a
lower actual value of the enthalpy, e.g. during the change from once-through operation to low load
operation [38]. An exemplary structure of the enthalpy control at the end of the evaporator is shown in
Appendix A, Figure 1.
3.8.2.1.2 Temperature
The temperature of the slightly superheated steam at the end of the evaporator can also be used for
the control of the feed water mass flow instead of the enthalpy according to 3.8.2.1.1 [38].
3.8.2.1.3 Ratio of Attemperation Water Mass Flow to Feed Water Mass Flow
The steam temperature in once-through boilers is strongly influenced by the feed water mass flow.
However, the steam temperature is controlled not by the feed water mass flow, but by the
attemperation mass flow, since this leads to better control behavior. At the same time, the feed water
mass flow is regulated in such a way that the steam temperature is kept within the control range. The
ratio of the attemperation water mass flow to the feed water mass flow can be varied depending on the
load. This ratio can also be corrected depending on the temperature at the evaporator outlet in order
to ensure that the steam at the evaporator outlet is slightly superheated [38]. A structure of the feed
water control with ratio of attemperation water mass flow to feed water mass flow as a controlled
variable is shown in Appendix A, Figure 2.
a) 2800
Set Point
Enthalpy [kJ/kg]
Actual Value
2700
2600
2500
0 50 100 150 200 250 300 350
t [min]
b) 50
Enthalpy CD [kJ/kg]
-50
| -100
0 50 100 150 200 250 300 350
t [min]
Figure 61: Exemplary enthalpy measurement data and their negative control deviation
a)
T
Discretization 1 kJ/kg
80
60
40
20
0
-20 -16 -12 -8 -4 0 4 8 12 16 20
MVD [kJ/kg]
T
b) c)
[%]
Distribution of IAE [%]
100 100
Discretization 1 kJ/kg Discretization 50 kJ/kg
T
T
80 80
Distribution of ISE
60 60
40 40
20 20
0 0
0 4 8 12 16 20 24 28 32 36 40 0 200 400 600 800 1000 1200 1400
IAE [kJ/kg] ISE [KJ²/kg²]
T T
Importance
●●●●● Very high
●●●● High
●●● Middle
●● Low
● Very low
Figure 63: Indicators for the evaluation of the evaporator outlet temperature/enthalpy control loop
performance
Chapter 4 - Lifetime Consumption of Thick-Walled Components 63
4.1.1 Creep
Creep is the time dependent, thermally assisted deformation of a component operating under stress
[26]. The creep curve (see Figure 64) shows that the creep strain is a time-dependent strain, which
occurs when a material is subjected to a constant stress σop at a constant temperature ϑop for a longer
period. TR represents the lifetime of a component until its rapture.
ε Rapture
Creep strain
σop=const
ϑop=const
Tr t
σop increases
Creep strain
and/or
ϑop increases
However, thick-walled components don't operate at a constant stress and at a constant temperature,
because during the operation, the operating stress changes and the operating temperature fluctuates
due to occurring disturbances. Increase in operating stress σop and/or in operating temperature ϑop
accelerates the creep and reduces the lifetime of a component to its rupture [45] (see Figure 65).
Chapter 4 - Lifetime Consumption of Thick-Walled Components 64
Operating Steam
Pressure
di 1
σ op ,cyl = ( + )⋅ p (4.1)
2ems vL 2
2
di 1
σ op ,sph =( + )⋅ p (4.2)
4ems v A (ems + d i ) 2
The coefficients vL and vA are to be determined for each thick-walled component individually
depending on its geometrical dimensions by means of the guideline DIN EN 12952-3.
The theoretical lifetime of a component TR is determined as the intersection point of the line of
stress σop with the lower limit of the creep rupture strength (=0.8·Rm/T/ϑ) at the steam temperature ϑ, as
shown in Figure 67 [10].
Řm/T/ϑ=0.8·Rm/T/ϑ
Rm/T/ϑ
lg σ in N/mm²
Řm/104/ϑ
Řm/105/ϑ
Řm/2∙105/ϑ
σop
This intersection point can also be determined mathematically by following formulas [19]:
0.8 ⋅ Rm / 105 / ϑ
lg( )
σm
(lg TR )1 = 5 + (4.3)
0.8 ⋅ Rm / 104 / ϑ
lg( )
0.8 ⋅ Rm / 105 / ϑ
0.8 ⋅ Rm / 105 / ϑ
lg( )
σm
(lg TR ) 2 = 5 + 0.30103 (4.4)
0.8 ⋅ Rm / 104 / ϑ
lg( )
0.8 ⋅ Rm / 2⋅105 / ϑ
The smaller one of the two calculated values TR1 and TR2 represents the theoretical lifetime of a
component TR at the steam pressure p and the steam temperature ϑ [19].
The partial lifetime consumption of a thick-walled component due to creep damage Dcreep,k is calculated
by dividing the whole operating time Top by the theoretical lifetime of a component TR, both at the
steam pressure p and the steam temperature ϑ [10]:
Top / p / ϑ
Dcreep ,k = (4.5)
TR / p / ϑ
The total lifetime consumption of a thick-walled component due to creep damage Dcreep is calculated
as the sum of partial lifetime consumptions [10]:
4.1.3 Comparison of the Lifetime Consumption at the Steam Temperature Set Point
to the One at the Steam Temperature Actual Value
One of the aims of this project is to assess the influence of the steam temperature control loop
performance on the lifetime consumption of thick-walled components. For this aim the lifetime
consumption of a thick-walled component at the actual steam temperature Dcreep,AV is to be compared
to the one at the steam temperature set point Dcreep,SP. In doing so it is to be distinguished between the
absolute lifetime gain/loss ΔLabs calculated by formula (4.9) and the relative lifetime gain/loss ΔLrel
calculated by formula (4.10):
Dcreep , AV
∆Lrel = 100% − ⋅100% (4.8)
Dcreep , SP
If ΔLabs or ΔLrel equals zero, the actual steam temperature doesn't cause any additional lifetime
consumption compared with the steam temperature set point. If ΔLabs or ΔLrel is positive, the actual
steam temperature causes the lifetime gain compared to the steam temperature set point. If ΔLabs or
ΔLrel is negative, the actual steam temperature causes the lifetime loss compared to the steam
temperature set point.
Chapter 4 - Lifetime Consumption of Thick-Walled Components 66
4.1.4 Information Available in Literature about the Impact of the Steam Temperature
on Lifetime Consumption due to Creep Damage
As a rule, the lifetime of an exemplary new thick-walled component is assessed concerning its design
conditions at a constant temperature. As mentioned above, operating conditions differ from design
conditions. Accordingly, the lifetime of a thick-walled component at operating conditions differs from
the one at design conditions.
In order to represent the influence of steam temperature on lifetime consumption due to creep
damage, the lifetime at operating temperature Top is compared with the lifetime at design
temperature Tde. In addition, it should be noted that, the design temperature ϑde, which is described
below, is the design temperature of a steam-water mixture and not the design temperature of a
material. Designing the temperature of a steam-water mixture, the possible deviations of operating
temperature from its set point are taken into account in order to prevent a damage of the material.
Therefore, the design temperature of a steam-water-mixture is always smaller than the design
temperature of a material.
Referring to [25], if the operating temperature ϑop is constant and equals the design temperature ϑde
(see Figure 68), then the lifetime at operating conditions Top is equal to the one at design conditions
with a constant temperature Tde (see (4.9)).
ϑ(t)
ϑop=ϑde
t
ϑde Design temperature
ϑop Operating temperature
Figure 68: Idealised operating steam temperature curve ϑop = ϑde [25]
However, the operating steam temperature ϑop fluctuates during the operation due to occurring
disturbances. Thereby the average of the operating temperature ϑaop could be smaller or bigger than
the design temperature or it could equal the design temperature ϑde. In the following, all these three
cases are described.
Figure 69 represents the average operating temperature ϑaop, which is smaller than the design
temperature ϑde. According to [25], a permanent negative deviation of the average operating
temperature ϑaop from the design temperature ϑde has a positive impact on the lifetime of a material, in
a way that the lifetime at operating conditions Top is higher than the lifetime at design conditions Tde.
(see (4.10)).
At the same time, such a deviation has a negative impact on the efficiency of a power plant unit. In
[18] is indicated that an increase of the operating temperature ϑop by 1 K corresponds to an efficiency
increase by 0.1 %. [11] gives the information that a decrease of the operating temperature ϑop by 10 K
causes an efficiency decrease by 0.5 % (without reheating process) and by 0.2 % (with reheating
process).
ϑ(t)
ϑaop ϑde
t
ϑde Design temperature
ϑaop Average operating temperature
Figure 69: Idealized operating steam temperature curve ϑaop < ϑde [25]
A durable positive deviation of the average operating temperature ϑaop from the design temperature ϑde
affects very unfavorably the lifetime consumption of a material due to creep damage (see Figure 70).
Referring to [25], due to a durable positive temperature deviation, which equals 10 K, the lifetime
calculated at operating conditions Top corresponds to the life calculated at design conditions Tde
approximately by half, depending on the kind of material and on the temperature level (see (4.11)).
In order to represent the sole impact of the steam temperature fluctuations on the lifetime consumption
of a material due to creep damage, the operating steam temperature ϑop is represented as an
idealized sine shaped curve (see Figure 71). Thereby, the average of the operating temperature ϑaop
equals the design temperature ϑde.
During the positive sine half wave, the operating steam temperature ϑop is higher than the design
temperature ϑde and leads to a decrease of the material lifetime. During the negative sine half wave,
the operating steam temperature ϑop is lower than the design temperature ϑde and leads to the
increase of the material lifetime. As the creep damage of a material is assessed by means of the
creep rapture strength curve and the creep rapture strength curves are non-linear, more lifetime gets
lost during the positive sine half wave, than it is possibly won during the negative sine half wave. After
a concluded sine curve, a small lifetime consumptions remains.
Chapter 4 - Lifetime Consumption of Thick-Walled Components 68
ϑ(t)
Δϑ1
ϑaop ϑde
t
ϑde Design temperature
ϑaop Average operating temperature
Figure 70: Idealized operating steam temperature curve ϑaop > ϑde [25]
Referring to [25], due to a sine-shaped temperature curve, which has the amplitude ±10 K, the lifetime
calculated at operating conditions Top is 10 to 20 % lower than the lifetime calculated at design
conditions Tde, depending on the kind of material and on the temperature level (see (4.12)).
ϑ(t)
Δϑo
ϑaop=ϑde
t
ϑde Design temperature
ϑop Operating temperature
ϑaop Average operating temperature
Figure 71: Idealized operating steam temperature curve ϑaop = ϑde [25]
In all represented examples, the operating steam temperature ϑop has a negative impact either on the
lifetime consumption of thick-walled components due to creep damage or on the efficiency of the
power plant unit. As a solution for this problem, Pich suggests in his paper [25] to keep the average
operating temperature ϑaop 2 or 3 K lower than the design temperature ϑde, thereby the maximum
operating temperature ϑop is higher than the design temperature ϑde. At the same time, the duration of
the positive sine wave will be shorter than the duration of the negative sine wave, so that during the
negative sine wave as much lifetime could be won as gets lost during the positive sine wave. If such
operating steam temperature is used, the lifetime at operating conditions Top will be approximately
equal to the lifetime calculated at design conditions at a constant temperature Tde and the efficiency
loss will be very small.
Chapter 4 - Lifetime Consumption of Thick-Walled Components 69
4.1.5.1 Impact of the Steam Pressure on the Lifetime Consumption due to Creep Damage
This subchapter demonstrates the impact of the steam pressure on the lifetime consumption of an
exemplary thick-walled component due to creep damage. The considered thick-walled component is a
superheater collector made of X10CrMoVNb9-1 steel. Table 10 shows the geometrical dimensions of
the collector which are used for the assessment of the operating stress σop. The operating stress σop is
assessed using formula (4.1).
Variable Value Unit
vL 0.8 [-]
di 250 [mm]
ems 95 [mm]
Table 9: Geometrical dimensions of the exemplary superheater collector used for the assessment of σop
It is assumed that the power plant operates at load levels of 100%, 80%, 60% and 40%. Besides, it is
assumed that the power plant operates over a period of 22.8 years (200000 hours / (365 days x 24
hours)) at each load. The operation time at every load operation amounts to 4500 hours per year. The
power plant considered is assumed to operate at the live steam pressure of 285 bar during full load
operation. If the load is lower, the steam pressure is lower too, like shown in Table 10. Using formulas
(4.5) the life time consumption of the steam collector at the steam temperature of 600 °C is assessed.
Table 10 shows that the lifetime consumption during full load operation (100%) is much higher than
the one during the load of 80% and 60%. The lifetime consumption during the load of 40% is very low
and negligible compared to the one at full load operation. This example shows that the steam pressure
level influences heavily the lifetime consumption due to creep damage. Since the highest steam
pressure level is available during full load operation, it is the most critical load for the lifetime
consumption due to creep damage.
Steam Temperature [°C]
Operating Steam
Load Time Pressure 600
Table 10: Lifetime consumption due to steam pressure during different power plant loads
4.1.5.2 Impact of the Steam Temperature Mean Value Deviation on the Lifetime Consumption
due to Creep Damage
This subchapter is about the impact of the permanent steam temperature mean value deviation on the
lifetime consumption of an exemplary thick-walled component. The thick-walled component
considered is the same one as in the chapter 4.1.5.1.
In this example it is assumed that the power plant operates over a period of 22.8 years in an typical
load regime of a base-load power plant, which is shown in Table 11. Besides, the power plant
Chapter 4 - Lifetime Consumption of Thick-Walled Components 70
considered is assumed to operate at the live steam pressure of 285 bar during full load operation. If
the load changes the steam pressure changes too, as shown in Table 12.
Load Operating Time (1 Year) Operating Time (22.8 Years)
[%] [hours] [hours]
100 4500 102600
80 1200 27360
60 1100 25080
40 800 18240
25 600 13680
Table 11: Typical load regime of a base-load power plant
The steam temperature set point is assumed to be 600 °C (see Table 12). Other steam temperatures
values in the Table 12 represent the steam temperature with a mean value deviation from the set
point. Using formulas (4.5), (4.6), (4.7) and (4.8) the life time consumption of the steam collector at all
these temperatures as well as the absolute and the relative lifetime gain/loss are assessed. Table 12
represents the results of this assessment. Table 12 shows that even a small permanent positive steam
temperature mean value deviation, which amounts to 1 K, causes the absolute lifetime loss of 2.1%
and the relative lifetime loss of 11.4%. The negative steam temperature mean value deviations lead to
lifetime gain, but have a negative influence on the power plant efficiency.
Operating Steam
Load Time Pressure Steam Temperature [°C]
595 596 597 598 599 600 601 602 603 604 605
[%] [hours] [bar] Lifetime Consumption Dcreep,k [%]
100% 102600 285 9,33 10,57 11,97 13,56 15,35 17,38 19,36 21,57 24,05 26,83 29,94
80% 27360 228 0.34 0.39 0.45 0.52 0.6 0.69 0.77 0.87 0.97 1.09 1.23
60% 25080 171 0.02 0.03 0.03 0.04 0.05 0.05 0.06 0.07 0.08 0.09 0.1
40% 18240 114 0 0 0 0 0 0 0 0 0 0 0
∑Lifetime Consumption Dcreep [%] 9.69 10.99 12.45 14.12 16 18.12 20.19 22.51 25.1 28.01 31.27
Absolute Lifetime Loss ΔLabs [%] 8.4 7.1 5.7 4.0 2.1 0 -2.1 -4.4 -7 -9.9 -13.2
Relative Lifetime Loss ΔLrel [%] 46.5 39.3 31.3 22.1 11.7 0 -11.4 -24.2 -38.5 -54.6 -72.6
Table 12: Lifetime consumption due to steam temperature mean value deviation
After the consideration of this example it can be concluded, that the steam temperature mean value
shouldn't exceed the steam temperature set point under any operating conditions. The control loop
performance indicator, which is useful for the evaluation of the impact of the steam temperature mean
value deviation on the lifetime consumption due to creep damage, is the indicator mean value
deviation MVDT (see 2.3.3).
4.1.5.3 Impact of the Steam Temperature Fluctuations on the Lifetime Consumption due to
Creep Damage
This subchapter is about the sole impact of the steam temperature fluctuations on the lifetime
consumption of an exemplary thick-walled component. The thick-walled component considered is the
same one as in the chapter 4.1.5.1 and the load regime considered is the same as in chapter 4.1.5.2.
The steam pressure set point amounts to 285 bar and the steam temperature set point amounts to
600 °C during full load operation. In order to represent the steam temperature fluctuations, the steam
Chapter 4 - Lifetime Consumption of Thick-Walled Components 71
temperature curve is represented as an idealized curve (see Figure 72). At the same time, the
average of the fluctuating steam temperature equals the steam temperature set point (see Figure 72).
Table 13 shows the assessed lifetime consumption ΔDcreep,k at load levels of 100%, 80% and 60% as
well as the total lifetime consumption Dcreep representing the sum of ΔDcreep,k at different load levels.
The results are represented for the steam temperature set point 600°C and for the fluctuating steam
temperatures with deviations in the range from ±2 K to ±10 K. Table 13 shows that the lifetime
consumption during full load operation is much higher than the one during part load operation.
Besides, the higher the steam temperature deviations are, the higher is the lifetime consumption
ΔDcreep,k. This fact is obvious concerning ΔDcreep,k at full load operation. However, even large steam
temperature deviations have only a little influence on the lifetime consumption during part load
operation and are negligible during the load of 40%. This example shows that the steam temperature
deviations are to be kept as low as possible during full load operation and may be allowed to be larger
during part load operation.
608
606
Steam Temperature [°C]
604
602
600
598
596
594
592
0 5 10 15 20 25 30 35
t [min]
Actual Value
Set Point
Defining the control loop performance indicator for the evaluation of the steam temperature
fluctuations influence on the life time consumption due to creep damage it is to be considered, that
unexpected disturbances can occur during the power plant operation. These disturbances can lead to
large steam temperature deviations. Even if the deviations have occurred for a few seconds, the
application of such indicators like positive overshoot O+, negative overshoot O- or peak-to-peak value
PTP to the measurement data, leads to the conclusion that the control loop performance is bad. In
Chapter 4 - Lifetime Consumption of Thick-Walled Components 72
order to avoid such conclusions, more sophisticated indicator is to be defined. This indicator should
consider not only the amplitude of the steam temperature deviation but also the duration of the
deviations. It would be reasonable to define some temperature ranges and to allow the steam
temperature deviations to stay within these ranges for a certain time. For example:
• one range could be the range of desired steam temperature deviations,
• another range could be the range of acceptable steam temperature deviations,
• one more range could represent the range of acceptable temperature deviations in case of
disturbances and
• the last one could represent the largest acceptable temperature deviations in case of
disturbances.
The width of a range is to be defined depending on the level of load, e.g. the bandwidth during part
load operation can be allowed to be larger than the one during full load operation. The duration of
deviations within the ranges is to be defined individually. However, the duration of deviations within the
last two ranges is to be considerably lower than the one within the first two ranges.
In order to check the reasonability of such a consideration, the life time consumption of thick-walled
components is to be considered. For this aim the influence of an exemplary steam temperature curve
on the steam collector is considered in scenario 1 (see Table 14). The steam temperature curve
consists of four curve parts with different deviations. The first curve part includes the temperature
deviations of ±3 K, the second part of ±5 K, the third part of ±7 K and the last part of ±9 K (see
Figure 73). The total operating time in this example amounts to 102600 hours during full load
operation, like in the example before. The deviations of ±3 K and of ±5 K are allowed for 39% of the
total time considered in each case (see Table 14). The deviations of ±7 K are allowed for 20% and the
deviations of ±9 K for 2% of the total time. Table 14 shows the assessed lifetime consumption of the
steam collector caused by each steam temperature part ΔDcreep,k as well as the one caused by the
total steam temperature curve at level load of 100% Dcreep,100%. Assessment shows that Dcreep,100%
caused by steam temperature curve consisting of four parts amounts to 18.04% at full load operation
and corresponds to the lifetime consumption caused by the temperature curve with deviations only in
the range of ±5 K (see Table 13).
610
Steam Temperature [°C]
605
600
595
590
0 20 40 60 80 100
t [%]
At the same time four different ranges of steam temperature deviations are defined (see the second
part of Table 14). The duration of deviations within the first range from -3 to 3 K amounts to 77.4% of
Chapter 4 - Lifetime Consumption of Thick-Walled Components 73
the total time considered. The duration of deviations within the second range amounts to 17.7%, within
the third range to 4.6% and within the fourth range to 0.3%.
The operating time during the load of 80% amounts to 27360 hours and during the load of 60% to
25080 hours, like in the example before. The percentage distribution of operating time concerning
steam temperature deviations during part load operation corresponds to the one during full load
operation. However, the steam temperature deviations are allowed to be 1 K larger during the load of
80% and even 2 K larger during the load of 60%. Therefore, the ranges of steam temperature
deviations increased by 1 K at the load of 80% and by 2 K at the load of 60%. Finally, the lifetime
consumption at load level of 80% Dcreep,80% and at load level of 60% Dcreep,60% are assessed. The total
lifetime consumption Dcreep is the sum of Dcreep,100%, Dcreep,80% and Dcreep,60%. Even though different
steam temperature deviations were allowed to take place during different load levels, the total lifetime
consumption amounts to 18.82% and corresponds to the lifetime consumption caused by a steam
temperature with deviations in the range of ±5 K, which amounts to 18.8% (see Table 13). The
absolute lifetime loss calculated by formula (4.7) amounts to 0.7% and is lower than 1%.
Lifetime Ranges of
Operating Operating Steam Steam Temperature Operating
Load Time Time Pressure Temperature Consumption Deviations Time
[%] [hours] [%] [bar] [°C] [%] [K] [%]
100% 40014 39 285 600±3 6.84 [-3;3] 77.4
→
100% 40014 39 285 600±5 7.03 [-5;-3) & (3;5] 17.7
→
80% 10670 39 228 600±6 0.28 [-6;-4) & (4;6] 15.1
→
60% 9781.2 39 171 600±7 0.02 [-7;-5) & (5;7] 13.2
Table 19 shows the results of the lifetime assessment of the steam collector for scenario 2. The
difference between scenario 1 and scenario 2 is the fact that the percentage distribution of operating
time concerning steam temperature deviations is different from the one in scenario 1. The deviations
of ±3 K and of ±5 K in scenario 2 are allowed for 42% of the total time in each case (see Table 14).
The deviations of ±7 K are allowed for 14% and the deviations of ±9 K for 2% of the total time. The
total lifetime consumption Dcreep in scenario 2 amounts to 18.76% and is lower than the one in
scenario 1 and lower than the one due to steam temperature deviations in the range of ±5 K (see
Chapter 4 - Lifetime Consumption of Thick-Walled Components 74
Table 13. This comparison shows that the lower is duration of large steam temperature deviations, the
lower is the lifetime consumption. However, the difference between the absolute lifetime loss in
scenario 1 and scenario 2 is not very big and amounts to 0.06%.
It can be concluded that using such a consideration of a steam temperature curve not only the
amplitude of steam temperature deviations but also their duration are considered. This allows more
detailed consideration of steam temperature deviations.
Operating Operating Steam Steam Lifetime Bands of Temperature Operating
Load Time Time Pressure Temperature Consumption Deviations Time
→
100% 43092 42 285 600±5 7.57 [-5;-3) & (3;5] 17.0
→
80% 11491.2 42 228 600±6 0.31 [-6;-4) & (4;6] 14.5
→
60% 10533.6 42 171 600±7 0.02 [-7;-5) & (5;7] 12.6
4.2.1 Fatigue
Fatigue is a phenomenon occurring in the material due to load cycles of the material stress. Repetitive
material stress changes lead to structural changes in the material and finally to the material cracking.
Stress
PT1
ϑi ϑm Δϑ
T=220
-
applied to every succession of three material stress values σm1, σm2, σm3 using the following method: if
the Boolean expression
is true, then the material stress value σm2 is a relative maximum or a relative minimum and is to be
saved within the chronological sequence of detected extreme values of material stress as shown in
Figure 76 [10].
σm(t) σm(t)
t t
Figure 76: Material stress values and detected extreme values of material stress
Using the chronological sequence of the detected extreme values of the material stress, the stress
load cycles will be detected. The detection and counting of material stress load cycles is based on the
‘Range-Pair-Method’. According to this method, if a stress change (σm1 - σm2) is interrupted by a
smaller stress change in the reverse direction (σm2 - σm3), then this smaller change (σm2 - σm3) forms a
closed hysteresis loop in the stress–strain curve (see Figure 77). Both extremes involved in the
smaller stress change (σm2 - σm3) form a stress load cycle, which is to be classified and counted as
shown in Table 16. Detected stress load cycles with a range lower than 190 N/mm² should not be
added up [10]. According to the guideline DIN EN 19252-4, Δσm≈190 N/mm² is the upper limit of the
range of material stress load cycle, which doesn’t cause any fatigue.
The extreme values σm2 and σm3, which belong to the load cycle that was detected and added up, are
to be deleted from chronological sequence of detected extreme values of stress. The operation of
detection and classification of stress load cycles as well as the operation of deletion of corresponding
extreme values is to be repeated until no more stress load cycles are available within the whole
sequence of detected extreme values.
σm2
σm3
Figure 77: Detection of stress load cycles and deletion of corresponding load cycle values
Chapter 4 - Lifetime Consumption of Thick-Walled Components 77
Figure 78: Methods for the assessment of fatigue, which is caused by RSE [10]
The material fatigue caused by the RSE cannot be assessed in the same way as the material fatigue
caused by the material stress load cycles detected. According to the Annex B.6 of the guideline DIN
EN 12952-4 there are five different methods for the assessment of the material fatigue, which is
caused by RSE:
a) RSE is neglected.
b) Rain-Flow-Method: Each change from an extreme value to an extreme value will be classified
and added up as a half-stress load cycle. The change between the biggest maximum extreme
value and the smallest minimum extreme value will be classified and added up as one stress
load cycle (see Figure 78-b).
c) Each increasing change from an extreme value to an extreme value will be classified and
added up as one stress load cycle (see Figure 78-c).
d) Each decreasing change from an extreme value to an extreme value will be classified and
added up as one stress load cycle (see Figure 78-d).
e) The change from an extreme value to an extreme value with the biggest range will be counted
as one stress load cycle. The extreme values of this stress load cycle are to be deleted from
the RSE (see Figure 78-e). This operation is to be repeated until within the RSE no more
stress load cycles are available
Chapter 4 - Lifetime Consumption of Thick-Walled Components 78
(2 Rm ) 2
2 f a= 2 f va ⋅ f 3 ⋅ (4.16)
(2 Rm )2 − (2 R p 0.2 / ϑ − 2 f va ) 2
• In the plastic range (2fva>2Rp0.2/ϑ)
(2 f va ) 2
2 f a= f 3 ⋅ (4.17)
2 R p 0.2 / ϑ
The temperature ϑwi* is to be calculated by following formula:
ϑwi* = 0,75 Max {ϑwi(σm2), ϑwi(σm3)} + 0,25 Min {ϑwi(σm2), ϑwi(σm3)} (4.18)
Whereas ϑwi(σmi) is the material temperature, which is measured simultaneously to the extreme
value σmi.
Table 16: Example for the fatigue assessment using classified stress load cycles [10] (t*=ϑwi*)
The fatigue damage from each class (i,k) of the material stress load cycles is calculated by following
formula:
ni k
∆DFIk = (4.19)
Ni k
Chapter 4 - Lifetime Consumption of Thick-Walled Components 79
in which ni k is the number of material stress load cycles in the class (i,k) and Ni k is the number of the
material stress load cycles in the class (i,k) until the first crack formation. Ni k is to be calculated
according to Figure B-9 in Annex B of the guideline DIN EN 12952-3.
The total fatigue damage is calculated by following formula:
The steam temperature deviation from its set point is rather small during full load operation (see
Figure 79-b). Afterwards, the temperature deviation becomes bigger during the negative load change
and during part load operation. During the positive load change there is a big negative deviation of the
steam temperature from its set point. This big negative temperature deviation results in a large peak-
to–peak value of the steam temperature, which amounts to 18.6 K.
The temperature difference in the steam collector’s wall Δϑw, which is necessary for the calculation of
the thermal stress, is assessed using the temperature model, shown in Figure 75, by means of the
steam temperature measurement data. More precisely methods for the assessment of the temperature
difference in the wall of a thick-walled component Δϑw are described in [20], [19], [24], [46]. Figure 80
reveals that the bigger the peak-to-peak value of the steam temperature (see Figure 79-a), the bigger
the peak-to-peak value of the temperature difference in the wall of the steam collector Δϑw (see
Figure 80-b). At the same time, the peak-to-peak value of the temperature difference in the wall of the
Chapter 4 - Lifetime Consumption of Thick-Walled Components 80
steam collector Δϑw is lower than the one of the steam temperature. The thermal stress of material σϑ
is proportional to the temperature difference Δϑw (see Figure 80-c and Figure 80-b).
a) 300
LS Pressure [bar]
250
200
150
0 200 400 600 800 1000 1200
t [min]
b) 560
LS Temperature [°C]
550
540
530
520
0 200 400 600 800 1000 1200
t [min]
a) 560
LS Temperature [°C]
550
540
18.6 K
530
520
0 200 400 600 800 1000 1200
t [min]
b)
Temperature Difference [K]
10
5
13.3 K
0
-5
-10
0 200 400 600 800 1000 1200
t [min]
c)
Thermal Stress [N/mm²]
100
50
83 N/mm²
0
-50
0 200 400 600 800 1000 1200
t [min]
Figure 80: Steam temperature, assessed temperature difference and thermal stress
Figure 81 shows the assessed compressive stress, which is proportional to the steam pressure inside
a thick-walled component.
Chapter 4 - Lifetime Consumption of Thick-Walled Components 81
a) 300
LS Pressure [bar]
250
150
0 200 400 600 800 1000 1200
t [min]
b)
Compressive Stress [N/mm²]
220
200
180
50.1 N/mm²
160
140
0 200 400 600 800 1000 1200
t [min]
a)
220
200
180
160
140
0 200 400 600 800 1000 1200
t [min]
Thermal Stress [N/mm²]
b) 100
50
-50
0 200 400 600 800 1000 1200
t [min]
c)
Material Stress [N/mm²]
250
200
150
100
0 200 400 600 800 1000 1200
t [min]
Figure 82 shows the material stress as a sum of the assessed thermal and compressive stresses. It is
obvious that the thermal stress could be minimized, if the steam temperature deviations are very
small. The compressive stress by contrast is inevitable for power plant load changes.
According to the method described in chapter 4.2.2.2 the extreme values of the assessed material
stress were detected (see Figure 91).
Chapter 4 - Lifetime Consumption of Thick-Walled Components 82
260
240
200
180
160
140
120
0 200 400 600 800 1000 1200
t [min]
Figure 83: Chronological sequence of detected extreme values of assessed material stress
Figure 84 shows the detection of the stress load cycles within 11 steps.
Chapter 4 - Lifetime Consumption of Thick-Walled Components 83
1)
Stress [N/mm²] Stress [N/mm²] Stress [N/mm²] Stress [N/mm²] Stress [N/mm²]
300
200
100
0 200 400 600 800 1000 1200
t [min]
2) 300
200
100
0 200 400 600 800 1000 1200
t [min]
3)
300
200
100
0 200 400 600 800 1000 1200
t [min]
4)
300
200
100
0 200 400 600 800 1000 1200
5) t [min]
300
200
100
0 200 400 600 800 1000 1200
t [min]
6)
Stress [N/mm²] Stress [N/mm²] Stress [N/mm²] Stress [N/mm²] Stress [N/mm²] Stress [N/mm²]
300
200
100
0 200 400 600 800 1000 1200
t [min]
7)
300
200
100
0 200 400 600 800 1000 1200
t [min]
8)
300
200
100
0 200 400 600 800 1000 1200
t [min]
9)
300
200
100
0 200 400 600 800 1000 1200
10) t [min]
300
200
100
0 200 400 600 800 1000 1200
11) t [min]
300
200
100
0 200 400 600 800 1000 1200
t [min]
In order to detect the stress load cycles within the calculated RSE, the Rain-Flow-Method (see
Figure 78-b) was used. Using this method, the highest material stress load cycle, which amounts to
124 N/mm², was detected (see Figure 93).
260
220
200
124 N/mm²
180
160
140
120
0 200 400 600 800 1000 1200
t [min]
Figure 85: Detection of material stress load cycles in the calculated RSE
The number and size of all material stress load cycles, which were detected within the assessed
material stress, are shown in Figure 86. There are a lot of stress load cycles in the range of 0 and 40
N/mm². According to the guideline DIN EN 12952-4, these stress load cycles don’t cause any fatigue.
The highest stress load cycle amounts to 124 N/mm² and also doesn't belong to stress load cycles,
which cause fatigue of the material. However, this example shows that the biggest stress load cycle
occurred during the positive load change of the power plant due to the big peak-to-peak values in the
compressive and the thermal stress at the same time.
400
Number of Material Stress Load Cycles
350
300
250
200
150
100
50
0
0 20 40 60 80 100 120 140 160 180 200
Size of Material Stress Load Cycles [N/mm²]
Figure 86: Number and size of all detected material stress load cycles
4.2.4 Control Loop Performance Indicator for the Assessment of the Allowable Peak-
to-Peak Steam Temperature Value during Power Plant Load Changes
Previous example shows that big stress load cycles occur mostly during the power plant load
changes. The bigger the power plant load change the bigger the peak-to-peak value of the
Chapter 4 - Lifetime Consumption of Thick-Walled Components 85
compressive stress and the material stress load cycle. If the peak-to-peak value of the thermal stress
is also big at the same time then the material load cycle will be even bigger and can exceed the limit of
190 N/mm². To avoid this excess the peak-to-peak value of the thermal stress is to be minimized
keeping the peak-to-peak value of the temperature difference in the wall of a thick-walled component
in a defined range. However, this range isn't any general defined constant value, but depends on the
size of the power plant load change, on geometrical dimensions of a thick-walled component and on
physical properties of material. Depending on these factors and considering the fact that the range of
material stress is to be kept under 190 N/mm², it is possible to assess the allowable temperature
difference in the wall of a thick-walled component Δϑw by following formula:
∆σ m≤ 190
σ m1 − σ m 2 ≤ 190
σ m1 − σ m 2 = σ p1 + σ ϑ1 − σ p 2 − σ ϑ 2
d ms β
σ m1 − σ m 2 = α m ( p1 − p 2 ) + α ϑ Lϑ (∆ϑ w1 − ∆ϑ w 2 )
2e ms 1 −ν (4.21)
d ms β
αm ∆p + α ϑ Lϑ ∆ϑ w ≤ 190
2e ms 1 −ν
d ms
190 − α m ∆p
2e ms
∆ϑ w ≤
β
α ϑ Lϑ
1 −ν
Using this formula, the allowable Δϑw was assessed for two exemplary thick walled components steam
collector 1 and steam collector 2. Both components are made of the X10CrMoVNb9-1 steel. Their
geometrical dimensions as well as the physical properties of the X10CrMoVNb9-1 steel, which were
used for the assessment of allowable Δϑw1 and Δϑw2, are given in Table 18.
Table 18: Values used for the assessment of Δϑw1 and Δϑw2
It is assumed in this example that the steam pressure equals 306 bar during full load operation.
According to this value, the peak-to-peak values of the steam pressure Δp were calculated for different
possible power plant load changes ΔP. Table 19 shows the assessed allowable Δϑw1 for steam
collector 1 and allowable Δϑw2 for stem collector 2 during different power plant load changes. It can be
seen that the higher the load change, the lower the values of allowable Δϑw1 and Δϑw2. Besides Δϑw1
differs widely from Δϑw2 due to different geometrical dimensions. Therefore, defining the allowable Δϑw,
the most critical thick-walled component is to be considered.
Chapter 4 - Lifetime Consumption of Thick-Walled Components 86
The peak-to-peak value of the temperature difference in the wall of a thick-walled component Δϑw is
always lower than the peak-to-peak value of the steam temperature (see Figure 80-a and
Figure 80-b). Therefore, the allowable Δϑw, calculated by formula (4.21), can be used as an indicator
for the determination of the allowable steam temperature peak-to-peak value PTPϑ,fatigue during
different power plant load changes. Defining the allowable steam temperature peak-to-peak value
using this method, a big temperature safety margin is considered.
d ms
190 − α m ∆p
2e ms
PTPϑ , fatigue ≤ (4.22)
β
α ϑ Lϑ
1 −ν
5.1.1 Definitions
Benchmarks are used in a very wide field of application. Hence, there is no consistent definition of the
term ‘benchmark’. A very general definition without a specific professional background is given in [44]:
‘Benchmark: A standard of excellence, achievement, etc., against which similar things
must be measured or judged.’
Most definitions describe the notion of benchmark from a business point of view, so does for example
Spendolini in ‘The Benchmark Book’ [45]:
‘A benchmark is a continuous analytical process for comparing the business practices of
companies that are acknowledged as best-in-class for the purpose of organizational
improvement.’
Summarizing these definitions, a benchmark can be described as:
‘comparison of the performance of a given process to some kind of reference value in
order to assess the potential for improvement of the process.’
By performance the control loop performance is meant, process refers to the considered control loop.
The control loop performance is evaluated with the aid of some control loop performance indicator,
see chapter 2.3. The reference value is the benchmark value as described below.
Having established a definition of the notion of ‘benchmark’, some related terms need to be specified
in this context:
The benchmark value is the reference value for comparison. It is usually some kind of best value
(maximum performance) or an upper limit to what is achievable by an existing process.
Benchmark describes the entire event of establishing a reference value and comparing it to the
investigated process. Oftentimes it is also used in the same way as ‘Benchmark Value’, but not in the
context of this project since it may lead to confusion.
Benchmarking is the process of comparing a benchmark value and the performance of the physical
process under consideration.
Chapter 5 - Benchmarks for Control Loop Performance 89
No transfer
2.) Measurement data Simulation data 2.)
This scheme consists of two distinct areas, both of which are constructed in a similar way:
Chapter 5 - Benchmarks for Control Loop Performance 90
On the one hand the ‘real world’ (shown on the left hand side) where the real control loop performance
is evaluated. It is furthermore characterised by the power plant, which is a real process, and the
corresponding controller. These two are subject to limitations. Some of these limitations are of physical
nature. Valve positions for example cannot be changed arbitrarily fast. Moreover they are limited to a
certain range. Other limitations stem from a control point of view, like robustness issues for example.
Measurement data from the power plant will enable the evaluation of the real control loop
performance. This is done via a performance indicator, see section 2.3.
On the other hand there is the ‘benchmark world’. This is where the benchmark value will be derived.
The ‘benchmark world’ is basically an approximation of the ‘real world’. Instead of the power plant and
the implemented controller in the ‘real world’, process models of the power plant and benchmark
controllers are used in the ‘benchmark world’. These benchmark controllers are designed under
different conditions compared to the ‘real world’. As the focus concentrates on a best achievable
performance, the benchmark value, some limitations can be discounted. This is true for robustness
issues, which only matter for implemented controllers. In addition, all system variables can be
considered to be known, implying that there is no need for observer based controller design etc.
These simplifications involve a very important aspect: Due to these optimal conditions, no direct
transfer from the ‘benchmark world’ to the ‘real world’ is feasible. The use of a benchmark controller is
restricted to the determination of a benchmark value for the control loop performance whereas it is not
meant to be implemented in a real power plant. In analogy to the ‘real world’, the benchmark value is
calculated on the basis of simulation data along with a control loop performance indicator.
In a last step, benchmark value and real control loop performance will be compared, representing the
actual benchmarking according to the definition.
The live steam temperature control loop is mostly realised in a cascade structure (see Figure 22). As
the subordinated control loop (follow-up controller and spray water injection) is very fast in comparison
to the superheater control (superheater and master controller), a cascade based control structure is
the first choice. Therefore, the control structure itself shall remain untouched within this benchmark. As
for the follow-up controller, a simple PI controller is used in most cases. It controls the input of the
superheater sufficiently well so that the assumption that ϑI is at its set point permanently is justified.
Hence, there is no important potential to improve the injection, which is why, in the frame of this
benchmark, the focus is entirely on the master controller.
The benchmark is derived via model based optimisation of the controller. Thus a superheater model is
required in the first place. The following subchapters are dedicated to this model, the benchmark
controller and finally the simulation results of the ensemble.
Chapter 5 - Benchmarks for Control Loop Performance 91
It consists of three first order delays in series interconnection with equal time constants, which add up
to the total superheater time constant T100. Splitting up time delay into three equal parts permits the
consideration of the locally distributed heat flux across the heating surfaces. The time constant T100 is
valid at full load operation (100%). Since such systems show slower dynamics at part load operation,
the time constant is adapted automatically to fit the altering conditions. Nonetheless, the model is to
be parameterized with the full load time constant T100. Another first order delay takes into account the
delay imposed by the measurement device. The model is enthalpy based instead of temperature
guaranteeing a linear system. The block diagram of the model with its two inputs hI and the flue gas
heat qF as well as the output hO is shown in Figure 88.
Flue gas heat qF
hI h* hO
Superheater Sensor
The combination of superheater and sensor results in a fourth order model. The states of the system
are the specific enthalpies of the steam, three states within the superheater itself and a fourth state
representing the measurement device. It is solely parameterized by two parameters, the total
superheater time constant T100 and the time constant of the sensor.
The controller gains are calculated by quadratic optimisation, compromising squared control error and
actuator effort. This prevents the controller from being overly aggressive while ensuring very good
control behaviour.
The control structure with the benchmark controller as well as the additional state of the model is
shown in Figure 89.
For further information on state feedback control and associated topics see [4] and [21] and the
references therein.
qF
ϑfeed ϑI ϑO
superheater
xSystem
C1
Follow-up
ϑI,SP xε ϑSP
controller C2
According to the explanations in section 5.3.1, the follow-up controller is realised as a PI-controller.
The simulation model is fed with measurement data of an exemplary power plant. It was operating in
steam generator in control mode while frequency control was activated. Information about
disturbances like soot blowing was not available. These measurement data are (see also Figure 89):
1. Feed steam temperature ϑfeed: The temperature of the steam that leaves the preceding
superheater stage.
2. Flue gas heat: The flue gas heat is not really a measurement value because it is not
measurable. Instead it is calculated via inlet and outlet temperature of the superheater and the
superheater model.
3. Outlet temperature setpoint: The outlet temperature set point is also taken from the
measurement data, in this case it is a constant value.
The simulation results in terms of live steam temperature ϑO (benchmark controller) and live steam
temperature set point ϑO,SP are visualised in Figure 90. The corresponding measurement data (real
plant) is also plotted in order to compare it to the simulation results. The output temperature oscillates
around the constant set point value. Fast oscillations result from a more aggressive controller
(compared to standard PI-techniques), on the other hand the amplitude is remarkably small. Figure 90
shows both measured live steam temperature and the simulated value (referred to as ‘benchmark
controller’).
Chapter 5 - Benchmarks for Control Loop Performance 93
546
545
[°C]
544
543
Set point
542
Benchmark controller
Real plant
541
0 1000 2000 3000 4000 5000
Time [s]
The enthalpy set point value for the subordinated follow-up controller, i.e. for the enthalpy at the
superheater inlet is plotted in Figure 91. The improved control behaviour comes at the expense of
higher control action. This also translates into higher demands regarding the attemperation system,
i.e. spray water injection.
Set point for follow-up controller
510
Benchmark controller
Real plant
500
Temperature [°C]
490
480
470
460
450
0 1000 2000 3000 4000 5000
Time [s]
Figure 91: Control variables (enthalpy set point for superheater inlet)
Best
ISE = 0
performance
ISEBV = 4,70
ISERP = 11,49
Less
performance
It is to notice that not only the control loop performance of the power plant (ISERP) but also the
benchmark value (ISEBV) are heavily dependent on the measurement data, i.e. on the specific plant it
was evaluated for. For this reason, the benchmark value will be different for each power plant,
nonetheless it can serve as a common basis for the comparison of different power plants.
mT
mF mStG pSt
TStG TSto
yT
Unlike proposed by the guideline, the supplied steam mass flow mT is not fed back to the steam
storage. This is due to the fact that in turbine-in-control mode, the power output is controlled by the
turbine inlet valve, which in turn makes the valve position and the resulting steam mass flow mT a
disturbance variable for the steam pressure. Consequently, the turbine inlet valve position yT is not
considered which makes the live steam pressure the actual output signal. These adaptations comply
with the removal of the signal paths indicated in gray colour.
The combination of steam generation and steam storage results in a fourth order state space model.
Three states describe the dynamic behaviour of the steam generation and the fourth state represents
the storage of the steam in the tubes of the entire steam generator. The system variables are
normalized which implies that the model is parameterized solely based on two parameters, namely the
two time constants given in the table below.
255 220
[bar]
[kg/s]
210
245 measurement data measurement data
simulation data simulation data
200
0 5000 10000 15000 0 2000 4000 6000
time [s] time [s]
mT
xSystem
C1
The performance of the live steam pressure control loop is heavily depending on the control variable
mF (fuel mass flow). By allowing for large variations in the fuel mass flow and its gradient, an arbitrarily
Chapter 5 - Benchmarks for Control Loop Performance 97
good control performance can be achieved with this model. Therefore, the dynamic behaviour of the
fuel mass flow has to be limited in some way. Limiting the gradient of the fuel mass flow is obviously a
valid approach since limitations arising from actuators are first and foremost limitations of rate of
change. This is done within the controller design step where the maximum gradients of the simulation
data are limited to the maximum gradients of the measurement data. The results are verified based on
gradient plots.
Supplied steam mass flow mT
675
665
[kg/s]
655
645
0 10000 20000 30000
time [s]
Figure 97 shows the comparison of the live steam pressure of the real plant and the simulation model.
Obviously, the benchmark controller causes faster oscillations around the constant set point but with
reduced amplitude.
Live steam pressure
262
258
[bar]
254
Real plant
Benchmark controller
Set point
250
0 2000 4000 6000
Time [s]
remain upper and lower bounds to the fuel mass flow gradient as well. Thus, the desired limitations
are fulfilled and as a consequence, the data is qualified for the evaluation of the control loop
performance.
Gradient of measurement data
0.04
0.02
0
-0.02
-0.04
0 1000 2000 3000 4000 5000 6000
Step 2: Evaluation: The control loop performance has two be assessed for both measurement data
and simulation data. This is accomplished by applying the chosen control loop performance indicator,
in this case the Integral of Squared Error (ISE), to the data. This was done for approximately 8 hours
of data of an exemplary power plant operating at full load in turbine-in-control mode. The results are
displayed in Figure 99 with the benchmark value ISEBV = 144,8 and the performance of the real
process ISERP = 261,8. For an illustrative reason, the Integral of Squared Error of zero is also
indicated. An ISE of zero implies that the live steam pressure does not deviate from its set point at all
time, which is of course not possible.
Best
ISE0 = 0
performance
ISEBV
ISERP
Less
performance
1. As stated before, the heat flux across the heating surfaces is not measurable. Nevertheless, it
is required for the simulation and thus for the derivation of the benchmark value. The heat flux
is estimated based on measurement data of the steam temperature and a superheater model.
The heat flux signal consists of two parts: The actual heat flux and superposed effects that
arise from the discretisation and sampling time of the measurement data. Since these effects
are not estimable, it is not possible to derive the “correct” heat flux.
2. The second shortcoming relates to the generality of a benchmark: in order to have a valid
benchmark value, it has to be correct for all power plants of a certain class operating in the
same operational mode. This implies independence of the specific measurement data.
Assuming perfect knowledge of the heat flux, some kind of “standard heat flux” could be set
up. This in turn would allow the derivation of a benchmark value that is actually valid for all the
power plants of the class it was designed for.
Summarizing these aspects we can state that the quality of the measurement data is crucial for the
derivation of the benchmark value. Assuming perfect knowledge of the system variables (i.e. “ideal”
measurement data), the benchmark would yield reliable results. Given the fact that this is not the case,
the mathematically derived benchmarks do not qualify for the evaluation of the measurement data.
Nevertheless, this is only a minor loss for the evaluation of the measurement data because best-
practice benchmarks do not face these kinds of problems.
Chapter 6 - Results of Measurement Data Evaluation and Analysis 100
Four hard coal fired power plant units have a nominal power output that is lower than 300 MW. Two of
these units have a drum boiler and are operated in fixed pressure operation. Another two have a once-
through boiler and are operated either in modified sliding pressure operation or natural sliding
pressure operation. The units with once-through boiler are additionally fired with biomass.
Another four hard coal fired power plant units with a nominal power output in the range from 301 to
600 MW have a once-through boiler and are operated either in modified sliding pressure operation or
in natural sliding pressure operation. One of these units is additionally fired with biomass.
Four of five lignite fired power plant units have a nominal power output in the range from 301 to
600 MW and a once-through boiler. Two of these units are operated in natural sliding pressure
operation and another two in fixed-pressure operation.
The other lignite fired power plant units has a nominal power output in the range from 601 to 900 MW.
This unit has a once-through boiler and is operated either in modified sliding pressure operation or
natural sliding pressure operation.
The combined cycle power plant unit has a nominal power output that is lower than 300 MW and a
drum boiler.
primary control or secondary control or both controls are activated. The larger the amount of required
control energy is, the larger are the peak-to-peak values of the power output set point. In some
measurement data sets, the peak-to-peak values of the power output set point were up to 40 % (see
Figure 100). These large peak-to-peak values of the power output set point can take place not only
during full, part or low load operation, but also during load changes. The peak-to-peak values of power
output set point during grid control operation are called ‘grid influence’.
95
90
85
Power Output [%]
80
75 36 %
Actual Value
70 Set Point
65
60
55
0 50 100 150 200
t [min]
Set Point
Actual Value
Figure 100: Exemplary power output measurement data of a power plant unit that provides control energy
during part load operation
Since the control loop performance can vary depending on power plant type, operating conditions,
etc., the results of measurement data evaluation were classified depending on power plant type,
power output, operational mode, operating mode, boiler type, etc. Thus, the classification cases are
defined by different power plant types, operational modes, power output, etc. The classification cases
and the results of the measurement data evaluation for these cases are summarized in Appendix C.
The evaluation results are given in the form of tables with the best performance, the worst
performance and the mean performance of evaluated control loops. Additionally, all known boundary
conditions (e.g. time frame evaluated, number of power plant units evaluated, boiler type, operating
mode, grid influence, etc.) are given for each case. Moreover, it is indicated whether the primary
control and/or the secondary control were activated. Besides, the value of the grid influence (peak-to-
peak value of the power output set point) is given. If the quality of the provided measurement data is
not good, it will be also mentioned.
The results of the measurement data evaluation are described briefly in the following subchapters.
during load changes without grid influence are smaller than during load changes with grid influence.
Moreover, the larger the grid influence during a load change is, the larger is the power output control
deviation.
set point, the reheater steam temperature control is often not in operation. Besides, this lead to an
efficiency loss of a power plant unit.
In some power plant units, different reheater steam control loop performance was determined in
different tracks. For example, large negative mean value deviations of the reheater steam temperature
took place in three of four tracks and a good performance was achieved in the fourth track. Therefore,
the reheater steam temperature control loop performance should be determined for each track
individually.
The meaning of the symbols that are used in formula (7.1) are described in the Table 23.
Symbol Meaning
qcm
IND norm
normalized indicator of an indicator q for control loop c in operational mode m
qcm
IND act
actual value of an indicator q for control loop c in operational mode m
qcm
IND ref,mean
average value of the reference indicator INDref of an indicator q for control loop c in
operational mode m
σ IND
qcm
standard deviation of the reference indicator INDref of an indicator q for control loop c in
ref
operational mode m
temperature control had ideally worked, the best PTPT -value would have been equal to zero. This
PTPT -value represents the best theoretical control loop performance (see Table 24). Applying the z-
scores method to the PTPT -values given in Table 24, the normalized indicator values are defined. The
normalized indicator values represent a scale of achievable live steam temperature control loop
performance for the indicator peak-to peak value PTPT (see Figure 102).
Table 24: Values and normalized values of the distribution shown in Figure 101
Using this scale, it is possible to identify whether an exemplary evaluated PTPT-value represents the
performance, which is better or worse:
• than the best practice performance,
• than the average practice performance or
• than the worst practice performance.
If the normalized indicator of an exemplary PTPT-value amounts to -2.0 as shown in Figure 103, then
the evaluated performance will be very good. It will be even better than the best practice performance.
-2.0
If the normalized indicator of an exemplary PTPT-value amounts to -0.5 as shown in Figure 103, then
the evaluated performance will be better than the average practice performance and worse than the
best practice performance (see Figure 104).
-0.5
are already planned. However, the increased steam pressure and steam temperature require high-
quality materials for the boiler and for the steam turbine as well as new high-tech measurement
instrumentation. All this results in high investment costs and influences the economic efficiency of a
plant.
The power plant control can make a contribution to achieve the best possible power plant efficiency,
keeping important parameters such as live steam temperature, live steam pressure and reheater
steam temperature as good as possible at their set points. The sole impact of the improved control
loop performance on the power plant efficiency can be demonstrated on the basis of an exemplary
lignite-fired power plant unit with the power output of 300 MW, in which a renewal of the I&C system
was performed. This exemplary power plant unit is operated in the operating mode 'Steam generator
in control (Turbine Following)' with initial-pressure operation. The control loop performances that were
achieved in the power plant unit before and after the I&C renewal are shown in Table 25. Before the
renewal of the I&C system the fluctuations of the live steam and of the reheater steam temperature
were in the range of ±5.5 K. The fluctuations of the live steam pressure were kept in the range of ±3
bar by means of turbine valve throttling (see Table 26). After the renewal of the I&C system the steam
temperature fluctuation were reduced up to ±3 K and the live steam pressure fluctuations remained in
the range of ±3 bar without turbine throttling. Due to the improved control loop performance, it was
possible to increase the live steam temperature and the reheater steam temperature set points, to
avoid the throttling losses and to influence positively the efficiency of the power plant unit in this way
(see Table 26).
Control Loops State prior to the renewal of State after the renewal of Improvement of
the I&C system the I&C system Efficiency
LS Temperature +/- 5.5 K +/- 3 K 0.038 %-Points
RHS Temperature +/- 5.5 K +/- 3 K 0.020 %-Points
LS Pressure +/- 3 bar (With throttling) +/- 3 bar (Without throttling) 0.010 %-Points
Table 26: Improvement of control loop performance and its impact on the power plant efficiency
Figure 109 shows a diagram, which is based on the data of an operator and represents the
components causing unplanned unavailability of fossil-fired power plants in 4th quarter 2010.
According to this diagram 24 % of power plant unavailability was caused by boiler tube leaks and
further 19 % by boiler. Further components causing a large part of power plant unavailability are
generator, steam turbine and transformer/switchgear.
One more operator gave the information that 40.5 % of the unplanned unavailability of their power
plants was caused by boiler, without distinction between the boiler and boiler tube leaks (see
Figure 110). Further unavailability was caused by steam turbine, generator and actuators.
Another operator gave the information that the superheater damage leads often to unplanned
unavailability of a plant.
It can be seen that the diagrams shown in Figures 105, 106, 107 give the information about the
components, which cause the power plant unavailability. However, they don't give the information
about the reasons of components outage. For example, such boiler problems as tube leaks are
typically caused by hydrogen absorption, erosion due to impacts from solid ash particles, corrosion
fatigue, overheating, etc. The steam turbine unavailability is typically caused by turbine blades, turbine
bearings, turbine generator vibration, main stop valves or control valves. In order to find out, if the
outage of these components could be prevented by a high control loop performance, more detailed
information is necessary.
HF - Bunker, feeder, pulverizing system 20.47 %
HA - Pressure system, feed water, steam sections 9.16 %
HH - Main firing system 2.83 %
HN - Flue gas exhaust 1.72 %
20.47 % HD - Ash, slag and particulate removal 1.65 %
HL - Combustion air system 1.5 %
28.53 %
HB - Support structure,enclosure, steam generator interior 1.13 %
H - Other components of conventional heat generation 7.65 %
9.16 % MA - Steam turbine 7.16 %
MK - Generator 1.23 %
2.59 % M - Main machine sets (other components) 0.66 %
2.83 %
1.85 % 1.72 % LA - Feed water system 4.19 %
1.62 % 7.65 % 1.65 % LB - Steam system 3.09 %
7.16 % 1.5 % P - Cooling water systems 2.18 %
2.18 %
1.13 % E - Fuel supply and residues disposal 1.62 %
3.09 %
C - Instrumentation and control equipment 1.85 %
4.19 % 1.23 %
U - Structures 0.07 %
0.66 %
A - Grid and distribution systems 2.59 %
No KKS - Other 28.53 %
Figure 108: Components causing unplanned unavailability of fossil-fired power plants in 2009 [41]
Chapter 8 - Control Loop Performance and Economic Efficiency of a Power Plant 111
1.5 %
Boiler 19 %
21 % 19 %
Boiler Tube Leaks 24 %
Generator 10 %
29 %
2% Steam Turbine 9 % Boiler 40.5 %
Balance of Plant 3 % 40.5 % Steam Turbine 29 %
2%
Pollution Control Equipment 1 % Generator 29 %
9% 24 % Transformer/Switchgear 9 % Actuators 1.5 %
Gas Turbine 2 %
Fuel Firing 2 %
1% 10 %
9% Other 21 % 29 %
3%
Figure 109: Components causing unplanned Figure 110: Components causing unplanned
power plant unavailability of operator 1 in 2010 power plant unavailability of operator 2 in 2010
9 Summary
A methodology for definition and verification of the performance of the most important control loops in
fossil-fired power plants was developed in this VGB research project. The most important control loops
in fossil-fired power plants are:
• power output control loop,
• live steam pressure control loop,
• live steam temperature control loop,
• reheater steam temperature control loop,
• enthalpy control loop / drum level control loop.
The methodology can be used to evaluate the control loop performance during power plant operation,
during the commissioning of new control concepts and retrofits, etc.
The methodology is based on meaningful control loop performance indicators, which were defined
within the framework of this research project. These control loop performance indicators were
classified for the above mentioned control loops with respect to the requirements on these control
loops. In order to classify the indicators, the respective control loops were investigated in detail.
Moreover, the influence of their control loop performance on the power plant efficiency, on the lifetime
of thick-walled components and on the economic efficiency of power plant operation was investigated.
These investigations showed that control loop performance indicators have a different meaning for
different control loops as well as varying importance during different operating conditions.
The applicability of each control performance indicator was shown using real power plant
measurement data. For this, measurement data of hard coal fired power plant units, lignite fired power
plant units and combined cycle power plant units were collected. In total, the measurement data of
fourteen European power plant units over a long period of time were available for the investigations.
The defined control loop performance indicators were applied to the available measurement data in
order to evaluate the control loop performance. To make the evaluation results comparable with each
other, the indicators were applied by means of the 'sliding time window'. Due to this application
methodology it was possible to determine the best and the worst achieved control loop performance
as well as its mean for the above mentioned control loops. The results of the measurement data
evaluation were classified depending on:
• power plant type,
• power output of a power plant,
• operational mode,
• operating mode and
• boiler type.
Based on the measurement data evaluation and analysis, some optimization potential was revealed in
some power plant units. A feedback regarding the results of the measurement data evaluation and
analysis was given to the operators of each participating power plant. Besides, it was shown that the
measurement data evaluation by means of different control loop performance indicators provides more
significant and meaningful results than the measurement data evaluation by means of a single control
loop performance indicator.
The best achieved values of the control loop performance can be used as benchmarks for the
comparison of existing power plants, for the evaluation of the control loop performance during the
Chapter 9 - Summary 113
commissioning of new control concepts and retrofits, for the establishment of requirements on the
control loop performance during the tendering phase as well as for the control loop performance
monitoring during power plant operation.
Chapter 10 - References 114
10 References
[1] Blyth, W.; Yang, M.; Bradley, R.:
Climate Policy Uncertainly and Investment Risk. International Energy Agency (IEA).
OECD/IEA, Paris, 2007.
[2] Black, J.; Hashimzade, N.; Myles, Gareth D.:
A dictionary of economics. New York: Oxford University Press, 2009.
ISBN 978-0-19-923705-0
[3] Boyce, Meherwan P.:
Handbook for Cogeneration and Combined Cycle Power Plants. New York: The American
Society of Mechanical Engineers, 2002
[4] Boyd, S.; Barratt, C.:
Linear Controller Design: Limits of Performance. Prentice-Hall, 1991
[5] Bunzenmeier, A.:
Ein praxisorientiertes Inbetriebnahmekonzept für Zustandsregler im Bereich der
Dampftemperaturregelung. VGB Kraftwerkstechnik 76 (1996), Heft 11
[6] Crastan, V.:
Elektrische Energieversorgung 2, Energie- und Elektrizitätswirtschaft, Kraftwerkstechnik,
alternative Stromversorgung, Dynamik, Regelung und Stabilität, Betriebsplanung und -
führung. 2. Aufl. Berlin Heidelberg: Springer Verlag, 2009
[7] Deutsche Energie-Agentur GmbH:
Energiewirtschaftliche Planung für die Netzintegration von Windenergie in Deutschland an
Land und Offshore bis zum Jahr 2020 (DENA Netzstudie), Köln, 2005.
[8] DIN 19226:
DIN 19226 Teil 5: Leittechnik; Regelungstechnik und Steuerungstechnik; Funktionelle
Begriffe
[9] DIN EN 12952-3:
Water-tube boilers and auxiliary installations - Part 3: Design and calculation for pressure
parts; German version prEN 12952-3:2008
[10] DIN EN 12952-4:
Water-tube boilers and auxiliary installations - Part 4: In-service boiler life expectancy
calculations; German version prEN 12952-4:2008
[11] Effenberger H.:
Dampferzeugung. Berlin Heidelberg: Springer Verlag, 2000
[12] Flynn, D.:
Thermal Power Plant Simulation and Control. London, United Kingdom: The Institution of
Electrical Engineering, 2003
[13] Herzog, R.; Kägi, U.:
Betriebserfahrungen mit einem Zustandsregler mit Beobachter an einer Überhitzer-
Temperaturregelung. Regelungstechnische Praxis (rtp) 26. Jahrgang, 1984 Heft 8
[14] Kakac, S.:
Boilers, Evaporators and Condensers. New York, Chichestra, Brisbane, Toronto,
Singapore: John Wiley& Sons, Inc., 1991.
[15] Kehlhofer, R.; Hannemann, F.; Stirnimann, F.; Rukes, B.:
Combined–Cycle Gas & Steam Turbine Power Plants. Tulsa, Oklahoma: PennWell
Corporation, 2009.
Chapter 10 - References 115
Appendix A
Figure 111: Feed water control with enthalpy h at the end of the evaporator as a controlled variable
a) Control structure
b) Diagram of enthalpy set value
1 Controller for enthalpy after evaporator
2 Controller for feed water flow
Appendix A 119
Figure 112: Feed water control with ratio of attemperation water mass flow to feed water mass flow as a
controlled variable
Appendix B 120
Appendix B
During negative load change it is a problem. During steady-set operation it is not a problem,
otherwise the power plant unit is designed wrongly.
much higher than the one at the working point of 30-40%. The small movements of the control
valve at this critical working point are more important than the number of shifts in direction.
How is the actuator wear detected?
The valve as a mechanical component cannot be monitored directly. It can be monitored only
by the drive software concerning how often the valve was used.
The drive monitoring system that was performed in a power plant as a test project had to be
stopped, because theory and practice are too far apart. Theoretical consideration led often to
maintenance problems.
If such well known problems as excessive wear of actuators due to vibrations of the complete
power plant appear often, it is desirable to revise the control loops.
• Examples
Different power plant blocks can hardly be compared or are rather not comparable, because even the
differences in the used coal lead to considerable differences in the results.
The example shows reachable values of the control loop performance, but it doesn't give
precise values for the power plant blocks.
Permanent mean value deviation of 0.7 K in the reheater steam temperature measurement data. Why
does this permanent mean value deviation take place?
The evaluation shows a surprising result that hasn't been seen in such a form before. The
question is, whether the same measurement was used for the control system. Usually, in each
track 2 of 3 measurements are taken and their average value is used as the actual value for
the control system. It is quite possible that single measurements differ from each other. This
difference can be in the range from 2 to 3 K over a longer period of time. However, the entire
track shouldn't have any considerable difference between the reference value and the set
point.
Comments on the Defined Control Loop Performance Indicators
It is necessary to select max. 3 simple indicators, according to which the control loop
performance is to be determined. All not intuitively understandable indicators are not
necessarily desired.
It is quite reasonable to find alternatives to an indicator that takes into account only the fixed
peak-to-peak value. Otherwise a constant steam temperature offset, which is in the range of
the acceptance band, is accepted. Therefore, the mean value deviation is to be considered
necessarily. The international guideline IEC 60045-1 "Steam turbines" contains e.g.
reasonable requirements on the steam temperature control performance, like ranges of
acceptable steam temperature deviations and ranges that may be exceeded only for a few
hours in a year.
The control loop performance of the steam temperature control is much more interesting than the one
of the steam pressure control. A steam pressure variation due to a dynamic steam pressure set point
is for example not informative. Considering the lifetime of thick-walled components, only the steady-
state operation is interesting.
• Enthalpy Control
Limitations of the enthalpy are not desirable. Up to now enthalpy control was used consciously to
influence the feed water mass flow. The performance of the enthalpy control was satisfying as long as
the position of attemperation valves was within the control range.