331 - Final Report PDF

Institute of Combustion and
Power Plant Technology

Director: Prof. Dr. techn. G. Scheffknecht
Pfaffenwaldring 23 • 70569 Stuttgart • Germany
Phone +49 (0) 711-685 63487 • Fax +49 (0) 711-685 63491
Definition and Verification of the Control Loop

Performance for Different Power Plant Types
Dipl.-Ing. Nataliya Knierim-Dietz

Dipl.-Ing. Lutz Hanel
Dipl.-Ing. Joachim Lehner
Research Project Supported by VGB Power Tech
May 2012
Index
1 Introduction .......................................................................................................... 1
1.1 Problem Formulation ........................................................................................ 1
1.2 Project Aims ..................................................................................................... 2
1.3 Structure of the Report ..................................................................................... 3
2 Theoretical Background ....................................................................................... 4
2.1 Considered Process Technology ...................................................................... 4
2.2 Definition of Control Loop Variables ................................................................. 6
2.3 Control Loop Performance Indicators and Application Methodology ................ 6
3 Control Loops Considered.................................................................................. 19
3.1 Live Steam Temperature Control Loop ........................................................... 19
3.2 Reheater Steam Temperature Control Loop................................................... 33
3.3 Power Output Control Loop ............................................................................ 33
3.4 Live Steam Pressure Control Loop ................................................................. 43
3.5 Correlation of Power Output and of Live Steam Pressure Control Loop
Performances ................................................................................................. 50
3.6 Feed Water Control Loop ............................................................................... 53
3.7 Feed Water Control in a Drum Boiler .............................................................. 53
3.8 Feed Water Control in a Once-Through Boiler ............................................... 58
4 Lifetime Consumption of Thick-Walled Components .......................................... 63
4.1 Creep Damage ............................................................................................... 63
4.2 Fatigue Damage ............................................................................................. 74
4.3 Influence of Steam Pressure and Steam Temperature Control on Lifetime

Consumption of Thick-Walled Components.................................................... 86
5 Benchmarks for Control Loop Performance ....................................................... 88
5.1 Introduction to Benchmarks ............................................................................ 88
5.2 The Development Scheme ............................................................................. 89

5.3 Live Steam Temperature Control Loop ........................................................... 90
5.4 Live Steam Pressure Control Loop ................................................................. 94
5.5 Conclusion on Benchmark .............................................................................. 98
6 Results of Measurement Data Evaluation and Analysis ................................... 100
6.1 Overview of Power Plant Units that Provided Measurement Data ................ 100
6.2 Classification of Cases for Measurement Data Evaluation ........................... 100
7 Normalization of Control Loop Performance Indicators .................................... 104
7.1 Normalization Method ................................................................................... 104
7.2 Application of Normalization Method ............................................................ 104
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
8.2 Power Plant Efficiency .................................................................................. 107
8.3 Part Load Capability of a Power Plant .......................................................... 108
8.4 Load Change Capability of a Power Plant .................................................... 109
8.5 Influence of Power Plant Unavailability on Economic Efficiency ................... 109
8.6 Power Output Control Loop Performance ..................................................... 111
9 Summary .......................................................................................................... 112
10 References ....................................................................................................... 114
Appendix A .............................................................................................................. 118
Appendix B .............................................................................................................. 120

List of Symbols
Symbol Meaning Unit
ϑi superheater inlet temperature [°C]

ϑi,sp set point of superheater inlet temperature [°C]
ϑo superheater outlet temperature [°C]
ϑo,sp set point of superheater outlet temperature [°C]
ϑwi inside wall temperature of a thick-walled component [°C]
ϑwm middle wall temperature of a thick-walled component [°C]
D covered regulating distance [%/min]
DF total fatigue damage [%]
DF RSE fatigue damage caused by RSE [%]
dms mean diameter [mm]
di inside diameter [mm]
Dn regulating distance [%]
1
e(t) control deviation
ems mean wall thickness of carrier [mm]
Et Young’s module at design temperature [N/mm²]
I number of time windows
IAE integral of absolute error
IAET integral of absolute error
ISE integral of squared error
ISET integral of squared error
M- negative peak value of a controlled variable
M+ positive peak value of a controlled variable
MVDT mean value deviation
N number of samples [-]
ni k counted number of stress load cycles in the class i, k [-]
Ni k theoretical number of stress load cycles in the class i,k up to the first crack in the [-]
wall of a thick-walled component
O maximum deviation
O- negative overshoot
O+ positive overshoot
p steam pressure [bar]
ΔP Power plant load change [%]
Δp Steam pressure range [bar]
PTP peak-to-peak value
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.
1.1 Problem Formulation

Final technical approvals for instrumentations, controls and automation systems are generally
stipulated by contract for new power plant units under construction as well as for retrofits. However,
requirements regarding the performance of superposed and subordinated control loops in
conventional thermal power plants are not defined systematically. In practice, specifications for the
control performance are inconsistent and mostly based on one single indicator, for example the
maximum desired value deviation. The control performance of a live steam temperature control is
often quantified by the maximum temperature deviation from the set-point.
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.
1.2 Project Aims

The aims of the proposed research project can be formulated in the following main aspects:
a) Definition of control loop performance indicators depending on type and dynamical

behaviour of a control loop and requirements thereon,
b) Verification of applicability and significance of control loop performance indicators on the
basis of real measurements,
c) Development of a classification of achievable control loop performances (best practice
benchmarks) depending on power plant type, operational mode, control concept, actuator
dynamic, etc.
d) Economical evaluation, i.e. relevance of high control loop performances for the economic
efficiency of power plant operation (e.g. efficiency, flexibility, wear).
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).
Figure 1: Application framework
1.3 Structure of the Report

This report is divided into seven chapters. Following the introduction, the second chapter covers the
theoretical background, including the description of the considered process technology, the definition
of the control loop variables, which are used throughout the project and the definition of control loop
performance indicators. The third chapter includes the description of considered control loops and
some examples for the application of the control loop performance indicators to the measurement
data. The influence of the control loop performance on the lifetime consumption of thick-walled
components is described in the fourth chapter. The benchmarks for the control loop performance are
represented in the fifth chapter. The description of the evaluated control loop performance and the
description of the normalization of control loop performance indicators are given in chapters six and
seven. The eighth chapter covers the influence of the control loop performance on the economical
efficiency of a power plant.
Chapter 2 - Theoretical Background 4
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.
2.1 Considered Process Technology

When defining the control loop performance, it is important to consider the boiler type used in the
power plant unit. There are two types of boilers - drum boilers and once-through boilers (see Figure
2). Drum boilers are subdivided into natural circulation boilers and forced circulation boilers.
Once-through boilers are subdivided into Benson boilers and Sulzer boilers. The following
subchapters include the description of some boiler features, which are of importance for the definition
of the control loop performance.
Boiler Types
Drum Boiler Once-Through Boiler
Natural-Circulation Boiler Forced Circulation Boiler Benson Boiler Sulzer Boiler
Figure 2: Types of boiler
2.1.1 Drum Boiler

In a drum boiler, the feed water passes through the economiser to the drum, from which it flows down
the unheated down pipes to the evaporator, where it partially evaporates. After that, the steam-water-
mix flows to the boiler drum. The difference in the density of the steam and the water allows the steam
to rise and to pass through the superheaters. At the same time, the separated water flows again into
the down pipes and passes through the evaporator. This process is called the circulation process of a
boiler [17]. The circulation process in the forced circulation boiler in contrast to the natural circulation
boiler is aided by the feed water pump. This is the main difference between these boiler types.
However, the control system of the natural circulation boiler doesn’t differ from the one of the forced
circulation boiler. For this reason, both boiler types are described together in this chapter.
One of the important drum boiler features in terms of the control is that the steam mass flow generated
by the fuel mass flow is not exactly predictable. At the same time, changes of the fuel mass flow
influence directly the water level in the drum. Depending on the water level in the drum, the feed water
mass flow has to be changed. However, the feed water mass flow in a drum boiler doesn't affect the
steam temperature. Therefore, the steam temperature in drum boilers is controlled only by the
attemperation [17].
During the frequency control operation rapid changes of the power output are necessary. For this, the
boiler must be able to provide a certain amount of steam at a specific point of time. As a drum boiler
can store a lot of steam, the providing of the needed amount of steam is not problematic and will lead
only to small steam pressure changes. Therefore, not very high requirements are usually made on the
steam pressure control [17].
2.1.2 Once-Through Boiler

As mentioned above, once-through boilers are subdivided into Benson boilers and Sulzer boilers. The
main features of these boilers in terms of control are described in the following subchapters.
2.1.2.1 Benson Boiler
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].
2.1.2.2 Sulzer Boiler
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].
2.1.3 Combined Cycle Power Plant

The interest on combined cycle power plants has increased in recent years because of their high
efficiencies, relatively low investment costs and comparatively fast installation [12]. The main
subsystems of a combined cycle power plant are the gas turbine, the heat recovery steam generator
(HRSG) and the steam turbine. Each of these subsystems has its own control system.
The heat used for the boiling of the water and superheating of the steam in the combined cycle power
plants is derived from the exhausting gas of the gas turbine. The exhausting gas is led to the HRSG,
which can be equipped for the supplementary firing. The two types of HRSGs are forced circulation
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.
2.2 Definition of Control Loop Variables

A process control system consists of a process and a control that controls and monitors the process.
The current status of the process can be represented by different variables. The definitions and terms
of variables, which are further used in this report, are shown in Figure 3.
Feedforward
Controller uFF(t) Disturbances z(t)
wT(t) w(t) e(t) Feedback y(t)

Pre-Filter Process
Controller u (t) u(t)
FB
Figure 3: Standard control loop
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).
2.3 Control Loop Performance Indicators and Application Methodology

Control loop performance indicators are measures, which describe how well a control is achieving its
objectives. These indicators enable actual results achieved over time to be compared with planned
results and help to identify problems that can impede the achievement of control objectives. Some
control loop performance indicators and a methodology for their application to the measurement data
are described in the following subchapters. The application of the control loop performance indicators
to the control loops is covered in chapter 3.
2.3.1 Measurement Data Evaluation by means of Moving Time Windows

The length of measurement data sets can strongly influence the results of control loop performance
evaluation. In order to compare the evaluation results of data sets of different length, the data are to
be evaluated by means of a ‘moving time window’. For this aim the time window T is defined and the
indicators are applied to the measurement data over this time window. An example representing this
methodology is demonstrated on the basis of exemplary curves shown in Figure 4 and using the
exemplary control loop performance indicator 'negative overshoot O-T'. This indicator is described in
detail in subchapter 2.3.2.
w(t)
y(t), w(t)
y(t),
y(t) Actual value

w(t) Set point
Figure 4: Exemplary measurement data
a)
w(t)
y(t),w(t)
O-T
y(t),
y(t) Actual value

w(t) Set point
t1 t1+T t
b)
-T
O-T
O
0
t1 t1+T t
Figure 5: Measurement data evaluation by means of moving time windows - step 1
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.
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),
y(t) Actual value

w(t) Set point
t3 t3+T 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%.
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
Figure 8: a) Exemplary measurement data

b) Negative overshoots O-T of exemplary measurement data
c) Percentage distribution of O-T in form of a histogram
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).
PTPT = O+T + O−T (2.3)
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.
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

b) Positive overshoots O+T of exemplary measurement data
c) Percentage distribution of O+T in form of a histogram
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

b) Peak-to-peak values PTPT of exemplary measurement data
c) Percentage distribution of PTPT in form of a histogram
2.3.3 Mean Value Deviation

The mean value deviation from the set point MVDT is calculated by integrating the control deviation
e(t) over the time interval [t, t+T] and dividing the result by T.
t +T t +T
1 1
MVDT = −
T ∫t e(τ )dτ = − T ∫ (w(τ ) − y(τ ))dτ
t
(2.4)
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
Figure 11: a) Negative control deviation of exemplary measurement data

b) Mean value deviations from the set point MVDT
c) Percentage distribution of MVDT in form of a histogram
2.3.4 Integral of Absolute Error

The Integral of Absolute Error (IAE) is the area between the curve of the controlled variable y(t) and
the reference variable w(t). IAE is calculated by integrating the absolute value of the control deviation
from zero to infinity [29]. Due to the integration of the absolute value of the control deviation, both
positive and negative control deviations contribute to the IAE. Moreover, small and large control
deviations are equally weighted.
The Integral of Absolute Error IAET is calculated by integrating the absolute value of the control
deviation over a time interval [t, t+T] and dividing the result by the time span T.
1 t +T 1 t +T
IAE T = ∫ e(τ )dτ = ∫ w (τ ) − y (τ )dτ (2.5)
T t T t
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
Figure 12: a) Absolute control deviation of exemplary measurement data

b) Values of integral of absolute error IAET
c) Percentage distribution of IAET in form of a histogram
2.3.5 Integral of Squared Error

The Integral of Squared Error (ISE) is calculated by integrating the squared control deviation from zero
to infinity [29]. Due to the squaring down of the control deviation before the integration, large control
deviations are stronger penalized than small ones (see Figure 13).
The Integral of Squared Error ISET is calculated by integrating the squared control deviation over a
time interval [t, t+T] and dividing the result by the time span T.
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.
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
Figure 13: a) Squared control deviation of exemplary measurement data

b) Values of integral of squared error ISET
c) Percentage distribution of ISET in form of a histogram
2.3.6 Standard Deviation

The standard deviation STDT is a statistic indicator, which shows how tightly the measurement data
are clustered around the mean value of a signal in the time frame [t, t+T].
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,
• 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.
-3STDT -2STDT -STDT eT STDT 2STDT 3STDT
Figure 14: Standard deviation of normal distributed signals [23]
2.3.7 Distribution of Deviations

This indicator shows the percentage distribution PDkT of e(t)-samples in the time interval [t, t+T] within
defined ranges. At the same time PDkT corresponds to the percentage duration of deviations within
defined ranges. The number of ranges k is variable. The percentage distribution PDkT is calculated by
the following formula:
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
Distribution of PDT [%]


100 100 100
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 [%]


100 100
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: Ranges and their percentage distribution PDkT
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.
2.3.8 Covered Regulating Distance

The covered regulating distance D is the distance covered by an actuator within the considered time
frame T (see Figure 17). The calculation of the covered regulating distance D is described by following
formula:
N N
∑ 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:
x(t)
N-1
N
x(tn)
Dn
x(tn-Δt)
tn-Δt tn t
T
Figure 17: Actuator position
2.3.9 Number of Shifts in Direction

In order to calculate the number of shifts in direction of an actuator x(t), it is necessary to determine
the slope of the curve representing the position of an actuator.
The slope of the curve x'(tn) at the point of time tn is calculated by dividing the rise of actuator position
between two neighbouring points of time tn+Δt and tn by the sampling time Δt:
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:
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
_ _
_
Figure 19: Calculation of NSD within two steps

2.3.10 Summary of Control Loop Performance Indicators

The control loop performance indicators, which are described in detail in previous chapters, and their
features, are summarized in Table 1.
Control Loop Performance Indicator Features
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
Covered Regulating Distance D Indicates the distance covered by an actuator
Number of Shifts in Direction NSD Indicates the number of shifts in direction of an actuator
Table 1: Summary of control loop performance indicators and their features

Chapter 3 - Control Loops Considered 19
3 Control Loops Considered

Chapter 3 includes the description of the considered control loops and examples for the practical
application of control loop performance indicators.
3.1 Live Steam Temperature Control Loop
3.1.1 Process Description

The steam temperature control loop is described in the guideline VDI/VDE 3503 ‘Steam Temperature
Control in Fossil Fired Steam Power Stations’ [37]. This guideline includes a block diagram of the
control system, as shown in Figure 20.
Figure 20: Block diagram of the steam temperature control system [37]
This control system is affected by the following input variables [37]:

• Heating. The heat flux across the heating surface is a disturbance variable. In some setups it
can be manipulated by flue-gas side control interventions (flue-gas reflux, swivel burners). Its
impact on the controlled variable is modelled as a first order delay.
• Steam flow. The steam flow is another disturbance variable. Its impact on the controlled
variable is modelled as a first order delay too. In contrast to the heat flux, it is easily
measurable and can thus be used as a disturbance variable feed forward to improve the
control performance. It is generally not available as a manipulated variable.
• Inlet temperature. The temperature of the steam entering the control system is another
disturbance variable.
The transfer function of a superheater is a linear delay of high order (see Figure 21). Indication of a
time constant and the order of the control system is then sufficient to describe the control system [37].
Alternatively, the transfer function can also be characterized using the equivalent dead time Tu and the
balancing time Tg. The ratio Tu/Tg has a fixed relationship with the order of the transfer function. The
order of the control system remains constant, both during constant pressure and varying pressure
operation, while the equivalent dead time and balancing time increase inversely proportional to the
steam flow. High pressure superheaters with good control behaviour have delay times of
approximately 50 s and values for Tu/Tg of approximately 0.3 at full load [37].
Figure 21: Control transfer function of a superheater [37]
3.1.2 Control Structure

The superheater control problem is conventionally considered as a SISO control problem. A cascaded
control structure is mostly used for the control of the live steam temperature (see Figure 22) [17].
The outlet temperature ϑO is the controlled variable of this control loop. The master controller is a
superposed control loop, which controls the superheater outlet temperature ϑO and uses the
superheater inlet temperature set point ϑI,SP as a manipulated variable. The master controller must be
able to deal with the disturbances of the flue gas heat and the steam mass flow variability.
Disturbances of the flue gas heat
Steam
mass flow ϑI ϑO
Superheater
Follow-up ϑI, SP Master ϑO, SP

Control valve
controller controller
Attemperation PG Generator
spray water I Inlet
O Outlet
SP Set Point
Figure 22: Cascaded steam temperature control structure
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.
3.1.3 Information Available in Guidelines

The guideline VDI/VDE 3503 contains only rudimentary information about the control loop
performance and control loop performance indicators. According to the guideline ‘the most important
indicator for the assessment of the control loop performance is the overshoot of the controlled variable
in case of disturbances’. Besides, in the guideline VDI/VDE 3503 the information is given that, the
overshoot depends on the following quantities:
• value of disturbance, like heating fluctuations and changes in steam flow,
• balancing time TZ of the disturbing step function,
• equivalent dead time of disturbance step response,
• equivalent dead time TU of the control transfer function, balancing time Tg of the control
transfer function and the ratio TU/Tg,
• controller settings.
In the upcoming guideline VDI/VDE 3507 for technical approvals of steam generator control it is
additionally distinguished between the positive and negative overshoot.
3.1.4 Control Aims

The steam temperature control is one of the most discussed control problems in power plants. The
reasons for the extensive attention are mainly found in issues such as [12]:
• Plant lifetime. The steam temperature control has an essential influence on the variation of
the steam temperature and correspondingly on the lifetime consumption of thick-walled
components.
• Efficiency. Limiting steam temperature variations to only small deviations from the set point
allows for a higher live steam temperature set point and accordingly higher unit efficiency.
Increasing the live steam temperature by 10 K will increase the efficiency of a 400 MWe unit by
approximately 0.25 % [12].
• Load-following capability. In order to perform high load gradients a good steam temperature
control is required.
• Availability. A good steam temperature control reduces the probability of a forced power plant
outage due to the excess of security limits.
3.1.5 Practical Application of Control Loop Performance Indicators

A set of applicable indicators has to be selected for the evaluation of the live steam temperature
control loop performance with reference to the following points:
• efficiency of a power plant unit,
• lifetime consumption of thick-walled components,
• capability to compensate disturbances and
• actuator wear.
Examples for the application of control loop performance indicators to exemplary live steam
temperature measurement data are demonstrated in the following subchapters. The exemplary live
steam temperature measurement data are shown in Figure 23-a. The live steam temperature set point
is represented in black colour and the live steam temperature actual value is represented in orange
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.
a) LS Temperature 1 [°C] 560

Set Point
Actual Value
550
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.
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)
Distribution of PTPT [%]

100 100
Distribution of O-T [%]
Discretization 0.5 K Discretization 0.5 K

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-T [K] PTPT [K]
Figure 24: a) Distribution of O+T-values

b) Distribution of O-T-values
c) Distribution of PTPT-values
3.1.5.2 Distribution of Deviations

The results of the application of the indicator PDkT to the exemplary measurement data are shown in
Figure 24. For the application of the PDkT to the exemplary measurement data five steam temperature
ranges k are defined:
• range [-3; 3],
• range [-5;-3)&(3;5],
• range [-7;-5)&(5;7],
• range [-9;-7)&(7;9] and
• range [-∞;-9)&(9;+∞].
Using the indicator PDKT the percentage distribution of steam temperature samples in the defined
ranges is determined. It can be seen that the percentage distribution of the live steam temperature
samples in the range [-3;3] is from 75 to 100 %. If the percentage distribution in this range amounts to
100 %, then there are no live steam temperature samples in other ranges. If the percentage
distribution in this range amounts to 75 %, then 25 % of all live steam temperature samples in a time
window are distributed in other four ranges.
The percentage distribution of the live steam temperature samples in the range [-5;-3)&(3;5] is from 0
to 20 %. However, the percentage distribution from 0 to 5 % was determined in 62 % of all considered
time windows.
The percentage distribution of the live steam temperature samples in further three ranges is from 0 to
5 % in the most of the evaluated time windows. This means, that the most steam temperature samples
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 [%]
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
[% ]
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
[% ]
Figure 25: Distribution of PDkT-values
3.1.5.3 Integral of Absolute Error

IAET is a control loop performance indicator, in which positive and negative temperature deviations are
considered. The results of the application of the indicator IAET to the exemplary live steam
temperature measurement data are shown in Figure 26-b. It can be seen that, that the absolute values
of the live steam temperature control deviation are concentrated in the range from 0.5 to 2.5 K.
3.1.5.4 Integral of Squared Error

ISET is a control loop performance indicator that penalizes heavily all large live steam temperature
deviations. The results of the application of the indicator ISET to the exemplary live steam temperature
measurement data are shown in Figure 26-c. It can be seen that, the ISET-values are concentrated in
the range from 1 to 4 K² in about 80 % of all evaluated time windows. It means that there are no large
steam temperature deviations in these time windows. In 14 % of all evaluated time windows, the
values of ISET are larger and concentrated in the range from 12 to 13 K². It can be concluded that
there are some large steam temperature deviations that occurred shortly.
3.1.5.5 Mean Value Deviation

In order to evaluate the mean value deviation of the live steam temperature from its set point the
control loop performance indicator MVDT is used. Using MVDT the influence of the steam temperature
on the efficiency of a power plant unit and on the lifetime consumption of thick-walled components due
to creep damage can be assessed (see chapter 4.1.4 and chapter 4.1.5.1). The results of the
application of MVDT to the exemplary live steam temperature measurement data are shown in
Figure 26-a. The MVDT-values are concentrated in the range from -0.75 to -0.25 K in 83 % of all
evaluated time windows (see Figure 26-a). Referring to [25], the evaluated live steam temperature
curve doesn’t cause any additional lifetime consumption of thick-walled components due to creep
damage and leads to a slight efficiency loss.
a) Distribution of MVD [%]

T 100
80
60
40
20
0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
MVD [K]
T
b) c)
[%]
Distribution of ISE T [%]

100 100
Discretization 0.5 K Discretization 0.5 K²
T
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
Figure 26: a) Distribution of MVDT-values

b) Distribution of IAET-values
c) Distribution of ISET-values
3.1.5.6 Covered Regulating Distance and Number of Shifts in Direction
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.
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
These reasons will be contemplated in the following subchapters.

3.1.5.6.1 Dimensions and Characteristic Diagrams of Attemperator Control Valves

In order to check if the dimensions or characteristic diagrams of attemperator control valves are
different from each other, the documentation of a power plant unit is to be taken into account.
3.1.5.6.2 Asymmetry of Fire in a Boiler

In case of asymmetry of fire in a boiler, the steam in parallel tracks will be heated differently. In order
to keep the steam temperature at its set point, more water must be attemperated into the more heated
track. As for this aim, the attemperator control valve from the more heated track opens more than the
attemperator control valve from the lower heated track. Accordingly, the position mean values of these
control valves differ strongly from each other. Investigations revealed that, in case of asymmetry of fire
in a boiler these mean values will be strongly different at least in one of attemperation stages.
Therefore, if an assumption of fire asymmetry exists, the mean values of attemperator control valve
positions should be additionally considered.
3.1.5.6.3 Suitability of Follow-Up Controller Settings

According to the guideline VDI/VDE 3503, the fluctuations of the inflow steam temperature, of the
steam mass flow and of the flue gas heat, belong to the disturbances of the superheater. These
disturbances have negative influence on the superheater outlet temperature ϑO (see Figure 29). In
order to keep the outlet temperature ϑO at its set point ϑO,SP, the follow-up controller and the master
controller have to deal with these disturbances (see chapter 3.1.2). At the same time, the action of the
follow-up controller is to be suitable and should lead to a correct control valve reaction.
In order to check the suitability of the follow-up controller action, the control loop performance indicator
number of shifts in direction NSD will be applied not only to the measurement data of attemperator
control valve position, but also to the measurement data of the inflow steam temperature, superheater
inlet temperature ϑI, superheater inlet temperature set point ϑI,SP and control deviation in the input of
follow-up controller (see Figure 29).
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).
NSDϑI
NSDINFLOW
NSDCV
NSDCD NSDϑI,SP
Figure 30: Consideration of NSDCV and of NSDϑbeforeATT
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
Track 1
Track 2
Figure 31: NSD of attemperator control valve position and of steam temperature before attemperation
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.
a) b)

4 4
3 3
2 2
1 1
0 0
c) d)

4 4
3 3
2 2
1 1
0 0
ATT 1 ATT2 ATT3 ATT1 ATT2 ATT3
Track 1
Track 2
Figure 33: NSDCV, NSDCD, NSDϑI and NSDϑI,SP.
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
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).
Live Steam Temperature Control

Importance
Control Loop Performance Indicator Relevance Full Load Part Load Low Load Load
Operation Operation Operation Change
Mean value deviation MVDT Efficiency, creep damage ●●●●● ●●●● ●●● ●●●●
Distribution of deviations PDk Creep damage ●●●●● ●●●● ●●● ●●●●
Positive overshoot O+T
Negative overshoot O-T
Peak-to-peak value PTPT General CLPIs ●●●● ●●● ●● ●●●
Integral of absolute error IAET
Integral of squared error ISET
Peak-to-peak value PTP ≤ Allowable steam ● ● ● ●●●●
temperature peak-to-peak value PTPϑ,fatigue Fatigue damage
Importance Actuators
●●●●● Very high
●●●● High
●●● 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.
Live Steam Temperature Control

Importance
Control Loop Performance Indicator Relevance Full Load Part Load Low Load
LC1 LC2
Operation Operation Operation
Mean value deviation MVDT Efficiency, creep damage ●●●●● ●●●● ●●●●
Distribution of deviations PDkT Creep damage ●●●●● ●●●● ●●●●
Negative overshoot O-T
Peak-to-peak value PTPT General CLPIs ●●●● ●●● ●●●
Peak-to-peak value PTP ≤ Allowable steam ● ● ●●●●
temperature peak-to-peak value PTPϑ,fatigue Fatigue damage
Importance Abbrev. Meaning

●●●●● Very high LC1 Load change from full load operation to part load operation and vice versa
●●●● High Load change from full load operation to low load operation and vice versa
LC2
●●●
as well as
Middle Load change from part load operation to low load operation and vice versa
●● Low
● Very low
Actuators
Wear of actuators
Number of shifts in direction NSD
Table 3: Indicators for the evaluation of the steam temperature control loop performance, pSt set = const.
3.2 Reheater Steam Temperature Control Loop

On the reheater steam temperature control loop performance the same requirements are made as on
the live steam temperature control loop performance. Therefore, the indicators that are selected for
the evaluation of the live steam temperature control loop performance can be used for the evaluation
of the reheater steam temperature control loop performance (see chapter 3.1).
3.3 Power Output Control Loop

The unit control is the superordinated control system of a power plant unit, which includes two control
loops: the power output control loop and the live steam pressure control loop. Technical description of
the unit control, control aims of the power output control and practical application of control loop
performance indicators to power output measurement data are described in the following subchapters.

The unit control is described in detail in the guideline VDI/VDE 3508. This guideline includes a
simplified block diagram of the controlled system as shown in Figure 34. The controlled variables of
this control system are the generator output PG and the live steam pressure pSt. The manipulated
variables of this control system are the turbine valve opening yT and the fuel mass flow ṁF. The
assignment of the manipulated variables to corresponding controlled variables depends on the
operating mode used in a power plant unit (see chapter 3.3.2).
Referring to [40], the dynamic behaviour of a power plant unit can be divided into three subprocesses:
• 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.
Fuel supply and Steam storage Steam expansion and

steam generation generation of electrical energy
Figure 34: Representation of the dynamic relationship in a steam power plant unit [40]
The main disturbances of the unit control are:

• the turbine mass flow ṁT and
• fuel disturbances, e.g. fluctuations of the fuel calorific value or interrupted coal mass flow due to
sticking of the coal in a coal bunker.

The structure of the power output control depends on the operating mode. The operating modes differ
from each other in terms of the assignment of the manipulated variables (turbine valve opening yT and
fuel mass flow ṁF) to the controlled variable (generator output PG) (see Table 4).
Operating mode Controlled

PG
In control Steam pressure variable
Turbine Constant set point (fixed-pressure operation) yT

Set point dependent on output (modified-sliding yT
Turbine
pressure operation) Manipulated
Steam generator Constant set point (initial-pressure operation) ṁF variable
Steam generator Uncontrolled (natural-sliding pressure operation) ṁF
Table 4: Assignment of the manipulated variables yT and ṁF to the controlled variable PG
3.3.2.1 Turbine in Control (Steam Generator Following)
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].
3.3.2.1.1 Fixed-Pressure Operation
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]
3.3.2.1.2 Modified Sliding-Pressure Operation

During the modified sliding-pressure operation, the power output PG is controlled by the turbine valve
opening yT. The difference to the fixed pressure operation is that, the position of the turbine control
valve yT is not controlled as a function of the power output, but remains unchanged within the sliding-
pressure range (see Figure 36-a). It is to be mentioned that the turbine control valve should be opened
at a point of minimal throttling. In case of demand for the immediate-reserve capacity, the turbine
control valve opens and thus increases the steam mass flow to the turbine. If a constant immediate
reserve capacity is required over the entire sliding-pressure range, the characteristic as shown in
Figure 36-b will be chosen [40].
a) b)
Figure 36: Steady-state characteristic of the modified sliding-pressure operation [40]
3.3.2.2 Steam Generator in Control (Turbine Following)

During the initial pressure operation, the power output PG is controlled by the fuel mass flow ṁF. The
generator output set point PG set is the output signal of the target output limiter, like in the operating
mode ‘Turbine in Control' (see chapter 3.3.2.1). Depending on the magnitude of the power output
control deviation, the power output controller determines the set point for the thermal output of the
 Fset. In order to enhance the dynamic of the steam generator, the set point for the thermal
fuel Q
 Fset is pre-controlled by the fuel feed forward controller [40]. The fuel mass flow
output of the fuel Q
controller is intended to ensure that the steam generator is supplied with as much thermal output as it
is needed for the generator output PG in order to reach and to maintain its set point PG set [40].
3.3.2.2.1 Initial-Pressure Operation

During the initial pressure operation, the generator output control including the fuel forward control has
long time delays due to slow response of fuel supply, heat release and steam generation. However,
this operating mode is very robust from the technical point of view [40].
3.3.2.2.2 Natural Sliding-Pressure Operation

During the natural sliding-pressure operation changes of power output are carried out by changes of
the fuel mass flow. The power output changes take place with a large time delay due to slow response
of fuel supply. In order to diminish the disadvantages of the delayed control characteristic of natural
sliding-pressure operation, modified sliding-pressure operation is used (see chapter 3.3.2.1.2).
However, the natural-sliding-pressure operation is very economical, because there is no throttling of
turbine valves [40].

In the guideline VDI/VDE 3508 the performance of the unit control is determined by:
• ‘the time characteristic of power output, process steam output and district heat output
compared with their set points for interconnected-network operation and
• by the time characteristic of frequency, process steam pressure and heating water
temperature for island network operation.’
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.
Ramp-like output change without

Activation of the immediate-reserve
utilisation of the immediate-reserve
capacity (primary control)
capacity (secondary control)
Type of fuel Oil/gas Coal Oil/gas Coal
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]
3.3.4 Control Aims

In order to meet the demands of secure electricity supply, the energy suppliers are obligated to comply
with the existing load requirements at any time. The aim of the power output control is to maintain a
balance between the generation and consumption of the electricity. The frequency of the electrical
power grid is the characteristic state variable, which provides the information on the balance between
the energy generated and the one consumed. The frequency is to be kept stable at 50 Hz. Deviations
of the frequency from its set point can be caused by a higher electricity consumption in comparison to
electricity generation, by a power plant failure, by forecast changes, etc. The power plant units, that
offer grid control, have to increase or to decrease their load in order to stabilise the frequency. The
rules of load frequency control and requirements on power plant performance are given in the
TransmissionCode 2007 [35]. In general three kinds of frequency control are distinguished:
• Primary control. All power plants have to be capable of delivering a maximum primary control
reserve of 2% of the rated power within 30 seconds. The maximum reserve will be activated, if
the frequency deviation amounts to 200 mHz. The maximum reserve is to be maintained over
a period of 15 minutes.
• Secondary control restores primary control reserves and maintains a balance between
generation and consumption of electricity within each control area within a period ranging from
a few seconds to 15 minutes. The corresponding load deviations are to be compensated in the
control area within this timeframe. Secondary control is accomplished by increasing the fuel
input of a power plant unit. Therefore, the dynamic behaviour of power plant units is subject to
high requirements.
• Tertiary control reserve is required to restore the secondary control reserves. The tertiary
control reserve is usually activated manually after the activation of the secondary control.
2
These values are valid for power plant units with once-through boiler (circulation steam generators
are considerably slower during the sliding-pressure operation)
Influence of the power output control on the economic efficiency of the power plant unit is described in
chapter 8.

A set of applicable indicators has to be selected for the evaluation of the power output control loop
performance with reference to the following points:
• economical efficiency of a power plant unit,
• capability to compensate disturbances,
• ability to meet frequency control requirements.
Examples for the application of control loop performance indicators to exemplary power output
measurement data are demonstrated in the following subchapters. In these examples two exemplary
measurement data sets are used. The first data set includes the power output measurement data
during the steady-state operation (see Figure 37-a). The power output set point is constant. The
corresponding negative control deviation is shown in Figure 37-b.
In the second data set the power output measurement data under the influence of the secondary
control are considered (see Figure 38-a). The power output set point is not constant. The
corresponding negative control deviation is shown in Figure 38-b.
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
a) 110
Power Output [%]

100
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 [% ]
b) 100 c) 100
Distribution of O-T [%]
Discretization 0.5 % Discretization 0.5 %

80 80
60 60
40 40
20 20
0 0
0 1 2 3 4 5 6 7 8 9 10 0 2 4 6 8 10 12 14 16
O [% ] PTPT [% ]
-T
Figure 39: a) Distribution of O+T-values for the data set 1

b) Distribution of O-T-values for the data set 1
c) Distribution of PTPT-values for the data set 1
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
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 %.

IAET is a control loop performance indicator, in which positive and negative power output deviations
are considered. The results of the application of the indicator IAET to the exemplary power output
measurement data are shown in Figure 40-b and in Figure 41-b.
Figure 40-b shows that the absolute values of the power output control deviation in the first data set
are concentrated in the range up to 1 %. The absolute values of the power output control deviation in
the second data set are concentrated in the range from 1.5 to 3 %.

ISET is a control loop performance indicator that penalizes heavily all large power output deviations.
The results of the application of the indicator ISET to the exemplary power output measurement data
are shown in Figure 40-d and in Figure 41-d.
Figure 40-d shows that the ISET-values are lower than 1 %² in all evaluated time windows. This
signalizes that there are no large power output deviations in the first data set.
In Figure 41-d there are two clusters of ISET-values: one in the range from 3 to 4.5 %² and another in
the range from 7 to 12 %². The percentage distribution of the second cluster is bigger than the one of
the first cluster. This means that, the number of time windows with relatively large power output
deviations from the set point is higher than the number of time windows with low power output
deviations in the second data set.

The control loop performance indicator MVDT provides information about the influence of power output
control on the economical efficiency of a power plant unit. The evaluated MVDT–values for both data
sets are shown in Figure 40-a and in Figure 41-a.
Power output mean value deviations of the first data set are concentrated in the range from -0.25 to
0.25 % in 50 % of all evaluated time windows (see Figure 40-a). Further power output mean value
deviations are in the range from -0.75 to -0.25 % and in the range from 0.25 % to 0.75 % (see
Figure 40-a).
Power output mean value deviations of the second data set are concentrated in the range from -1.75
to -0.25 % in 93 % of all evaluated time windows (see Figure 40-a). However, there are some
MVDT-values in the range from -2.25 to -1.75 % and in the range from -0.25 to 0.25 % (see
Figure 40-a).
a) c)
Distribution of MVD T [%]
Distribution of STD T [%]

100 100
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
MVD [% ] STDT [% ]
T
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]
Figure 40: a) Distribution of MVDT-values for the data set 1

b) Distribution of IAET-values for the data set 1
c) Distribution of STDT-values for the data set 1
d) Distribution of ISET-values for the data set 1
a) c)
Distribution of STD T [%]
100 100
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)
100 100
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]

c) Distribution of STDT-values for the data set 2
d) Distribution of ISET-values for the data set 2
3.3.5.5 Standard Deviation

In order to determine the deviation of the measurement data from the mean value, the control loop
performance indicator standard deviation STDT is applied to the power output measurement data of
both data sets. The evaluated STDT-values are shown in Figure 40-c and in Figure 41-c.
In Figure 40-c the determined STDT-values of the first data set are concentrated in the range up to
0.5 % in 89 % of all evaluated time windows. In further 11 % of all time windows, the determined
STDT-values are in the range from 0.5 to 1 %.
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.
Power Output Control
Control Loop Performance Indicator Relevance
Mean value deviation MVDT Economic Efficiency

Negative overshoot O- T
Peak-to-peak value PTPT
Integral of absolute error IAET General CLPI
Standard deviation STDT
Table 6: Indicators for the evaluation of the power output control loop performance
3.4 Live Steam Pressure Control Loop

The technical description and the control aims of the live steam pressure control as well as the
practical application of control loop performance indicators to live steam pressure measurement data
are described in the following subchapters.

See chapter 3.3.1.

The structure of the live steam pressure control in fossil-fired power plants depends on the operating
mode of the power plant unit.
Operating mode Controlled

In control Steam pressure pSt variable
Turbine Constant set point (fixed-pressure operation) ṁF

Set point dependent on output (modified-sliding
Turbine ṁF
pressure operation) Manipulated
Steam generator Constant set point (initial-pressure operation) yT variable
Steam generator Uncontrolled (natural-sliding pressure operation) yT

Table 7: Assignment of the manipulated variables yT and ṁF to the controlled variables pSt
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).
3.4.2.1 Turbine in Control (Steam Generator Following)
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.
3.4.2.1.1 Fixed-Pressure Operation

During the fixed-pressure operation, the live steam pressure pSt is controlled by the fuel mass flow ṁF
and the live steam pressure set point pSt set is constant. Depending on the live steam pressure control
 Fset. The
deviation, the pressure controller determines the set point for the thermal output of the fuel Q
 Fset is additionally pre-controlled by the fuel feed forward controller in order
thermal output of the fuel Q
to enhance the dynamic of the steam generator. However, the control step response of fuel supply,
heat release and steam generation is very slow. Besides, external and internal disturbances influence
negatively the live steam pressure actual value. In order to keep the live steam pressure actual value
at its set point, the parameters of the live steam pressure controller are to be adjusted to the response
of the controlled system very precisely. This is difficult to realize due to the varying time response of
mills.
3.4.2.1.2 Modified Sliding-Pressure Operation

The difference between the fixed-pressure operation and the modified sliding-pressure operation is the
fact that, the live steam pressure set point pSt set is not constant and changes depending on the power
output PG. Apart from that, the operating behaviour during the modified sliding-pressure operation is
comparable to that during the fixed-pressure operation.
3.4.2.2 Steam Generator in Control (Turbine Following)
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.
3.4.2.2.1 Initial-Pressure Operation

During the initial-pressure operation, the live steam pressure pSt is controlled by the turbine valve
opening yT and the live steam pressure set point pSt set is constant (see Figure 35). The live steam
pressure control deviation is fed to the pressure controller, which changes the turbine valve opening yT
and adjusts the turbine steam mass flow ṁT to the steam mass flow generated ṁStG. Using this
operating mode the live steam pressure is regulated very precisely, due to the fast step response of
the turbine valve.
3.4.2.2.2 Natural Sliding-Pressure Operation

In natural sliding-pressure operation, the turbine valve is completely opened (see Figure 42).
Therefore the live steam pressure is not regulated and adjusts proportionally to the generated steam
mass flow ṁStG and to the turbine steam mass flow ṁT. This means that, the live steam pressure
varies as a function of the power output during this operating mode [40].
Figure 42: Steady-state characteristic of the natural sliding-pressure operation [40]

See chapter 3.3.3.
3.4.4 Control Aims

The steam pressure control is one of the most important controls in power plants due to the following
reasons:
• 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.

A set of applicable indicators for the evaluation of the live steam pressure control loop performance
has to be selected referring to the following points:
• efficiency of a power plant unit,
• lifetime consumption of thick-walled components,
• capability to compensate disturbances.
Examples for the application of control loop performance indicators to exemplary live steam pressure
measurement data are demonstrated in the following subchapters. In these examples two exemplary
measurement data sets are used. In the first data set the live steam pressure measurement data with
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
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
+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 [% ]
Distribution of PTP [%]

b) 100 c) 100
Distribution of O [%]
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) Distribution of O+T-values for the data set 1

b) Distribution of O-T-values for the data set 1
c) Distribution of PTPT-values for the data set 1
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).

IAET is a control loop performance indicator, in which positive and negative live steam pressure
deviations are considered. The results of the application of the indicator IAET to the exemplary live
steam pressure measurement data are shown in Figure 47-b and in Figure 48-b.
Figure 40-b shows that the most absolute values of the live steam pressure control deviation in the
first data set are concentrated in the range up to 1 %. The most absolute values of the live steam
pressure control deviation in the second data set are concentrated in the range from 0.5 to 1.5 % (see
Figure 48-b).

ISET is a control loop performance indicator that penalizes heavily all large live steam pressure
deviations. The results of the application of the indicator ISET to the exemplary live steam pressure
measurement data are shown in Figure 47-c and in Figure 48-c.
Figure 47-c shows that the most of ISET-values are lower than 1 %². This signalizes that there are no
large power output deviations in the first data set.
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.

The control loop performance indicator MVDT provides the information about the influence of the live
steam pressure on the efficiency of a power plant unit and on the lifetime consumption of thick-walled
components. The evaluated MVDT-values for both data sets are shown in Figure 47-a and in
Figure 48-a.
Live steam pressure mean value deviations of the first data set are concentrated in the range from -
0.25 to 1.25 % in all evaluated time windows (see Figure 47-a). This means that the evaluated live
steam pressure curve has a positive influence on the power plant efficiency and a slightly negative
impact on the lifetime consumption due to creep damage.
Live steam pressure mean value deviations of the second data set are concentrated in the range from
-0.75 to 1.25 % in 88 % of all evaluated time windows (see Figure 48-a). There are some
MVDT-values in the range from -1.25 to -1.75 % (see Figure 48-a).
a) 100
80
60
40
20
0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
MVDT [% ]
b) c)
100 100
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

c) Distribution of ISET-values for the data set 1
a)

100
80
60
40
20
0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
MVD [% ]
T
b) c)

100 100
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

c) Distribution of ISET-values for the data set 2
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.
Live Steam Pressure Control

Importance
Control Loop Performance Indicator Relevance Full Load Part Load Low Load Load
Efficiency,
Mean value deviation MVDT ●●●●● ●●●● ●●● ●●●●
creep damage
Negative overshoot O- T
Peak-to-peak value PTPT General CLPIs ●●●● ●●● ●● ●●●
Importance
●●●● High
●●● Middle
●● Low
● Very low
Figure 49: Indicators for the evaluation of the live steam pressure control loop performance
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 50: Considered measurement data

Live Steam Pressure [%]
Turbine Control Valve [%]
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.
Quadrant I: Quadrant II:

Power output too big Power output too small
LS pressure too low LS pressure too low
Live Steam Pressure Control Deviation [%]
Quadrant IV: Quadrant III:

LS pressure too high LS pressure too high
Power Output Control Deviation [%]
Figure 52: Control deviations of the power output and of the live steam pressure in form of a scatter
Power Output Range [%] Live Steam Pressure Range [%]

Quadrant I -0.4 1.3
Quadrant II 0.7 1.5
Quadrant III 0.5 -1.3
Quadrant IV -0.3 -1.5
Table 8: Ranges of the power output and live steam pressure deviations
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.

LS pressure too low 70% LS pressure too low
70%

Figure 53: 70 % of the power output samples close to zero and 70 % of the live steam pressure samples
close to zero

LS pressure too low LS pressure too low

Figure 54: Ranges of the power output and live steam pressure deviations
3.6 Feed Water Control Loop

In contrast to the previously discussed control loops, the feed water control depends on the type of the
boiler. The term 'feed water control' is by the way not quite correct, as the feed water mass flow is the
manipulated variable instead the controlled variable. Control variables are the drum level in a drum
boiler and e. g. the steam enthalpy/steam temperature at the end of the evaporator in the once-
through boiler. However, the term 'feed water control' is common and will be used in the following.
3.7 Feed Water Control in a Drum Boiler

The feed water control in drum boilers regulates the feed water flow to the boiler in order to keep the
level in the drum within the desired limits. These limits are specified by the boiler manufacturer. The
drum level actual value may neither exceed the permissible maximum water level nor fall below the
permissible minimum water level during the steady-state operation as well as during load changes
(see Figure 55). During the steady-state operation, the steam mass flow leaving the drum is relatively
constant. If the power output changes, the steam mass flow to the turbine changes too. Any sustained
difference between the steam and feed water mass flow can quickly empty or fill the drum. If the drum
level is too high, the water can enter the superheater and damage the superheater tubes and the
turbine. If the drum level is too low, boiler tubes will be damaged by overheating. For this reason, the
feed water mass flow into the drum must be adjusted to the steam mass flow leaving the drum.
Thereby, the steam mass flow is the main disturbance variable of the drum level control.
Figure 55: Measuring and control ranges of a drum [36]
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 control structures of the drum level control are classified as the single-component control, the
triple-component control and the cascaded triple-component control [36].
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].
Figure 56: Configuration of the drum level control loops [36]:

a) Single-component control
b) Triple-component control
c) Triple-component control with cascaded structure

According to the guideline VDI/VDE 3502 the drum level control loop performance 'is principally
characterised by the steady-state control deviation and the transitory control deviation (overshoot).'
Besides, this guideline gives the information that the steady-state and the transitory control deviations
may not exceed the maximum permissible limits during the steady-state operation as well as during
load changes. According to this guideline 'further requirements bring no operational advantages'.
VDI/VDE 3502 points out that the performance of the drum level control is influenced by:
• Configuration of the control loop,
• Transfer function of the controlled system,
• Disturbance transfer function of the controlled system and
• Properties of the measuring devices and control devices.
3.7.4 Control Aims

The drum level control is one of the most important power plant controls due to the following reasons:
• Plant lifetime. As long as the drum water level stays within desired limits, boiler components
are protected from overheating and from water entry. However, exceeding of these limits can
lead to damage of boiler components and of the steam turbine.
• Availability. A proper drum level control minimizes the downtime of power plant units and
reduces costs.

Regarding the control aims of the drum level control, a set of applicable control loop performance
indicators has to be selected according to the following information:
• plant lifetime,
• plant availability,
• capability to compensate disturbances.
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
Figure 57: Considered measurement data
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.
Water Level [%]

Max. permissible
100
water level 100% y
x
Set Point 0%
0
x
y
Min. permissible
100
water level 100%
t
0…x % Maximum permissible range of drum level steady-state control deviation

x…y % Maximum permissible range of drum level transitory control deviation
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.
Drum Level Control

Importance
Control Loop Performance
Considered Effect Full Load Part Load Low Load Load
Indicator
Damage of boiler components and of
Positive Overshoot O+T
a steam turbine
1*
Negative Overshoot O-T Overheating of boiler tubes 1*
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
3.8 Feed Water Control in a Once-Through Boiler

The control system of the once-through boiler differs completely from the one of the drum boiler. An
important feature of once-through boilers is that the feed water mass flow influences the steam mass
flow, the steam temperature and the steam pressure. The main task of the feed water control is to
adjust precisely the feed water mass flow to the fuel mass flow in order to [38]:
• ensure the stability of the fluid mass flow in the evaporator to prevent boiler components from
inadmissible steam temperatures,
• maintain the controllability of attemperation control valves within the specified control range,
• execute the changes from once-through operation to low load operation and vice versa
without any rapid steam temperature deviations,
• maintain steam temperature deviations within a permissible temperature range during load
changes as well as during frequency control operation.
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].

It is distinguished between the feed water control during once-through operation and during low load
operation, when considering the feed water control in once-through boilers.
3.8.2.1 Feed Water Control during Once-Through Operation

The feed water mass flow during once-through operation can be controlled either as a function of the
enthalpy/temperature at the evaporator outlet or depending on the ratio of the attemperation water
mass flow to the feed water mass flow. The methods of feed water control are described below.
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.
3.8.2.2 Feed Water Control during Low Load Operation

Recirculation takes place in a once-through boiler during low load operation. The aim of the
recirculation is to avoid that wet steam reaches the superheaters or even the turbine in order to
prevent these components from damage [17]. When leaving the evaporator, the water-steam mixture
doesn't flow to the superheaters, but is lead to the separator vessel. After that the physical separation
of the water and steam takes place in the sediment bowl. The recirculation pump pumps the separated
water through the recirculation valve to the feed water pipe. Thereby, the water level control has to
maintain the water level in the separator vessel within specified range and to avoid the oversupply of
the separator vessel [38].
At the same time, the minimum mass flow control has to ensure the minimum mass flow through the
evaporator. Even if the steam temperature actual value is lower than its set point, the evaporator mass
flow actual value must not be lower than the minimum value of the evaporator mass flow. The water
level control loop and the minimum mass flow control loop are coupled and influence each other [38].

According to the guideline VDI/VDE 3506 the quality of the feed water control during the once-through
operation is defined as the deviation of the particular controlled variable (i.e. the steam enthalpy/steam
temperature in the evaporator outlet or ratio of the attemperation mass flow to feed water mass flow)
from the specified set point value. Besides, this guideline gives information that all controlled variables
provide a delayed response to changes in the fuel and the feed water mass flow. Furthermore, the
achievable control performance depends on the disturbances and on control behaviour of the
controlled system [38].
According to the guideline VDI/VDE 3506, the fluctuation of the water level control in the sediment
bowl during low load operation may be permitted to avoid rapid temperature and enthalpy changes of
the slightly superheated steam. As for the minimum mass flow control, only low undershoots of the
evaporator mass flow from the minimum value of the evaporator mass flow may occur for a short time.
3.8.4 Control Aims

The evaporator outlet temperature/enthalpy control has to adjust as good as possible the feed water
mass flow to the fuel mass flow and thus to guarantee:
• Stability of the fluid mass flow in the evaporator in order to prevent unallowable steam
temperatures and thus to avoid the lifetime consumption of thick-walled components.
• Operation of attemperation control valves within the control range in order to enable a
proper steam temperature control.

A set of applicable indicators for the evaluation of the enthalpy control loop performance has to be
selected referring to the control aims of the enthalpy control.
Examples for the application of control loop performance indicators to exemplary enthalpy
measurement data are demonstrated in the following subchapters. The exemplary enthalpy
measurement data are shown in Figure 61-a). The corresponding negative control deviation is shown
in Figure 61-b. These data were evaluated by means of time windows with time interval T=60 min.
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

IAET is a control loop performance indicator, in which positive and negative enthalpy deviations are
considered. The results of the application of the indicator IAET to the exemplary enthalpy
measurement data are shown in Figure 62-b. This figure shows that the absolute values of the
enthalpy control deviation are concentrated in the range from 6 to 18 kJ/kg.

ISET is a control loop performance indicator that penalizes heavily all large enthalpy deviations. The
results of the application of the indicator ISET to the exemplary enthalpy measurement data are shown
in Figure 62-c. The values of ISET are in the range from 50 to 450 kJ²/kg² (see Figure 62-c). This
means that there are no large enthalpy deviations from the set point.

The control loop performance indicator MVDT provides the information about the influence of the
enthalpy on the stable mass flow in the evaporator and on the quality of the live steam pressure and
power output control. The evaluated MVDT-values are shown in Figure 62-a. Most of the enthalpy
mean value deviations are concentrated in the range from -4 to 4 kJ/kg (see Figure 62-a). This means
that the enthalpy control keeps the enthalpy mean value very well at its set point.
a)
Distribution of MVD [%]

100
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
Figure 62: a) Distribution of MVDT-values

b) Distribution of IAET-values
c) Distribution of ISET-values
3.8.6 Indicators for the Evaluation of Evaporator Outlet Temperature/Enthalpy

Control Loop Performance
The control loop performance indicators for the evaluation of the evaporator outlet
temperature/enthalpy control loop performance are summarized in Figure 63.
Control of the Evaporator Outlet Temperature/Enthalpy

Importance
Control Loop Performance Indicator Relevance Full Load Part Load
Load Change
Operation Operation
A stable mass flow in the evaporator,
Keeping of the attemperation control
Mean value deviation MVDT valves within the controlled ranges, ●●●● ●●● ●●●
Influence on the quality of the LS
Pressure and PO Controls
Negative overshoot O-T General CLPIs,
Peak-to-peak value PTPT Disturbance for steam temperature control ●●● ●● ●●
Integral of absolute error IAET loops of following supeheaters
Importance
●●●● 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 Lifetime Consumption of Thick-Walled Components

Within the framework of this research project, the lifetime consumption of thick-walled components is
taken into account. The thick-walled components considered are the pressure-retaining boiler
components, which are not heated from the outside (e.g. a drum). It is important to note that the aim of
this research project is not to calculate the lifetime consumption of an exemplary thick-walled
component, but to assess the impact of the steam temperature control loop performance on the
lifetime consumption of thick-walled components.
Referring to the guidelines DIN EN 12952-3 and DIN EN 12952-4, the lifetime consumption of thick-
walled components depends, amongst others, on creep damage and on fatigue damage of the
material. The following subchapters describe the influence of the live steam temperature on the
lifetime consumption of thick-walled components due to creep and fatigue damage.
4.1 Creep Damage
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
Figure 64: Creep curve [26]

ε
σop increases
Creep strain
and/or
ϑop increases
Figure 65: Accelerated creep curve [45]
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).
4.1.2 Assessment of the Lifetime Consumption due to Creep Damage

According to the guideline DIN EN 12952-4, the creep damage of a material can be assessed by
means of the creep rapture strength and depends on the absolute value of the operating steam
temperature, the absolute value of the operating stress and the kind of material (see Figure 66). The
creep rupture strength Rm/T/ϑ is the stress needed to cause fracture in a material at temperature ϑ
after T hours [43]. The operating stress σop is proportional to the steam pressure p and is calculated by
formula (4.1) for cylindrical shells and by formula (4.2) for spherical shells [19].
Life Time Consumption
Creep Damage Fatigue Damage
Creep Rapture Strength
Operating Steam Operating

Temperature Stress Kind of a Material
Operating Steam
Pressure
Figure 66: Lifetime consumption of thick-walled components due to creep damage
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
104 105 2·105 TR

lg T in h
Figure 67: Diagram for the determination of TR

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]:
Dcreep = ∑ Dcreep ,k (4.6)

k
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):
∆Labs = Dcreep , SP − Dcreep , AV (4.7)
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.
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)).
ϑop = ϑde → Top = Tde (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)).
ϑaop < ϑde → Top > Tde (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
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)).
∆ϑ1 = 10 K → Top = 0.4...0.5 ⋅ Tde (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.
ϑ(t)
Δϑ1
ϑaop ϑde
t
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)).
∆ϑo = 10 K → Top = 0.8...0.9 ⋅ Tde (4.12)
ϑ(t)
Δϑo
ϑaop=ϑde
t
ϑop 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.
4.1.5 Examples concerning Lifetime Consumption due to Creep Damage
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
[%] [hours] [bar] Lifetime Consumption Dcreep,k [%]

100% 102600 285 17,38
80% 102600 228 2,57
60% 102600 171 0,22
40% 102600 114 0,01
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
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
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
604
602
600
598
596
594
592
0 5 10 15 20 25 30 35
t [min]
Actual Value
Set Point
Figure 72: Idealized steam temperature curve

Operating Steam
Load Time Pressure 600 600±2 600±3 600±4 600±5 600±6 600±7 600±8 600±9 600±10
[%] [hours] [bar] Lifetime Consumption ΔDcreep,k [%]
100% 102600 285 17.38 17.41 17.54 17.74 18.03 18.39 18.84 19.38 20.01 20.74
80% 27360 228 0.69 0.69 0.69 0.7 0.71 0.73 0.75 0.78 0.81 0.84
60% 25080 171 0.05 0.05 0.05 0.05 0.06 0.06 0.06 0.06 0.06 0.07
40% 18240 114 0 0 0 0 0 0 0 0 0 0
∑Lifetime Consumption Dcreep [%] 18.12 18.15 18.28 18.49 18.8 19.18 19.65 20.22 20.88 21.65
Absolute Lifetime Loss ΔLabs [%] 0 -0.02 -0.16 -0.37 -0.68 -1.06 -1.53 -2.1 -2.76 -3.53
Relative Lifetime Loss ΔLrel [%] 0 -0.2 -0.9 -2 -3.8 -5.8 -8.4 -11.6 -15.2 -19.5
Table 13: Lifetime consumption due to steam temperature fluctuations
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
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
605
600
595
590
0 20 40 60 80 100
t [%]
Figure 73: Schematic representation of an exemplary steam temperature curve
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
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
100% 20520 20 285 600±7 3.77 [-7;-5) & (5;7] 4.6
100% 2052 2 285 600±9 0.4 [-9;-7) & (7;9] 0.3
∑Lifetime Consumption at 100%-Load Dcreep,100% [%] 18.04
80% 10670 39 228 600±4 0.27 [-4;4] 80.5
→
80% 10670 39 228 600±6 0.28 [-6;-4) & (4;6] 15.1
80% 5472 20 228 600±8 0.16 [-8;-6) & (6;8] 4.1
80% 547.2 2 228 600±10 0.02 [-10;-8) & (8;10] 0.3
60% 9781.2 39 171 600±5 0.02 [-5;5] 82.8
→
60% 9781.2 39 171 600±7 0.02 [-7;-5) & (5;7] 13.2
60% 5016 20 171 600±9 0.01 [-9;-7) & (7;9] 3.7
60% 501.6 2 171 600±11 0.00 [-11;-9) & (9;11] 0.3
Total Lifetime Consumption Dcreep [%] 18.82

Absolute Lifetime Loss ΔLabs [%] -0.7
Relative Lifetime Loss ΔLrel [%] -3.9
Table 14: Scenario 1
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
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
[%] [hours] [%] [bar] [°C] [%] [K] [%]

100% 43092 42 285 600±3 7.37 [-3;3] 79.3
→
100% 43092 42 285 600±5 7.57 [-5;-3) & (3;5] 17.0
100% 14364 14 285 600±7 2.64 [-7;-5) & (5;7] 3.4
100% 2052 2 285 600±9 0.4 [-9;-7) & (7;9] 0.3
80% 11491.2 42 228 600±4 0.29 [-4;4] 82.1
→
80% 11491.2 42 228 600±6 0.31 [-6;-4) & (4;6] 14.5
80% 3830.4 14 228 600±8 0.11 [-8;-6) & (6;8] 3.1
80% 547.2 2 228 600±10 0.02 [-10;-8) & (8;10] 0.3
60% 10533.6 42 171 600±5 0.02 [-5;5] 84.3
→
60% 10533.6 42 171 600±7 0.02 [-7;-5) & (5;7] 12.6
60% 3511.2 14 171 600±9 0.01 [-9;-7) & (7;9] 2.8
60% 501.6 2 171 600±11 0.00 [-11;-9) & (9;11] 0.3
Total Lifetime Consumption Dcreep [%] 18.76

Absolute Lifetime Loss ΔLabs [%] -0.64
Relative Lifetime Loss ΔLrel [%] -3.5
Table 15: Scenario 2
4.2 Fatigue Damage
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.
4.2.2 Assessment of the Lifetime Consumption due to Fatigue Damage

In order to determine the fatigue damage of a material it is necessary to assess the material stress
and to detect the material stress load cycles. Since only load cycles beyond a certain stress value
cause fatigue, the load cycles are to be classified. The following subchapters describe step-by-step
assessment of material stress, detection and classification of material stress load cycles, assessment
of fatigue and finally determination of material fatigue damage.
4.2.2.1 Material Stress

Annex B of the guideline DIN EN 12952-4 describes a method concerning the assessment of the
fatigue damage of boiler components during the power plant operation.
According to the guideline DIN EN 12952-4, the material stress σm is calculated as a sum of the
compressive stress σp and the thermal stress σϑ (see Figure 74 and Formula (4.13)). Both, the
compressive and the thermal stress depend on geometrical dimensions of a thick-walled component
and on physical properties of a thick-walled component's material [10]. The compressive stress σp is
proportional to the absolute value of the steam pressure inside a thick-walled component p and the
thermal stress σϑ is proportional to the temperature difference in the wall of a thick–walled
component Δϑw [10].
Life Time Consumption
Creep Damage Fatigue Damage
Stress
Compressive Stress Thermal Stress
Steam pressure inside Temperature difference in

a thick-walled component the wall of a thick-walled
component
Geometrical dimensions of a Physical properties of a

thick-walled component material
Figure 74: Lifetime consumption of thick-walled components due to fatigue damage

d ms β
σ m = σ p + σϑ = αm p + α ϑ Lϑ ∆ϑw (4.13)
2ems 1 −ν
The temperature difference in the wall of a thick-walled component Δϑw is defined as the
difference between the middle integral wall temperature ϑwm and the inside wall temperature ϑwi. The
inside wall temperature approaches the steam temperature inside a thick-walled component and the
middle wall temperature ϑwm follows the inside wall temperature ϑwi by a time delay. This behaviour is
shown in Figure 75 in a simple model. This model is often used as a first approximation for the
determination of Δϑw.
PT1
ϑi ϑm Δϑ
T=220
-
Figure 75: Approximation for determination of Δϑw
4.2.2.2 Detection of Stress Load Cycles

The fatigue of material doesn't depend on absolute values of material stress, but on the size and
number of material stress load cycles. In order to detect the material stress load cycles, it is necessary
to detect the extreme values of the assessed material stress. The detection of extreme values is
applied to every succession of three material stress values σm1, σm2, σm3 using the following method: if
the Boolean expression
(σm1<σm2 and σm3<σm2) or (σm1>σm2 and σm3>σm2) (4.14)
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.
σm(t) σm4 σm(t) σm4

σm6 σm6
σm2
σm3
σm1 σm5 σm1 σm5

t t
Figure 77: Detection of stress load cycles and deletion of corresponding load cycle values
4.2.2.3 Remaining Sequence of Relative Extremes

The chronological sequence of extreme values of material stress, which doesn’t contain any stress
load cycles, is called remaining sequence of relative extremes (RSE) [10].
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
4.2.2.4 Classification of Detected Material Stress Load Cycles

The detected material stress load cycles are to be classified depending on the range of stress 2fa and
on the temperature ϑwi*, as shown in Table 16. The range of stress load cycle 2fva (see Annex B of the
guideline DIN EN 12952-3, equations (B-2) and (B-5)) is:
2 f va= σ m 2 − σ m3 (4.15)
2fa is to be calculated out of 2fva as follows:

• In the elastic range (2fva<2Rp0.2/ϑ)
(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*)
4.2.2.5 Assessment of the Lifetime Consumption due to Fatigue Damage
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
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:
DF = ∑∑ ∆DFi k + DF RSE (4.20)

i k
in which DF RSE is the fatigue damage, which is caused by the RSE.
4.2.3 Examples concerning Lifetime Consumption due to Fatigue Damage
4.2.3.1 Material Stress of an Exemplary Thick-Walled Component

In this subchapter the influence of the steam pressure and steam temperature on the material stress
of an exemplary thick-walled component during different operational modes is shown. The exemplary
thick-walled component is a superheater collector made of X10CrMoVNb9-1 steel. Table 17 shows the
exemplary geometrical dimensions of the collector and the physical properties of the X10CrMoVNb9-1
steel, which are used for the assessment of the material stress.
The exemplary steam pressure and steam temperature measurement data are shown in Figure 79-a
and Figure 79-b. On the basis of the steam pressure measurement data it can be seen that the power
plant considered was operated at full load operation for the first 8 hours, followed by a negative load
change and part load operation for 4.5 hours (see Figure 79-a). Afterwards, there is a positive load
change.
Variable Value Unit
αm 3.5 [-]
αϑ 2.0 [-]
di 230 [mm]
ems 55 [mm]
β -6
12.4x10 for 20...550°C [1/K]
Lt
E 174000 for 550°C [N/mm²]

t
ν 0.3 [-]
Table 17: Geometrical dimensions of the exemplary superheater collector and physical properties of the
material
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
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]
Figure 79: Steam pressure and steam temperature
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.
a) 300
LS Pressure [bar]
250
200 60.6 bar
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]
Figure 81: Steam pressure and assessed compressive stress

Compressive Stress [N/mm²]
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: Assessed compressive, thermal and material stress
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).
260
240
Material Stress [N/mm²]

220
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.
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]
Figure 84: Detection of material stress load cycles

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
Material Stress [N/mm²] 240
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
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.
Variable Steam Collector 1 Steam Collector 2 Unit

αm 3.5 [-]
αϑ 2.0 [-]
di 250 225 [mm]
ems 95 116.5 [mm]
βLt 12.6x10-6 for 20...600°C [1/K]
Et 167000 for 600°C [N/mm²]
ν 0.3 [-]
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.
ΔP Δp Allowable Δϑw1 Allowable Δϑw2

[%] [bar] [K] [K]
5 15.3 30 30.3
10 30.6 28.4 29
15 45.9 26.8 27.7
20 61.2 25.1 26.4
25 76.5 23.5 25.1
30 91.8 21.9 23.8
35 107.1 20.3 22.5
40 122.4 18.7 21.2
45 137.7 17 19.9
50 153 15.4 18.5
55 168.3 13.8 17.2
60 183.6 12.2 15.9
65 198.9 10.6 14.6
Table 19: Assessed Δϑw1 and Δϑw2 depending on Δp
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
4.3 Influence of Steam Pressure and Steam Temperature Control on Lifetime

Consumption of Thick-Walled Components
The following subchapters describe the influence of the steam pressure and of the steam temperature
control on the lifetime consumption of thick-walled components during different operational modes.
4.3.1 Full Load Operation

For economical 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. Thereby, the
corresponding compressive stress in material of thick-walled components is relatively high. In order to
protect thick-walled components from the creep damage, the live steam temperature is to be regulated
as good as possible. For this aim the live steam temperature mean value should not exceed the set
point and the live steam temperature deviations should be kept as low as possible. As very high
demands are made on the live steam temperature control during full load operation, the live steam
temperature controller should be optimized for full load operation.
4.3.2 Part Load Operation

During part load operation, the dynamics of the steam generation process is much slower than during
full load operation. Therefore, the response behaviour of the live steam temperature with respect to
the control action of the attemperation is also slower, consequently leading to higher steam
temperature control deviations. Since, the live steam pressure and the corresponding compressive
stress during part load operation are significantly lower than during full load operation, the live steam
temperature mean value deviation and the live steam temperature deviations during part load
operation can be allowed to be bigger than during full load operation.
4.3.3 Low Load Operation

During low load operation, the dynamics of the steam generation process is slower than during part
load operation. Therefore, the response behaviour of the live steam temperature with respect to the
control action of the attemperation is also slower, consequently leading to higher steam temperature
control deviations. Since, the live steam pressure and the corresponding compressive stress during
low load operation are lower than during part load operation, the live steam temperature deviations
and the live steam temperature mean value during low load operation can be allowed to be bigger
than during part load operation.
4.3.4 Positive and Negative Load Changes

Positive and negative load changes require big live steam pressure changes, which cause big
compressive stress load cycles. When additionally fast temperature changes occur, big thermal stress
load cycles arise in the 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.
Chapter 5 - Benchmarks for Control Loop Performance 88
5 Benchmarks for Control Loop Performance

This chapter is dedicated to benchmarks for the control loop performance of different control loops of
steam power plants.
First of all, an introduction to benchmarks is given in section 5.1. General definitions along with some
adaptations to the area of control loop performance lay the groundwork for the subsequent
investigations.
In a second step, the benchmark development scheme is introduced to illustrate the general approach
of how to perform a benchmark (section 5.2).
Afterwards, individual benchmarks for various control loops are considered independently one after
another, starting with the live steam temperature control loop in section 5.3.
5.1 Introduction to Benchmarks
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.
5.1.2 Benchmark Classification

Within this research project, two different types of benchmarks are distinguished. In the following, they
are discussed briefly:
1. Best-practice benchmarks: In the case of best-practice benchmarks, the predefined
benchmark value is subjectively chosen. The control loop performance of an existing power
plant, which is considered to be good, is used as benchmark value and serves as reference
value for the control loop performance of other power plants. Since the benchmark value
arises from an existing process, it is also achievable by the processes that are subject to the
comparison and may even be outvalued by such a process.
2. Mathematically derived benchmark: A mathematically derived benchmark value is used for
the comparison. It is usually found by means of an optimisation problem. Therefore, the
benchmark value is optimal with respect to formulated optimisation problem, which implies
that the benchmark is not achievable by a real process. It is rather an upper limit to what is
possible in reality.
5.1.3 Goals of Benchmarks

The goals of benchmarks in the case of control loop performance investigations are twofold:
Firstly, in order to compare different power plants against each other, a common basis for the
comparison is helpful. This common basis can be a benchmark value as described above.
Secondly, the benchmark value provides a reference point not only for the comparison, but also to
assess and valuate the actual performance of a plant. Since the benchmark value indicates the best
possible performance from a theoretical point of view, the deviation from this value shows the
theoretical potential for improvement of the process.
5.2 The Development Scheme

The procedure of performing a benchmark for the control loop performance is discussed with the aid
of the so-called ‘benchmark development scheme’, introduced in Figure 87.
‘Real World’ ‘Benchmark World’

Approximation
Limitations Optimal Conditions
• Power plant • Process model
1.) 1.)
• Controller • Benchmark controller
No transfer
2.) Measurement data Simulation data 2.)
3.) Real control loop Benchmark Value 3.)

performance
Comparison
i.e. Benchmarking
Figure 87: Benchmark Development Scheme
This scheme consists of two distinct areas, both of which are constructed in a similar way:
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.
5.3 Live Steam Temperature Control Loop

The live steam temperature control loop being discussed in detail in chapter 3.1, the explanation at
this point is limited to a brief recall and some relevant aspects for the benchmarking process.
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.
5.3.2 Superheater Model

The superheater model is a simple IO-model based on guideline VDI/VDE 3503 ‘Steam Temperature
Control in Fossil Fired Steam Power Stations’ [37]. Its parameters are listed in Table 20.
Variable Description Type

hI Specific enthalpy of input steam Input signal
*
h Specific enthalpy of steam before sensor (Output Signal)
hO Specific enthalpy of live steam (output) Output signal
qF Specific flue gas heat Input signal
T100 Total superheater time constant (100%) Parameter
Ts Sensor time constant Parameter
Table 20: Variables of super heater model
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
Figure 88: Superheater model according to VDI/VDE 3503
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.
5.3.3 Benchmark Controller

The benchmark controller is designed as an optimal state feedback controller. A state feedback
controller computes the control variable as the weighted sum of the states of the system, where the
weights are the controller gains. This results in mere proportional feedback. To avoid permanent
control errors, an integrator is inevitable. That is why an additional state, the integral of the control
error, is added to the model and taken into account in the controller design. Thus, the combination of
C1 and C2 represents the entire benchmark controller.
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.
5.3.4 Simulation Results

In order to demonstrate the performance of the benchmark controller, it is simulated in the closed live
steam temperature control loop with the superheater model described in section 5.3.2. An overview of
the simulation setup is given in Figure 89. Blue colour signifies measurement data (consistent with the
‘real world’) whereas green colour refers to simulation data, i.e. the ‘benchmark world’. The main
interest lies in the output temperature of the superheater ϑO.
qF
ϑfeed ϑI ϑO
superheater
xSystem
C1
Follow-up
ϑI,SP xε ϑSP
controller C2
Figure 89: Simulation setup with benchmark controller
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’).
Live Steam Temperature

547
546
545
[°C]
544
543
Set point
542
Benchmark controller
Real plant
541
0 1000 2000 3000 4000 5000
Time [s]
Figure 90: Live steam temperature and set point
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
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)
5.3.5 Realisation of the Benchmark

For both measurement data and simulation data the control loop performance has to be assessed.
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 almost four hours of data of an
exemplary power plant operating at full load in frequency control mode. The benchmark value ISEBV
adds up to 4,70, the Integral of Squared Error of the real control loop performance ISERP reaches the
total of 11,49. The comparison is illustrated on the following scale in Figure 92. For an illustrative
reason, the Integral of Squared Error of zero is also indicated. An ISE of zero implies that the live
steam temperature does not deviate from its set point at all time, which is of course not possible.
Best
ISE = 0
performance
ISEBV = 4,70
ISERP = 11,49
Less
performance
Figure 92: Benchmark value and performance of the real process
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.
5.4 Live Steam Pressure Control Loop

In order to perform a benchmark for the live steam pressure control loop, the power station is assumed
to operate in turbine-in-control mode. In this operational mode, the power output is controlled by the
turbine inlet valve. The live steam pressure, while being disturbed by the fluctuating mass flow to the
turbine, is controlled by the fuel supply. In the case of fixed pressure mode, the reference value of the
live steam pressure is a constant set point.
Again, the benchmark is derived via model based optimisation of the controller. This requires a steam
generator model, which is discussed next. Afterwards, the benchmark controller is derived for this
model and simulation results of the closed loop lead to the final subchapter which is dedicated to the
realisation of the benchmark.
5.4.1 Steam Generator Model

The steam generator is based on VDI/VDE guideline 3508 ‘unit control of thermal power stations’.
Nonetheless, some adaptations are needed for the benchmark problem. A block diagram of the model
is shown in Figure 93, its parameters are listed in Table 21. According to the VDI/VDE guideline, the
dynamic behaviour of heat release and steam generation correspond to a higher-order delay which is
characterised by its dead time Tu and its balancing time Tg. In order to keep the model as simple as
possible, a third order delay with time constant TStG is chosen. The generated steam is input to the
subprocess of steam storage. The steam pressure pSt results from balancing the generated steam and
the steam mass flow supplied to the turbine mT. The storage effect is represented by an integrator with
storage time constant TSto, depending on the volume of the steam accumulator.
mT
mF mStG pSt
TStG TSto
yT
Figure 93: Steam generator model according to VDI/VDE guideline 3508
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.
Variable Description Type

mF Fuel supply (mass flow) Input signal
mStG Generated steam mass flow Inner Signal
mT Supplied steam mass flow (turbine) Input signal
pSt Live steam pressure Output signal
TStG Time constant of steam generator Parameter
TSto Steam storage time constant Parameter
Table 21: Variables of steam generator model
5.4.2 Model Validation

In order to validate the chosen model, it is simulated in closed pressure control loop with
measurement data from an exemplary lignite fired power plant operating at full load. The actual live
steam pressure serves as set point for the simulation model. The simulated pressure run along with its
set point are shown in Figure 94. Deviations arise not only from the changing set point value but more
importantly from the disturbance variable mT. The simulated pressure being in conformance with the
measurement data implies that the fuel rates should both show analogous behaviour. The fuel rates in
Figure 94 confirm this supposition and the model is proved valid. The time scales of the plots differ
due to an illustrative reason.
Live steam pressure Fuel rate

265 230
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]
Figure 94: Validation of steam generator model
5.4.3 Benchmark Controller

The benchmark controller is a state feedback controller designed in analogy to the benchmark
controller for the live steam pressure control loop. The block diagram of the closed loop is shown in
Figure 95 with the benchmark controller highlighted in green colour. For a more detailed description
see chapter 5.3.3.
5.4.4 Simulation Results

The performance of the benchmark controller is demonstrated by means of simulating the closed live
steam pressure control loop with the steam generator model described in section 5.4.1 along with the
benchmark controller. The simulation setup is shown in Figure 95. The simulation model is fed with
measurement data from an exemplary power plant operating at full load in fixed pressure mode. Blue
colour signifies measurement data (consistent with the ‘real world’, see Figure 95). The live steam
pressure is the simulation result, hence highlighted in green colour (according to the ‘benchmark
world’).
The measurement data are:

1. Live steam pressure set point: The live steam pressure set point is taken from
measurement data, but since normalized quantities are used in this model, the
constant set point is one.
2. Supplied steam mass flow (turbine): The supplied steam mass flow to the turbine is
a disturbance variable. The run of the steam mass flow of an exemplary power plant is
shown in Figure 96.
mT
pSt,SP xε mF Steam generation pSt

C2
and storage
xSystem
C1
Figure 95: Simulation setup with benchmark controller
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
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 96: Supplied steam mass flow (measurement data)
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
Set point
250
0 2000 4000 6000
Time [s]
Figure 97: Live steam pressure and set point
5.4.5 Realisation of the Benchmark

The evaluation of the benchmark is carried out in two steps:
Step 1: Qualification: The control variable fuel mass flow has to be judged in a first step. This is done
by comparing the gradients shown in Figure 98. The benchmark controller was designed such that its
output (in green colour) and the gradient of the real fuel mass flow are of equal maximum amplitude.
Besides some outliers (here approximately 1% of the data) the values remain within the red bounds.
Applying these bounds to the fuel mass flow of the benchmark controller, it can be seen that these
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
Gradient of simulation data (benchmark controller)

0.04
0.02
0
-0.02
-0.04
0 1000 2000 3000 4000 5000 6000
time [s]
Figure 98: Fuel mass flow gradients
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
Figure 99: Benchmark value and performance of the real process
5.5 Conclusion on Benchmark

In chapters 5.3 and 5.4, the development of benchmarks for the live steam temperature control loop
as well as the live steam pressure control loop was presented.
These are valid approaches, although they suffer from practical limitations:
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
6 Results of Measurement Data Evaluation and Analysis

The results of the measurement data evaluation are described in chapter 6.
6.1 Overview of Power Plant Units that Provided Measurement Data

Within the framework of this research project the measurement data of hard coal fired, lignite fired and
combined cycle power plant units were collected. Not all investigated control loops were evaluated for
each power plant unit, because some provided measurement data sets were not complete. The
measurement data were evaluated by means of the sliding time window T=60 min. Therefore, the
measurement data sets, which cover at least a period of one hour, were required.
In total, the measurement data of fourteen European power plant units were evaluated:
• eight hard coal fired power plant units,
• five lignite fired power plant units and
• one combined cycle power plant unit.
An overview of these power plant units is given in Table 22. These power plant units are classified
depending on power plant type and power output (see Table 22).
Nominal Power Output [MW]
≤300 301...600 601...900
h - hard coal fired power plant
I II III
b - lignite fired power plant
c - combined cycle Ih Ic IIh IIb IIIb
Number of Power Plant Units 4 1 4 4 1
Table 22: Power plant units that provided measurement data
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.
6.2 Classification of Cases for Measurement Data Evaluation

A power plant unit can be operated in different operational modes. During steady-state operation of a
power plant the power output set point is constant. The power plant operation, during which a power
plant provides control energy, is called grid control operation. During the grid control operation either
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.
6.2.1 Power Output Control Loop Performance

A very good power output control loop performance was achieved in some power plant units that were
operated in natural sliding and in modified sliding pressure operation during steady-state and during
grid control operation. As for the power output control loop performance during grid control operation,
the following can be concluded: the larger the grid influence is, the larger the power output control
deviations are.
As for positive and negative load changes it can be seen that the larger a positive or a negative load
change is, the larger is the power output control deviation. Generally, power output control deviations
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.
6.2.2 Live Steam Pressure Control Loop Performance

A very good live steam pressure control loop performance was achieved in power plant units that are
operated in modified sliding pressure operation. On the basis of evaluation results it can be seen that
the larger the grid influence is, the larger are the deviations of the live steam pressure from the set
point. Besides, the fluctuations of the live steam pressure are larger during part load operation than
during full load operation.
Generally, a good live steam pressure control loop performance was achieved in power plants that are
operated in fixed-pressure operation. However, in some of these power plant units the live steam
pressure actual value was on average 2 % larger than the given set point during all evaluated
operational modes.
6.2.3 Live Steam Temperature Control Loop Performance

On the basis of the evaluation results it can be concluded that the live steam temperature control loop
performance can be very different not only in different power plant units but also in different tracks. For
example, in an evaluated power plant unit with four tracks there was a very good performance of the
live steam temperature control loop in two of four tracks. At the same time very large negative mean
value deviations of the live steam temperature from the set point took place in another two tracks. For
this reason, the live steam temperature control loop performance should be determined for each track
individually.
In some power plant units a very good performance of the live steam temperature control was
achieved. Based on the evaluation results, it can be seen that the fluctuations of the live steam
temperature becomes larger with decreasing load. However, a good live steam temperature control
loop performance was achieved in some power plant units even during low load operation
The average of the live steam temperature was approximately 3 K below the set point in many power
plant units during different operational modes. This effect was especially evident during grid control
operation. Sometimes, the average of the live steam temperature was up to 35 K below the given set
point. This influences negatively the efficiency of a power plant unit.
As for positive and negative load changes, it can be generally concluded that the fluctuations of the
live steam temperature become larger during load changes. It is striking that the average of the live
steam temperature is often lower than the given set point during negative load changes.
6.2.4 Reheater Steam Temperature Control Loop Performance

The performance of the reheater steam temperature control loop was often very poor in many power
plant units. Large negative mean deviations of the reheater steam temperature from the set point, (up
to -50 K) took place during different operational modes. In some power plant units these large
negative mean value deviations of the reheater steam temperature took place regardless the
operational mode. Due to the fact that the reheater steam temperature is often lower than the given
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.
6.2.5 Enthalpy Control Loop Performance

Based on the evaluation results it can be seen that generally not very large enthalpy mean value
deviations from the set point but large enthalpy fluctuations take place in different power plant units
during different operational modes. Besides, the enthalpy fluctuations become larger with decreasing
load. Moreover, the enthalpy fluctuations during positive and negative load changes are larger than
during steady-state operation.
6.2.6 Drum Level Control Loop Performance

Drum level control deviations were always within the range of permissible drum level control deviations
in all evaluated power plant units.
Chapter 7 - Normalization of Control Loop Performance Indicators 104
7 Normalization of Control Loop Performance Indicators

In order to make control loop performance indicators comparable with each other their values are to
be normalized. For this aim, a reference histogram is created from individual histograms of different
power plants. The reference histogram has to be created for every control loop performance indicator.
When creating the reference histogram, it should be distinguished between different power plant
types, operational modes, etc.
7.1 Normalization Method

Several methods could be used for normalization of indicator values ([22],[28]). One of the possible
normalization methods is the 'Standardization' or the 'Z-Scores-Method'. In order to normalize the
values of an indicator using this method, the average value and the standard deviation of the
reference indicator values are to be determined. The normalized indicator is calculated as the
difference between the actual indicator value and the average indicator value divided by the standard
deviation (see (7.1)). Values of the normalized indicator can be positive, negative or equal to zero.
Depending on the application they represent whether the actual performance is better or worse than
the average performance or equal to the average performance.
IND qcm − IND ref,

qcm
mean
IND qcm
= act
(7.1)
norm
σ IND qcm
ref
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
Table 23: Meaning of the symbols used in formula (7.1)
7.2 Application of Normalization Method

The application of the described normalization method to an exemplary distribution of indicator values
is represented in this subchapter. The exemplary distribution of a control loop performance indicator
values is shown in Figure 101. These values were determined by evaluating the control loop
performance of the live steam temperature control loop by means of the control loop performance
indicator 'peak-to-peak value PTPT'. It is assumed that these values are the reference values for hard
coal fired power plants during full load operation.
The minimum value of PTPT, which was determined during the evaluation amounts to 3.81 K. This
value represents the best practice control loop performance (see Table 24). The maximum value of
PTPT amounts to 22.56 K and represents the worst practice control loop performance. The mean of
determined PTPT -values amounts to 7.89 K and represents the average control loop performance.
The standard deviation of the values shown in Figure 101 amounts to 3.26 K. If the live steam
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).

100
80
60
40
20
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
PTPT [K]
Figure 101: Distribution of exemplary PTPT-values
Live Steam Temperature, T=60 min.

Best Minimum Mean Maximum
Theoretical Standard
CLP Indicator Best Practice Average Practice Worst Practice Unit
Performance Deviation
Performance Performance Performance
PTPT 0 3.81 7.89 22.56 3.26 [K]

PTPT ,full load
IND norm -2.42 -1.25 0 4.5 - [-]
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.
Best Theoretical Best Practice Average Practice Worst Practice

Performance Performance Performance Performance
-2.42 -1.25 0 4.5
Figure 102: Scale of achievable control loop 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.

-2.42 -1.25 0 4.5
-2.0
Figure 103: Measuring of performance by means of performance scale - example 1

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).

-2.42 -1.25 0 4.5
-0.5
Figure 104: Measuring of performance by means of performance scale - example 2

Chapter 8 - Control Loop Performance and Economic Efficiency of a Power Plant 107
8 Control Loop Performance and Economic Efficiency of a Power Plant
8.1 Economic Efficiency of a Power Plant Operation and Influencing Factors

In "A dictionary of economics" [1], the term economic efficiency is defined as "a general term for
making the maximum use of available resources". According to the operator questionnaire, the
economic efficiency of a power plant unit is strongly influenced by the following factors:
• electricity market,
• power plant efficiency,
• fuel costs,
• lifetime consumption,
• operational costs,
• power plant availability,
• power plant reliability,
• maintenance,
• power plant flexibility,
• residue disposal.
Operators noticed that depending on power plant type these factors can have different influence on
the economic efficiency of a power plant unit. For example, such factors as power plant availability
and power plant efficiency are significant for base or middle load power plants. Such factors as power
plant reliability and power plant flexibility are significant for peak load power plants.
8.2 Power Plant Efficiency

Efficiency of a power plant producing only electrical power is defined as the quotient of the electrical
power output and the supplied fuel power. The efficiency of a power plant unit is influenced by various
single efficiencies like the steam generator efficiency, steam turbine efficiency, generator efficiency,
thermal efficiency and auxiliary power efficiency [33]. Moreover, the power plant efficiency depends on
the power plant type, fuel type used, operating load level, technology employed, power plant
availability, etc. According to [33], the efficiency during operation of power plants is about 2 % - 3 % (in
absolute terms) lower than the design efficiency. This is due to the efficiency losses, which are caused
by part load operation, start-ups and shutdowns.
The efficiency of power plant units has been improved and most likely will be continuously improving
in future. The reasons are obvious: high power plant efficiency is good for economy and environment.
Operators agreed that having high power plant efficiency allows to generate each kWh using less fuel,
which results in fuel savings and accordingly in fuel cost savings. Furthermore, they mentioned that
high power plant efficiency is accompanied by a reduction of emissions, which is a significant
argument in order to achieve the acceptance of coal-fired power plants. Moreover, it was noticed that
the power plants with better efficiency are better ranked in the merit order and are the first to switch on
when power is required. Thus the power plant efficiency has a great influence on the economic
efficiency of a power plant.
It is known that the power plant efficiency depends strongly on steam conditions before the turbine. In
order to achieve the highest possible power plant efficiency, in the concepts of coal-fired power plants
that are currently under construction increased steam pressure and steam temperature parameters
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
8.3 Part Load Capability of a Power Plant

There is a strong efficiency benefit from operating power plant at the highest load possible. To operate
a plant below the highest load possible is generally benign to a plant, but entails a loss of economic
efficiency due to higher fuel consumption per power output.
According to the questionnaire, since the supply of the renewable energy increases permanently, it is
of high interest for power plant operators to have power plant units being able to operate at a very low
load level (e.g. operation with one coal mill without additional oil firing).
As the steam generation process is slower with decreasing load level, a power plant unit responds
differently to parameter changes during part or low load operation than during full load operation. In
order to operate a power plant unit at the lowest desired load level, the control must be adapted to a
very slow steam generation process. If control cannot deal with very slow steam generation
processes, the minimum load level has to be increased. It can be seen that the proper control is
necessary for each load level. Thus the control loop performance affects the quality of energy
production at each load level.
8.4 Load Change Capability of a Power Plant

The contribution from renewable energy sources to total energy consumption in Europe grows from
year to year. In 2020 the share of renewable energy sources is expected to amount to 20 % [34]. A
challenge for all fossil power plants is the expected decrease in their full load hours due to the growing
shares of renewable energy and their priority access and priority dispatch. Thermal power plants are
currently the most important sources of balancing power in the electricity system and to ensure a safe
grid operation they have to be flexible. One of the factors influencing the operational flexibility of
thermal power plants is the maximum load change ratio [34]. The maximum load change ratio of a
plant is a part of the power plant concept. It represents the ability of a power plant unit to balance the
load in the grid and thus gains more and more in importance.
Considering the influence of increasing proportion of renewable energy and possible forecast errors,
one operator is of the opinion that "demand response is one of the issues that gain in importance and
has an economic value". Another operator added that "power plants are to be more dynamic and to be
able to provide high load gradients". One more operator pointed out the economic benefit of high
power plant load change ratio as it results "in safer qualification for primary and secondary control and
in fuel savings". Moreover, it is desirable "having units with large load variance band". In the opinion of
another one operator "it makes no sense to operate base load or middle load power plant units as a
peak load power plant unit. For instance, if a coal-fired power plant unit has to start-up within a few
hours, it is necessary to keep a boiler filled with hot water. This means to burn fuel for a long period
just to keep it warm, which makes no sense considering efficiency. Furthermore, often load changes
as well as frequent start-ups and shutdowns lead to higher lifetime consumption of power plant
components. Some studies have been carried out to evaluate the influence of the start-ups increase
on the lifetime consumption of a boiler. For instance, on the basis of 700 MWe oil power plant units
was noted, that a boiler tube leak could occur after 6 start-ups on average. It is to be mentioned, that
the lifetime consumption of a boiler depends on the temperature of a boilers before the start-up".
Operators agreed that the control loop performance doesn't influence the load change capability of a
power plant, as other limitations define the load change rate. But they indicated that the control loop
performance has an influence on the quality of power plant load changes, which can be of economic
value. For example, the high steam temperature control loop performance during load changes can
protect thick-walled components from the lifetime consumption due to fatigue damage. The high power
output control loop performance can help to meet exactly the grid demands and to ensure a stable grid
operation.
8.5 Influence of Power Plant Unavailability on Economic Efficiency

Power plant unavailability reduces the utilization time of a plant per year and reduces thus the income
of operators. On the basis of power plant data of VGB member companies, VGB analyses every year
the power plant unavailability and its causes. According to [41], the planned unavailability of fossil-fired
power plants amounted to 9.3 % in 2009. The unplanned unavailability of these power plants
amounted to 9.1 % in 2009. Therewith, the complete unavailability amounts to 18.4 % in 2009. The
major part of unplanned unavailability was caused by bunker, feeder and components of pulverizing
system, followed by components of pressure system, feed water system and steam sections as well
as by steam turbine (see Figure 108) [41].
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]
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
8.6 Power Output Control Loop Performance

Information which was given by operators to this topic leads to the conclusion that power output mean
value deviation from the set point has an influence on the economic efficiency of a plant. Operators
indicated that the 15-minutes-averages of the power output actual value are to be equal to the
predetermined value due to balancing settlement (the time periods for settlement in European
countries can be different from each other). If the power output mean value of a power plant unit
deviates from its set point and leads to unbalance in the portfolio of the load dispatcher, the load
dispatcher intervenes and solves this problem by means of other power plants units. If it is impossible
to keep the portfolio of load dispatchers in balance and a power plant operator supplies grid with less
or more energy than it has been agreed, it can lead to additional costs:
• In case of grid undersupply, the grid operator has to compensate the energy lack. As this
energy lack was caused by the power plant operator, the power plant operator has to
compensate it financially.
• In case of grid oversupply the power plant operator uses more fuel than it is necessary. The
surplus energy is not necessarily paid at the same time. This means that the additional fuel
consumption is not compensated financially.
It can be seen that in both cases the operation costs increase without compensation by additional
revenues.
Since it is necessary to keep the 15-minutes averages of the power output actual value at the
predetermined value, small power output fluctuations lose their importance. It was pointed out in the
questionnaire that the higher and prolonged power output fluctuations are often balanced throughout
the power plant fleet. Thus, significant power output deviations in one power plant can have an impact
on the whole power plant fleet. If a power plant unit provides control energy, it is important that the
output actual value follows as precisely as possible the given set point.
Furthermore, operators mentioned that the changes in fuel quality are to be compensated by the
correction controller. If the influence of fuel change cannot be corrected, it leads to the difference
between the planned and actual grid supply and influences thus economic efficiency of a plant.
Chapter 9 - Summary 112
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.
[16] Klebes, J.:

High-Efficiency Coal-Fired Power Plants Based on Proven Technology. VGB Power Tech,
Nr. 3, 2007.
[17] Klefenz, G.:
Die Regelung von Dampfkraftwerken. 4. Aufl. Mannheim: Wien: Zürich: BI-Wiss.-Verl.,
1991
[18] Lausterer, G.:
Optimierung der Kraftwerksregelung durch gezielten Einsatz mathematischer Modelle.
rtp 25, Heft 11, 1983
[19] Lehne, F.:
Ermittlung von Wandtemperaturdifferenzen in dickwandigen Bauteilen zur
Lebensdauerverbrauchsbestimmung. Dissertation, TU Braunschweig, 2003.
ISBN 3-9808121-1-1
[20] Leithner, R.; Steege, F.; Pich, R.; Erlmann, K.; Chi Trung Nguyen:
Vergleich verschiedener Verfahren zur Bestimmung der Temperaturdifferenz in
dickwandigen Bauteilen für die Lebensdauerberechnung. VGB Kraftwerkstechnik 70,
Heft 6, 1990
[21] Lunze, J.:
Regelungstechnik 2: Mehrgrößensysteme, Digitale Regelung. Springer-Verlag, 2008
[22] Mendenhall, W.; Sincich, T.:
Statistics for engineering and the sciences. 5th Edition. Pearson Prentice-Hall, Inc. Upper
Saddle river, NJ, USA 2006. ISBN: 0131877062
[23] Peck, R.; Olsen, C.; Devore, J.:
Introduction to Statistics & Data Analysis. 3rd Edition. Brooks/Cole, Cengage Learning
2009
[24] Pich, R.:
Allgemeine Betrachtungen über instationäre Wärmespannugen in krümmungsbehinderten
Platten, Hohlzylindern und Hohlkugeln mit ebenen symmetrischen Temperaturfeldern.
Dissertation, Universität Wien, 1993
[25] Pich, R.:
Der Einfluss von Heißdampftemperaturschwankungen auf die rechnerische
Lebensdauererschöpfung druckführender Bauteile. VGB Kraftwerkstechnik 65, Heft 8,
1985
[26] Roesler, J.; Harders, H.; Baecker, M.:
Mechanical behaviour of engineering materials. Metals, ceramic, polymers and
composites. Berlin, Heidelberg: Springer Verlag, 2007
[27] Rosen, J.:
The future role of renewable energy sources in European electricity supply. Dissertation.
Karlsruhe, 2007
[28] Sceaffer, R.; Mulekar, M.; McClave, J.:
Probability and statistics for engineers. 5th Edition. Brooks/Cole, Cengage Learning.
20011. ISBN-13: 978-0-534-40302-7
[29] Schlitt, H.:
Regelungstechnik in Verfahrenstechnik und Chemie : Bauelemente, Regelkreise, Planung
und Verwirklichung. 1. Aufl. Würzburg: Vogel Verlag, 1991
[30] Sindelar, R.:

Frischdampf-Temperaturregelung mit Zustandsregler aus Ingenieursicht.
Anwendungsaufsätze, at – Automatisierungstechnik 44 (1996) Heft 3.
[31] Sindelar, R.; Hanus, B., Hernych, M.:
Die Vorteile eines Reglers mit nachgebender Rückführung. VGB Kraftwerkstechnik
7/2000
[32] Singh, S. K.:
Process Control: Concepts, Dynamics and Applications. New Delhi: PHI Learning Private
Limited, 2009
[33] Spliethoff, H.:
Power Generation from Solid Fuels. Heidelberg, Dordrecht, London, New York: Springer
Verlag, 2010
[34] Official Journal of European Union:
DIRECTIVE 2009/28/EC OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL
on the promotion of the use of energy from renewable sources and amending and
subsequently repealing Directives 2001/77/EC and 2003/30/EC. 2009.
[35] TransmissionCode 2007;
TransmissionCode 2007 - Netz- und Systemregeln der deutschen
Übertragungsnetzbetreiber. Berlin: Verband der Netzbetreiber VDN e.V. beim VDEW,
2007
[36] VDI/VDE 3502:
VDI/VDE-Richtlinie 3502: Drum Level Control in Fossil Fired Steam Power Stations.
Berlin: Verein Deutscher Ingenieure, Beuth Verlag, 1996
[37] VDI/VDE 3503:
VDI/VDE-Richtlinie 3503: Steam Temperature Control in Fossil Fired Steam Power
Stations. Berlin: Verein Deutscher Ingenieure, Beuth Verlag, 1996
[38] VDI/VDE 3506, Part 1:
VDI/VDE-Richtlinie 3502: Feedwater Control for Once-Through Boilers in Fossil-Fired
Power Stations. Berlin: Verein Deutscher Ingenieure, Beuth Verlag, 1997
[39] VDI/VDE 3507:
VDI/VDE-Richtlinie 3507: Abnahme von Regelanlagen für Dampferzeuger. Berlin: Verein
Deutscher Ingenieure, Beuth Verlag, 1966
[40] VDI/VDE 3508:
VDI/VDE-Richtlinie 3508: Unit Control of Thermal Power Stations. Berlin: Verein
Deutscher Ingenieure, Beuth Verlag, 2003
[41] VGB, Eurelectric:
Technical-Scientific Report TW 103 AE: Analysis of Unavailability of Thermal Power
Plants 2000-2009. Edition 2010.
[42] Webster, G.A.; Ainsworth, R.A.:
High Temperature Component Life Assessment. Chapman & Hall, 1994
[43] Roesler, J.; Harders, H.; Baeker, M.:
Mechanical Behaviour of Engineering Materials. Metals, Ceramics, Polymers, and
Composites. Berlin, Heidelberg, New York: Springer Verlag, 2007.
ISBN 978-3-540-73446-8.
[44] Summers, E.; Isaacs, A.; Butterfield, J.; Holmes A.; Law, J.:
Collins English dictionairy: complete and unabridged, HarperCollins Publishers, 2010

[45] Spendolini, M.J.:
The Benchmarking Book. Amacom Books, 1992
[46] Zindler, H.; Hauschke, A.; Leithner, R.:
Impact of the HP Preheater Bypass on the Economizer Inlet Header. 5th International
Conference on ENERGY, ENVIRONMENT, ECOSYSTEMS and SUSTAINABLE
DEVELOPMENT (EEESD'09). Athens, Greece, September, 2009
Appendix A 118
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
Minutes of the Workshop with Manufacturers on

"Definition and Verification of the Control Loop
Performance for Different Power Plant Types"
Appendix B 121
Part 1: Steam Temperature Control

Up to now, the control loop performance was defined using one fixed value for all power plant types,
all operational modes, all operating modes, without distinction between load changes or disturbances.
That’s why the controller settings are very restrictive and the required control loop performance is not
always reachable. It is necessary to distinguish between different power plant types, operational
modes and load changes and to define reasonable control loop performance indicators. Defining
indicators it is necessary to find an alternative to the peak-to-peak value method. New methodology
should involve such indicators like mean value deviation or excess of certain ranges. Besides, it is
necessary to clarify which requirements should be made on new indicators.
Manufacturers deliver only one component of the power plant, but still have to guarantee a certain
control loop performance without knowing which further components will be added.
The temperature control is a slow control. Therefore, an error is often shown, even though the
controller reacted already. The controller shouldn't be set too precisely, because of the time delay. Too
precise controller settings lead to heavy wear of actuators, but no improved control loop performance.
• Efficiency / Lifetime of Thick-Walled Components

High control loop performance corresponds to high power plant efficiency. Therefore, the goal is to
reach higher control loop performances. Moreover, the temperature level gets higher and the distance
to the protection limits becomes smaller at the same time. If the steam temperature excesses the
protection limit, it can lead to the failure of a power plant unit, which reduces the availability of the unit.
Therefore, an additional control loop performance indicator “Availability of the power plant” should be
considered. Eventually, the protection limits should be considered too.
• Considering Actuators
Is it considered whether the manipulating variables are in a defined range and actuators are able to
execute the control?
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.
Which factors cause the wear of actuators?

Actuators wear is often caused by vibrations of the plant. The drives could be broken due to
vibrations. Water pressure shocks, heating problems or fuel value fluctuations are additional
problems. Considering actuators rather simple indicators are to be taken into account, like the
number of shifts in direction or the transient/rise time.
How does control system cope with worn actuators?
Actuators are generally less considered by operators than other power plant components. It
means that certain accuracy of actuators is required for being used in tests, but the long-time
operation of actuators isn't considered. If an actuator is broken, it is replaced.
Continuous vibrations, which often occur in control valves, lead to short movements of the
valves. Due to these movements the lubricator is pressed out and the valve is no longer
lubricated. As these short movements lead to a heavy wear of the valve, it makes sense to
penalize especially short movements in a control loop performance indicator, since it is not
possible to grease the valve again by passing through the complete valve lift in practice.
Moreover, more attention should be given to small movements of the valve, when the valve is
almost closed. The wear of the valve due to small valve movements at this working point is
Appendix B 122
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.
Part 2: Live Steam Pressure Control

When considering live steam pressure there are two different theories: in Germany the steam pressure
deviations are not heavily penalized, in China by contrast the pressure is to be controlled very
precisely.
Appendix B 123
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.
Part 3: Power Output Control

The inconsistency of gradients using the indicator "consistence of gradient's signs CGS" is caused by
set-point changes. The power output actual value can follow the power output set point only with a
time delay due to the large dead time. Rapid changes of the power output actual value and of the
power output set point within the period ΔT lead to corresponding gradients in opposite directions.
Note to Directive 3507: power output gradients, such as 5%/min. in the mean load range, are only
possible if everything is optimized for this case. However, it is to be considered how often this case
actually takes place. A peak-to-peak value consideration is not very reasonable, because it can
possibly cause problems concerning vibrations in the boiler, which are not desired and will not be
penalized by any indicator.
Part 4: Feed Water Control
• Drum Level Control

The drum acts as an integrator. Therefore, the drum can be affected in such a way that it will vibrate.
The oscillations of the drum influence other parameters. Therefore, consideration of fluctuations of the
feed water mass flow is very reasonable.
• 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.
Part 5: General Impression and Comments

Generally, new indicators bring more information for operators and allow as well clearer specifications
for manufacturers.
Priorities concerning control loop performance indicators:
• It is necessary to select only a few indicators for the evaluation and verification of the
performance, like mean value deviation, integral squared error or standard deviation.
• Consideration of ranges would be very reasonable, e.g. a range of allowable deviations, a
range that may be exceeded for 20% of the time considered and a range that may be
exceeded for 5% of the time considered.
• Comparison of gradients (CGS) seems to be too complicated.
Causal research would be interesting.

331 - Final Report PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

331 - Final Report PDF

Hochgeladen von

Copyright:

Verfügbare Formate

Institute of Combustion and

Power Plant Technology

Definition and Verification of the Control Loop

Dipl.-Ing. Nataliya Knierim-Dietz

Research Project Supported by VGB Power Tech

1.1 Problem Formulation ........................................................................................ 1

1.2 Project Aims ..................................................................................................... 2

1.3 Structure of the Report ..................................................................................... 3

2 Theoretical Background ....................................................................................... 4

2.1 Considered Process Technology ...................................................................... 4

2.2 Definition of Control Loop Variables ................................................................. 6

2.3 Control Loop Performance Indicators and Application Methodology ................ 6

3 Control Loops Considered.................................................................................. 19

3.1 Live Steam Temperature Control Loop ........................................................... 19

3.2 Reheater Steam Temperature Control Loop................................................... 33

3.3 Power Output Control Loop ............................................................................ 33

3.4 Live Steam Pressure Control Loop ................................................................. 43

3.6 Feed Water Control Loop ............................................................................... 53

3.7 Feed Water Control in a Drum Boiler .............................................................. 53

3.8 Feed Water Control in a Once-Through Boiler ............................................... 58

4 Lifetime Consumption of Thick-Walled Components .......................................... 63

4.1 Creep Damage ............................................................................................... 63

4.2 Fatigue Damage ............................................................................................. 74

4.3 Influence of Steam Pressure and Steam Temperature Control on Lifetime

5 Benchmarks for Control Loop Performance ....................................................... 88

5.1 Introduction to Benchmarks ............................................................................ 88

5.2 The Development Scheme ............................................................................. 89

5.4 Live Steam Pressure Control Loop ................................................................. 94

5.5 Conclusion on Benchmark .............................................................................. 98

6 Results of Measurement Data Evaluation and Analysis ................................... 100

6.2 Classification of Cases for Measurement Data Evaluation ........................... 100

7 Normalization of Control Loop Performance Indicators .................................... 104

7.1 Normalization Method ................................................................................... 104

7.2 Application of Normalization Method ............................................................ 104

8.2 Power Plant Efficiency .................................................................................. 107

8.3 Part Load Capability of a Power Plant .......................................................... 108

8.4 Load Change Capability of a Power Plant .................................................... 109

8.5 Influence of Power Plant Unavailability on Economic Efficiency ................... 109

8.6 Power Output Control Loop Performance ..................................................... 111

9 Summary .......................................................................................................... 112

10 References ....................................................................................................... 114

Appendix A .............................................................................................................. 118

Appendix B .............................................................................................................. 120

ϑi superheater inlet temperature [°C]

1.1 Problem Formulation

1.2 Project Aims

a) Definition of control loop performance indicators depending on type and dynamical

Figure 1: Application framework

1.3 Structure of the Report

2.1 Considered Process Technology

Drum Boiler Once-Through Boiler

Natural-Circulation Boiler Forced Circulation Boiler Benson Boiler Sulzer Boiler

Figure 2: Types of boiler

2.1.1 Drum Boiler

2.1.2 Once-Through Boiler

2.1.2.1 Benson Boiler

2.1.2.2 Sulzer Boiler

2.1.3 Combined Cycle Power Plant

2.2 Definition of Control Loop Variables

wT(t) w(t) e(t) Feedback y(t)

Figure 3: Standard control loop

2.3 Control Loop Performance Indicators and Application Methodology

2.3.1 Measurement Data Evaluation by means of Moving Time Windows