definition and verification of the control loop performance for …report-p... · 2012. 11. 12. ·...

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 [%]

e(t) control deviation 1

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

Rm/2·105/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.

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].

Drum Boiler Once-Through Boiler

Natural-Circulation Boiler Forced Circulation Boiler Benson Boiler Sulzer Boiler

Boiler Types


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.

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.

ProcessFeedbackController

Disturbances z(t)

e(t) y(t)u(t)

Feedforward Controller

w(t)

uFF(t)

uFB(t)Pre-Filter

wT(t)


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.

Figure 4: Exemplary measurement data

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.

y(t)

, w(t)

t

y(t),

w(t)

y(t) Actual value w(t) Set point

0

O-T

t

y(t)

, w(t)

tt1 t1+T

t1 t1+T

a)

b)

y(t),

w(t) O-T

O-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%.

0

O-T

t

y(t)

, w(t)

tt2 t2+T

t2 t2+T

O-T

O-T

y(t),

w(t)

a)

b)


0

O-T

t

y(t)

, w(t)

t

O-T

t3 t3+T

t3 t3+T

O-T

y(t),

w(t)

a)

b)



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.

wMO −= ++ TT (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].

TT −− −= MwO (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).

TTT −+ += OOPTP (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.

00

50

100

Dis

tribu

tion

of O

-T [%

]

O-T

0 O

-T

t y

(t), w

(t)

t

O-T

Dis

tribu

tion

ofO

-T[%

]O

-T

a)

b)

c)

0 T

0 T

y(t),

w(t)



b) Positive overshoots O+T of exemplary measurement data

c) Percentage distribution of O+T in form of a histogram


b) Peak-to-peak values PTPT of exemplary measurement data

c) Percentage distribution of PTPT in form of a histogram

00

50

100

Dis

tribu

tion

of O

+T [%

]

O+T

0 O

+T t

y(t)

, w(t)

t

O+T

Dis

tribu

tion

of O

+T[%

]O

+T

a)

b)

c)

0 T

0 T

y(t),

w(t)

0

0

50

100

Dis

tribu

tion

of P

TPT [%

]

PTPT

0

PTP

T

t

y(t)

, w(t)

t

PTPT

Dis

tribu

tion

ofP

TPT

[%]

PTP

T

a)

b)

c)

0 T

0 T

y(t),

w(t)


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.

τττττ dywT

deT

MVDTt

t

Tt

t

))()((1)(1T −−=−= ∫∫

++

(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.

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.

τττττ dywT

deT

IAETt

t

Tt

t∫∫++

−== )()(1)(1T (2.5)

0

-e(t)

t

0

MVD

T

t

00

50

100

Dis

tribu

tion

of M

VDT [%

]

MVDT

MVDT

Dis

tribu

tion

of M

VD

T[%

]M

VD

T-e

(t)a)

b)

c)

0 T

0 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.

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.

∫∫++

−==Tt

t

Tt

t

dywT

deT

ISE τττττ ))²()((1)²(1T (2.6)

Figure 13 shows the determined ISET-values of exemplary measurement data and their percentage distribution.

0

|e(t)

|

t

0

IAE T

t

T

00

50

100

Dis

tribu

tion

of I

AE

T [%

]

IAET

IAET

Dis

tribu

tion

of IA

ET

[%]

IAE

T|e

(t)|

a)

b)

c)

0 T

0 T


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

nTnT ee

NSTD

1

2)(1 (2.7)

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 Te is the arithmetic mean value of the control deviation in the time frame [t, t+T] defined as:

∑=

=N

nnT e

Ne

1

1 (2.8)

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 Te ±STDT,

0

e²(

t)

t

0

ISE T

t

0

0

50

100

Dis

tribu

tion

of I

SET [

%]

ISETISET

Dis

tribu

tion

of IS

ET

[%]

ISE

Te²

(t)

a)

b)

c)

T

T

0

0


• 95.45 % of the measurement data are within the range of Te ±2STDT, • and 99.73 % of the measurement data are within the range of Te ±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:

%100T

kkT ⋅=

NNPD (2.9)

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.

Figure 15: Negative control deviation of exemplary measurement data and defined ranges

Te STDT-STDT 3STDT-3STDT -2STDT 2STDT

0

-e(t)

t

A

-A

B

-B

CDE

-E-D-C

-e(t)

0 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:

T

ttxtx

T

DD

N

nnn

N

nn ∑∑

==

∆−−== 22

)()( (2.10)

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:

Dis

tribu

tion

ofP

DT

[%]

Dis

tribu

tion

ofP

DT

[%]

Dis

tribu

tion

ofP

DT

[%]

Dis

tribu

tion

ofP

DT

[%]

Dis

tribu

tion

ofP

DT

[%]

0 10 20 30 40 50 60 70 80 90 1000

50

100D

istri

butio

n of

PD

T [%

]

PD[-A;A] T [%]0 10 20 30 40 50 60 70 80 90 100

0

50

100

Dis

tribu

tion

of P

DT

[%]

PD[-B;-A)&(A;B] T [%]0 10 20 30 40 50 60 70 80 90 100

0

50

100

Dis

tribu

tion

of P

DT

[%]

PD[-C;-B)&(B;C] T [%]

0 10 20 30 40 50 60 70 80 90 1000

50

100

Dis

tribu

tion

of P

DT

[%]

PD[-D;-C)&(C;D] T [%]0 10 20 30 40 50 60 70 80 90 100

0

50

100

Dis

tribu

tion

of P

DT

[%]

PD(-∞ ;-D)&(D;+∞) 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:

tttxtxtx nn

n ∆∆−−

=)()()(' (2.11)

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.

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:

x(t)

t

x(tn-Δt)x(tn)

tntn-Δt

Dn

T

NN-1

tT

0+

++ + +

_ _ _0

x'(t)

x(t)

t

x(tn-Δt)x(tn)

tntn-ΔtT

++

+ __ +

+ _0a)

b)


( ) ( )

T

xxNSD

N

nnn∑

=−−

= 21'sgn'sgn

21

(2.12)

in which:

<−

>+=

0'1

0'1:)'sgn(

n

nn

xif

xifx (2.13)

Figure 19: Calculation of NSD within two steps

x'n

n

0+

++ + +

_ _ _

x'(t)

tT

0

Step 1

Step 2

++ + +

_ _ _0

+


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.

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.

Steam mass flow

Attemperation spray water

Control valve

Superheater

Master controller

ϑI ϑO

ϑO, SPϑI, SPFollow-up controller

Disturbances of the flue gas heat

PG GeneratorI InletO OutletSP Set Point


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.

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.

0 50 100 150 200 250 300 350 400 450530

540

550

560LS

Tem

pera

ture

1 [°

C]

t [min]

Set PointActual Value

0 50 100 150 200 250 300 350 400 450-20

-10

0

10

LS T

empe

ratu

re C

D [K

]

t [min]

a)

b)

|


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.

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

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of O

+T [%

]

O+T [K]

Discretization 0.5 K

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of O

-T [%

]

O-T [K]


0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 300

20

40

60

80

100

Dis

tribu

tion

of P

TPT [%

] PTPT [K]


a)

b) c)


are concentrated in the first two ranges (range [-3; 3] and range [-5;-3)&(3;5]) and that there are a few steam temperature samples in further three bands

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

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

100

Dis

tribu

tion

of P

DT [%

]

PD[-3;3] T [%]

Discretization 5 %

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

100

Dis

tribu

tion

of P

DT [%

]

PD[-5;-3)&(3;5] T [%]

Discretization 5 %

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

100

Dis

tribu

tion

of P

DT [%

]

PD[-7;-5)&(5;7] T [%]

Discretization 5 %

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

100

Dis

tribu

tion

of P

DT [%

]

PD[-9;-7)&(7;9] T [%]

Discretization 5 %

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

100

Dis

tribu

tion

of P

DT [%

]

PD(-∞ ;-9)&(9;+∞ ) T [%]

Discretization 5 %


curve doesn’t cause any additional lifetime consumption of thick-walled components due to creep damage and leads to a slight efficiency loss.

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.

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60

20

40

60

80

100D

istri

butio

n of

MVD

T [%]

MVDT [K]


0 1 2 3 4 5 6 7 8 9 10 11 120

20

40

60

80

100

Dis

tribu

tion

of I

AE

T [%

]

IAET [K]


0 2 4 6 8 10 12 14 16 18 20 22 240

20

40

60

80

100

Dis

tribu

tion

of I

SET [

%]

ISET [K²]

Discretization 0.5 K²

a)

b) c)


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.

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.

Track 1

ϑC21

ϑC11

ATT21

ATT11

ϑ21

ϑ11

ϑC31

ATT31

ϑ31

ϑC22

ϑC12

ATT22

ATT12

ϑ22

ϑ12

ϑC32

ATT32

ϑ32

Track 2

1 2 30

2

4

6

8

10

ATT1 ATT2 ATT31 2 3

0

1

2

3

4

ATT1 ATT2 ATT3

Track 1Track 2

a) b)


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).

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).

NSDCV

NSDINFLOW

NSDϑI,SPNSDCD

NSDϑI


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.

Figure 31: NSD of attemperator control valve position and of steam temperature before attemperation

NSDCV

NSDINFLOW

NSDϑI,SPNSDCD

NSDϑI

Num

ber o

f Shi

fts in

Dire

ctio

n [1

/min

]

ATT1 ATT2 ATT30

1

2

3

4

Num

ber o

f Shi

fts in

Dire

ctio

n [1

/min

]

ATT1 ATT2 ATT30

1

2

3

4

Track 1Track 2

ATT1 ATT2 ATT1 ATT2 ATT3ATT3

a) b)


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.

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.

NSDCV

NSDINFLOW

NSDϑI,SPNSDCD

NSDϑI


Figure 33: NSDCV, NSDCD, NSDϑI and NSDϑI,SP.

3.1.6 Indicators for the Evaluation of Steam Temperature Control Loop Performance

In the previous chapters different control loop performance indicators were applied to an exemplary steam temperature measurement data. The control loop performance indicator mean value deviation MVDT is useful for the evaluation of steam temperature influence on the power plant efficiency and on the lifetime consumption of thick-walled components due to creep damage (see

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

Num

ber o

f Shi

fts in

Dire

ctio

n [1

/min

]

ATT1 ATT2 ATT30

1

2

3

4

Num

ber o

f Shi

fts in

Dire

ctio

n [1

/min

]

ATT1 ATT2 ATT30

1

2

3

4

Track 1Track 2

ATT1 ATT2 ATT1 ATT2 ATT3ATT3

Num

ber o

f Shi

fts in

Dire

ctio

n [1

/min

]ATT1 ATT2 ATT3

0

1

2

3

4

ATT1 ATT2 ATT3

Num

ber o

f Shi

fts in

Dire

ctio

n [1

/min

]

ATT1 ATT2 ATT30

1

2

3

4

ATT1 ATT2 ATT3

a) b)

c) d)

Importance●●●●● Very high

●●●● High

●●● Middle

●● Low

● Very low

ActuatorsControl Loop Performance Indicator Considered Effect

Covered regulating distance DNumber of shifts in direction NSD Wear of actuators


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).


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

Control Loop Performance Indicator RelevanceImportance

Full Load Operation

Part Load Operation

Low Load Operation

Load Change

Mean value deviation MVDT Efficiency, creep damage ●●●●● ●●●● ●●● ●●●●Distribution of deviations PDk Creep damage ●●●●● ●●●● ●●● ●●●●Positive overshoot O+TNegative overshoot O-TPeak-to-peak value PTPTIntegral of absolute error IAETIntegral of squared error ISET

General CLPIs ●●●● ●●● ●● ●●●

Peak-to-peak value PTP ≤ Allowable steam temperature peak-to-peak value PTPϑ,fatigue

Fatigue damage ● ● ● ●●●●


●●●● High

●●● Middle

●● Low

● Very low

ActuatorsControl Loop Performance Indicator Considered Effect



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:

Live Steam Temperature Control


Full Load Operation

Part Load Operation LC1 Low Load

Operation LC2

Mean value deviation MVDT Efficiency, creep damage ●●●●● ●●●● ●●●●Distribution of deviations PDkT Creep damage ●●●●● ●●●● ●●●●Positive overshoot O+TNegative overshoot O-TPeak-to-peak value PTPTIntegral of absolute error IAETIntegral of squared error ISET

General CLPIs ●●●● ●●● ●●●

Peak-to-peak value PTP ≤ Allowable steam temperature peak-to-peak value PTPϑ,fatigue

Fatigue damage ● ● ●●●●


●●●● High

●●● Middle

●● Low

● Very low ActuatorsControl Loop Performance Indicator Considered Effect


Abbrev. MeaningLC1 Load change from full load operation to part load operation and vice versa

LC2Load change from full load operation to low load operation and vice versaas well asLoad change from part load operation to low load operation and vice versa


• Fuel supply and steam generation. The input variable of this subprocess is the thermal output

FQ . The output variable is the generated steam mass flow ṁStG. The thermal output, which is necessary for the evaporation of the feed water and for the generation of the steam, is generated by the combustion of fossil fuels with the necessary combustion air. This subprocess includes further subprocesses such as fuel supply (including the mills), heat release, heat transfer, evaporation of the feed water and the reheating of the steam generated. The changes in the supply of the fuel mass flow affect the process of the live steam generation with a time delay. Therefore, the transfer function of the fuel supply and steam generation is represented as a high order time-delay element in Figure 34. This transfer function is characterized using the equivalent dead time Tu and the balancing time Tg. The values of Tu and of Tg depend on the steam generation system and on the type of fuel used.

• Steam storage. The input variables of this subprocess are the steam mass flow generated ṁStG and the turbine valve opening yT. The output variables are the turbine mass flow ṁT and the live steam pressure pSt. This subprocess represents the steam mass storage of the steam generator. The steam mass storage is modelled as an integrator and its volume is characterized by the storage time constant TS. The difference between the steam mass flow generated ṁStG and the turbine mass flow ṁT determines the steam pressure pSt. At the same time, the steam pressure pSt and the turbine valve opening yT determine the steam mass flow ṁT supplied to the turbine.

• Steam expansion, generation of electrical energy. During the process of steam expansion and generation of electrical energy the thermal power is converted into electrical power. This process includes components such as the high-pressure section of the turbine, the reheater, the intermediate-pressure and low-pressure sections of the turbine and the generator. The time response of the turbine and of the generator is negligible compared to the time response of the steam generator. The reheater and the connecting pipelines are an additional steam mass storage, which has the first-order delay effect, as shown in Figure 34.

Figure 34: Representation of the dynamic relationship in a steam power plant unit [40]

The 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.

Fuel supply and steam generation

Steam storage Steam expansion and generation of electrical energy



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 PG Controlled

variable In control Steam pressure

Turbine Constant set point (fixed-pressure operation) yT

Manipulated variable

Turbine Set point dependent on output (modified-sliding pressure operation)

yT

Steam generator Constant set point (initial-pressure operation) ṁF

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].

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 fuel Q Fset. In order to enhance the dynamic of the steam generator, the set point for the thermal output of the fuel Q Fset is pre-controlled by the fuel feed forward controller [40]. The fuel mass flow 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].

a) b)



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.


Activation of the immediate-reserve

capacity (primary control)

Ramp-like output change without utilisation of the immediate-reserve

capacity (secondary control)

Type of fuel Oil/gas Coal Oil/gas Coal

Test stimulus : output change, % Pnorm

±(2 to 5) ±(2 to 5) 20 to 30 20 to 30

Test conducted within percentage of output range, % Pnorm

50 to 90 50 to 90 40 to 90 40 to 90

Reference values for achievable rates of output change, % Pnorm /min

Fixed pressure

Sliding pressure

Fixed pressure

Sliding pressure

Fixed pressure

Sliding pressure

Fixed pressure

Sliding pressure

20 20 20 20 7 to 20 6 to 12

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.

Figure 37: Exemplary power output measurement data and their negative control deviation – data set 1

0 50 100 150 200 250 300 350 40096

97

98

99

100

Pow

er O

utpu

t [%

]

t [min]


0 50 100 150 200 250 300 350 400-2

-1

0

1

2

Pow

er O

utpu

t CD

[%]

t [min]

a)

b)

|


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.

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

0 50 100 150 200 250 300 35070

80

90

100

110

Pow

er O

utpu

t [%

]

t [min]


0 50 100 150 200 250 300 350-20

-10

0

10

Pow

er O

utpu

t CD

[%]

t [min]

a)

b)

|

0 1 2 3 4 5 6 7 8 9 100

20

40

60

80

100

Dis

tribu

tion

of O

+T [%

]

O+T [%]

Discretization 0.5 %

0 1 2 3 4 5 6 7 8 9 100

20

40

60

80

100

Dis

tribu

tion

of O

-T [%

]

O-T [%]


0 2 4 6 8 10 12 14 160

20

40

60

80

100

Dis

tribu

tion

of P

TPT [%

]

PTPT [%]


a)

b) c)


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).


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



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 %.

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60

20

40

60

80

100

Dis

tribu

tion

of M

VDT [%

]

MVDT [%]


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of S

TDT [%

]

STDT [%]


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of I

AE

T [%

]

IAET [%]


0 2 4 6 8 10 12 14 16 18 20 22 24 26 280

20

40

60

80

100

Dis

tribu

tion

of I

SET [

%]

ISET [%²]

Discretization 0.5 %²

a)

b)

c)

d)

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60

20

40

60

80

100

Dis

tribu

tion

of M

VDT [%

]

MVDT [%]


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of S

TDT [%

]

STDT [%]


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of I

AE

T [%

]

IAET [%]


0 2 4 6 8 10 12 14 16 18 20 22 24 26 280

20

40

60

80

100

Dis

tribu

tion

of I

SET [

%]

ISET [%²]


a)

b)

c)

d)


The determined STDT-values of the second data set are concentrated in the range from 2.5 to 3.5 % in 80 % of all evaluated time windows (see Figure 41-c). In the remaining time windows, the values of STDT are in the range from 0.5 to 2.5 %.

3.3.6 Indicators for the Evaluation of Power Output Control Loop Performance

All control loop performance indicators that are suitable for the evaluation of the power output control loop performance are summarized in Table 6. The control loop performance indicator mean value deviation MVDT is useful for the assessment of the power output control influence on the economic efficiency of a power plant unit. Control loop performance indicators positive overshoot O+T, negative overshoot O-T, peak-to-peak value PTPT, integral of absolute error IAET and integral of squared error ISET are general control loop performance indicators, which indicate the power output deviations.

Table 6: Indicators for the evaluation of the power output control loop performance

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 pSt

Controlled variable In control Steam pressure

Turbine Constant set point (fixed-pressure operation) ṁF

Manipulated variable

Turbine Set point dependent on output (modified-sliding pressure operation) ṁF

Steam generator Constant set point (initial-pressure operation) yT

Steam generator Uncontrolled (natural-sliding pressure operation) yT

Table 7: Assignment of the manipulated variables yT and ṁF to the controlled variables pSt

Power Output Control

Control Loop Performance Indicator Relevance

Mean value deviation MVDT Economic Efficiency

Positive overshoot O+TNegative overshoot O- TPeak-to-peak value PTPTIntegral of absolute error IAETIntegral of squared error ISETStandard deviation STDT

General CLPI


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 deviation, the pressure controller determines the set point for the thermal output of the fuel Q Fset. The thermal output of the fuel Q Fset is additionally pre-controlled by the fuel feed forward controller in order 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.

Figure 43: Exemplary LS3

pressure measurement data and their negative control deviation – data set 1

Figure 44: Exemplary LS pressure measurement data and their negative control deviation – data set 2

3 LS means Live Steam

0 50 100 150 200 250 300 350 40094

95

96

97

98

LS P

ress

ure

[%]

t [min]


0 50 100 150 200 250 300 350 400-1

0

1

2

3

LS P

ress

ure

CD

[%]

t [min]

a)

b)

|

0 50 100 150 200 250 300 35070

80

90

100

LS P

ress

ure

[%]

t [min]


0 50 100 150 200 250 300 350-4

-2

0

2

4

LS P

ress

ure

CD

[%]

t [min]

a)

b)

|


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.

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.

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of O

+T [%

]

O+T [%]


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of O

-T [%

]

O-T [%]


0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 300

20

40

60

80

100

Dis

tribu

tion

of P

TPT [%

]

PTPT [%]


a)

b) c)


In Figure 48-c the ISET-values are concentrated in the range from 0.5 to 1.5 %² in all considered time window. This means that there are no large live steam pressure deviations from the set point.


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).



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

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60

20

40

60

80

100

Dis

tribu

tion

of M

VDT [%

]

MVDT [%]


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of I

AE

T [%]

IAET [%]


0 2 4 6 8 10 12 14 16 18 20 22 24 26 280

20

40

60

80

100

Dis

tribu

tion

of I

SET [

%]

ISET [%²]


a)

b) c)




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.

Figure 49: Indicators for the evaluation of the live steam pressure control loop performance

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60

20

40

60

80

100

Dis

tribu

tion

of M

VDT [%

]

MVDT [%]


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

20

40

60

80

100

Dis

tribu

tion

of I

AE

T [%]

IAET [%]


0 2 4 6 8 10 12 14 16 18 20 22 24 26 280

20

40

60

80

100

Dis

tribu

tion

of I

SET [

%]

ISET [%²]


a)

b) c)

Live Steam Pressure Control


Full Load Operation

Part Load Operation

Low Load Operation

Load Change

Mean value deviation MVDTEfficiency, creep damage ●●●●● ●●●● ●●● ●●●●

Positive overshoot O+TNegative overshoot O- TPeak-to-peak value PTPTIntegral of absolute error IAETIntegral of squared error ISET

General CLPIs ●●●● ●●● ●● ●●●


●●●● High

●●● Middle

●● Low

● Very low


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.

Figure 50: Considered measurement data

Actual Value1Actual Value2

Pow

er O

utpu

t [%

]

Actual ValueSet Point

LS P

ress

ure

[%]

0 50 100 150 200 250 300 350 400 450

96

98

100

t [min]

Turb

ine

Valv

e [%

]

Turbine Valve 1Turbine Valve 2

Ther

mal

Out

put [

%]

0 50 100 150 200 250 300 350 400 45090

95

100

t [min]

0 50 100 150 200 250 300 350 400 4500

100

200

t [min]

0 50 100 150 200 250 300 350 400 45090

100

110

t [min]


Figure 51: Impact of the turbine valve opening on the live steam pressure

In order to compare the power output control loop performance with the live steam pressure control loop performance, the control deviations of the power output and of the live steam pressure are shown in Figure 52. These control deviations are represented in form of a scatter, in which every power output sample is represented with its corresponding live steam pressure sample. Thereby, four quadrants of the scatter can be distinguished:

• Quadrant I, in which the power output actual value is too large and the live steam pressure actual value is too low

• Quadrant II, in which the power output actual value is too low and the live steam pressure actual value is too low

• Quadrant III, in which the power output actual value is too low and the live steam pressure actual value is too large

• Quadrant IV, in which the power output actual value is too large and the live steam pressure actual value is too large

As shown in Figure 52, the number of samples in the quadrants II and III is considerably larger than the number of samples in quadrants I and IV. This is due to the fact that, the power output actual value considered is quite often below its set point (see Figure 37 and Figure 50).

In order to find out the ranges, in which most of the power output samples close to zero and most of the live steam pressure samples close to zero are concentrated,

• a scatter part with 70 % of the power output samples close to zero and • a scatter part with 70 % of the live steam pressure samples close to zero

are taken into account, as shown in Figure 53. The crossover of both of these parts defines the required range. Such ranges are defined for all quadrants of the scatter considered (see Figure 54). The results of this definition are shown in Table 8.

Live

Ste

am P

ress

ure

[%]

Turbine Control Valve [%]


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.

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

Power Output Control Deviation [%]

Live

Ste

am P

ress

ure

Con

trol D

evia

tion

[%]

Quadrant II:Power output too small

LS pressure too low

Quadrant III:Power output too small

LS pressure too high

Quadrant I:Power output too bigLS pressure too low

Quadrant IV:Power output too bigLS pressure too high


Live

Ste

am P

ress

ure

Con

trol D

evia

tion

[%] 70%

70%


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.


Live

Ste

am P

ress

ure

Con

trol D

evia

tion

[%]


LS pressure too low






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).

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.

0 200 400 600 800 1000460

465

470

475

480

485

t [min]

Wat

er L

evel

in H

P-D

rum

[mm

]



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.

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.

0…x % Maximum permissible range of drum level steady-state control deviation

x…y % Maximum permissible range of drum level transitory control deviation

Min. permissible water level 100%

Max. permissible water level 100%

Set Point 0%

Water Level [%]

t

100

100

0

yx

xy

100 200 300 400 500 600 700 800 900 1000-4

-2

0

2

4

Wat

er L

evel

[%]

t [min]

0 100 200 300 400 500 600 700 800 900 1000-100

-50

0

50

100

Wat

er L

evel

[%]

t [min]

O-=3.7%

O+=3%

a)

b)


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.

Drum Level Control

Control Loop PerformanceIndicator Considered Effect

Importance

Full Load Operation

Part Load Operation

Low Load Operation

Load Change

Positive Overshoot O+TDamage of boiler components and ofa 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 %


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.


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.

0 50 100 150 200 250 300 3502500

2600

2700

2800

Ent

halp

y [k

J/kg

]

t [min]


0 50 100 150 200 250 300 350-100

-50

0

50

Ent

halp

y C

D [k

J/kg

]

t [min]

a)

b)

|


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.

Figure 63: Indicators for the evaluation of the evaporator outlet temperature/enthalpy control loop performance

-20 -16 -12 -8 -4 0 4 8 12 16 200

20

40

60

80

100

Dis

tribu

tion

of M

VDT [%

]

MVDT [kJ/kg]

Discretization 1 kJ/kg

0 4 8 12 16 20 24 28 32 36 400

20

40

60

80

100

Dis

tribu

tion

of I

AE

T [%]

IAET [kJ/kg]


0 200 400 600 800 1000 1200 14000

20

40

60

80

100

Dis

tribu

tion

of I

SET [

%]

ISET [KJ²/kg²]


a)

b) c)

Control of the Evaporator Outlet Temperature/Enthalpy


Full Load Operation

Part Load Operation Load Change

Mean value deviation MVDT

A stable mass flow in the evaporator,Keeping of the attemperation control valves within the controlled ranges,Influence on the quality of the LS Pressure and PO Controls

●●●● ●●● ●●●

Positive overshoot O+TNegative overshoot O-TPeak-to-peak value PTPTIntegral of absolute error IAETIntegral of squared error ISET

General CLPIs,Disturbance for steam temperature control loops of following supeheaters

●●● ●● ●●


●●●● High

●●● Middle

●● Low

● Very low

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.

Figure 64: Creep curve [26]

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).

t

Rapture

σop=constϑop=const

Cre

ep s

train

ε

Tr

σop increasesand/or

ϑop increases

t

Cre

ep s

train

ε


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].

Figure 66: Lifetime consumption of thick-walled components due to creep damage

pve

dLms

icylop ⋅+= )

21

2(,σ (4.1)

pdeve

dimsAms

isphop ⋅+

+= )

21

)(4(

2

,σ (4.2)

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].

Figure 67: Diagram for the determination of TR

Creep Damage Fatigue Damage

Creep Rapture Strength

Operating Stress

Operating Steam Temperature

Operating Steam Pressure

Kind of a Material

Life Time Consumption

σop

TR

Řm/104

/ϑ

Řm/105

/ϑ

Řm/2∙105

/ϑ

Rm/T/ϑ

lg T in h

lgσ

in N

/mm

²

104 105 2·105

Řm/T/ϑ=0.8·Rm/T/ϑ


This intersection point can also be determined mathematically by following formulas [19]:

)8.08.0

lg(

)8.0

lg(5)(lg

/10/

/10/

/10/

1

5

4

5

ϑ

ϑ

ϑ

σ

m

m

m

m

R

RR

R

T

⋅⋅

⋅

+= (4.3)

)8.08.0

lg(

)8.0

lg(30103.05)(lg

/102/

/10/

/10/

2

5

4

5

ϑ

ϑ

ϑ

σ

⋅⋅⋅

⋅

+=

m

m

m

m

R

RR

R

T (4.4)

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]:

ϑ

ϑ

//

//,

pR

popkcreep T

TD = (4.5)

The total lifetime consumption of a thick-walled component due to creep damage Dcreep is calculated as the sum of partial lifetime consumptions [10]:

∑=k

kcreepcreep DD , (4.6)

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):

AVcreepSPcreepabs DDL ,, −=∆ (4.7)

%100%100,

, ⋅−=∆SPcreep

AVcreeprel D

DL (4.8)

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)).

deopdeop TT =→=ϑϑ (4.9)

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)).

deopdeaop TT >→<ϑϑ (4.10)

ϑ(t)

ϑop=ϑde

tϑde Design temperature

ϑop Operating temperature


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).

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)).

deop1 5.0...4.010 TTK ⋅=→=∆ϑ (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)

t

ϑdeϑaop

ϑde Design temperature

ϑaop Average operating temperature


Figure 70: Idealized operating steam temperature curve ϑaop > ϑde [25]

Referring to [25], due to a sine-shaped temperature curve, which has the amplitude ±10 K, the lifetime calculated at operating conditions Top is 10 to 20 % lower than the lifetime calculated at design conditions Tde, depending on the kind of material and on the temperature level (see (4.12)).

deopo 9.0...8.010 TTK ⋅=→=∆ϑ (4.12)

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.

ϑ(t)

ϑde

Δϑ1

ϑaop



ϑ(t) Δϑo

ϑaop=ϑde


ϑop Operating temperature



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.

Load Operating

Time Steam

Pressure

Steam Temperature [°C] 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.

Load Operating

Time Steam

Pressure Steam Temperature [°C]

[%] [hours] [bar] 595 596 597 598 599 600 601 602 603 604 605

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.

Figure 72: Idealized steam temperature curve

Load Operating

Time Steam

Pressure

Steam Temperature [°C]

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

0 5 10 15 20 25 30 35592

594

596

598

600

602

604

606

608

t [min]

Ste

am T

empe

ratu

re [°

C]



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).

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

0 20 40 60 80 100590

595

600

605

610

t [%]

Ste

am T

empe

ratu

re [°

C]


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%.

Load Operating

Time Operating

Time Steam

Pressure Steam

Temperature Lifetime

Consumption

[%]

Ranges of Temperature Deviations

Operating 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

∑Lifetime Consumption at 80%-Load Dcreep,80% [%] 0.73 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

∑Lifetime Consumption at 60%-Load Dcreep,60% [%] 0.05 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.

Load Operating

Time Operating

Time Steam

Pressure Steam

Temperature Lifetime

Consumption

[%]

Bands of Temperature Deviations

Operating 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±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±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

∑Lifetime Consumption at 60%-Load Dcreep,60% [%] 0.05 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].

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

wL

ms

msmpm p

ed ϑ

νβαασσσ ϑ

ϑϑ ∆−

+=+=12

(4.13)

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.

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

Creep Damage Fatigue Damage

Stress

Compressive Stress Thermal Stress

Steam pressure inside a thick-walled component

Temperature difference in the wall of a thick-walled

component

Physical properties of a material

Geometrical dimensions of a thick-walled component

Life Time Consumption

ϑi ΔϑPT1

ϑm

-T=220


applied to every succession of three material stress values σm1, σm2, σm3 using the following method: if the Boolean expression

(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].

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.

Figure 77: Detection of stress load cycles and deletion of corresponding load cycle values

σm(t)

t

σm(t)

t

t

σm2

σm3

σm4

σm5σm1

σm6

t

σm4

σm5σm1

σm6

σm(t) σm(t)

(σm1<σm2 and σm3<σm2) or (σm1>σm2 and σm3>σm2)

http://www.linguee.de/uebersetzung/deutsch/successively.html�


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:

322 mmvaf σσ −= (4.15)

2fa is to be calculated out of 2fva as follows:

• In the elastic range (2fva<2Rp0.2/ϑ)

( ) 2/2.0

2

2

3 )22(2)2(

22vapm

mvaa fRR

Rfff

−−⋅⋅=

ϑ

(4.16)

• In the plastic range (2fva>2Rp0.2/ϑ)

ϑ/2.0

2

3 2)2(

2p

vaa R

fff ⋅= (4.17)

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:

ki

kiFIk N

nD =∆ (4.19)


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:

∑∑ +∆=i k

RSEFkFiF DDD (4.20)

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

Lt 12.4x10-6 for 20...550°C [1/K]

Et 174000 for 550°C [N/mm²]

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).

Figure 79: Steam pressure and steam temperature

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.

0 200 400 600 800 1000 1200150

200

250

300

t [min]

LS P

ress

ure

[bar

]

0 200 400 600 800 1000 1200520

530

540

550

560

t [min]

LS T

empe

ratu

re [°

C]

a)

b)

0 200 400 600 800 1000 1200520

530

540

550

560

t [min]

LS T

empe

ratu

re [°

C]

0 200 400 600 800 1000 1200-10

-5

0

5

10

t [min]

Tem

pera

ture

Diff

eren

ce [K

]

0 200 400 600 800 1000 1200-50

0

50

100

t [min]

Ther

mal

Stre

ss [N

/mm

²]

a)

b)

18.6 K

13.3 K

83 N/mm²

c)


Figure 81: Steam pressure and assessed compressive stress

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).

0 200 400 600 800 1000 1200150

200

250

300

t [min]

LS P

ress

ure

[bar

]

0 200 400 600 800 1000 1200140

160

180

200

220

t [min]Com

pres

sive

Stre

ss [N

/mm

²]60.6 bar

50.1 N/mm²

a)

b)

0 200 400 600 800 1000 1200100

150

200

250

t [min]

Mat

eria

l Stre

ss [N

/mm

²]

0 200 400 600 800 1000 1200140

160

180

200

220

t [min]Com

pres

sive

Stre

ss [N

/mm

²]

0 200 400 600 800 1000 1200-50

0

50

100

t [min]

Ther

mal

Stre

ss [N

/mm

²]

a)

b)

c)


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.

0 200 400 600 800 1000 1200120

140

160

180

200

220

240

260

t [min]

Mat

eria

l Stre

ss [N

/mm

²]


Figure 84: Detection of material stress load cycles

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

1)

2)

3)

4)

5)

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

0 200 400 600 800 1000 1200100

200

300

Stre

ss [N

/mm

²]

t [min]

6)

7)

8)

9)

10)

11)


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).

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.

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

0 200 400 600 800 1000 1200120

140

160

180

200

220

240

260

t [min]

Mat

eria

l Stre

ss [N

/mm

²]

124 N/mm²

0 20 40 60 80 100 120 140 160 180 2000

50

100

150

200

250

300

350

400

Size of Material Stress Load Cycles [N/mm²]

Num

ber o

f Mat

eria

l Stre

ss L

oad

Cyc

les


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:

νβ

α

αϑ

ϑν

βαα

ϑϑν

βαασσ

σσσσσσ

σσ

σ

ϑϑ

ϑϑ

ϑϑ

ϑϑ

−

∆−≤∆

≤∆−

+∆

∆−∆−

+−=−

−−+=−

≤−

≤∆

1

2190

19012

)(1

)(2

190190

212121

221121

21

L

ms

msm

w

wL

ms

msm

wwL

ms

msmmm

ppmm

mm

m

pe

d

pe

d

ppe

d

(4.21)

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.

νβ

α

α

ϑϑ

ϑ

−

∆−≤

1

2190

,L

ms

msm

fatigue

pe

d

PTP (4.22)

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.

Figure 87: Benchmark Development Scheme

This scheme consists of two distinct areas, both of which are constructed in a similar way:

‘Benchmark World’Approximation

‘Real World’

No transfer

Benchmark ValueReal control loopperformance Comparison

i.e. Benchmarking

Simulation dataMeasurement data

• Power plant

• Controller

Limitations• Process model

• Benchmark controller

Optimal Conditions

1.)

2.)

3.)

1.)

2.)

3.)


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.

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.

Superheater Sensor

Flue gas heat qF

h*hI hO


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.

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’).

superheater

C1

C2

ϑO

ϑSP

xSystem

xε

qF

Follow-upcontroller

ϑIϑfeed

ϑI,SP


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.

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.

0 1000 2000 3000 4000 5000541

542

543

544

545

546

547

Time [s]

[°C

]

Live Steam Temperature

Set pointBenchmark controllerReal plant

0 1000 2000 3000 4000 5000450

460

470

480

490

500

510

Time [s]

Tem

pera

ture

[°C

]

Set point for follow-up controller

Benchmark controllerReal plant


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.

Best performance

Lessperformance

ISERP = 11,49

ISEBV = 4,70

ISE = 0


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.

mF pSt

mT

yT

mStG

TStG TSto


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.

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

0 5000 10000 15000

245

255

265Live steam pressure

time [s]

[bar

]

measurement datasimulation data

0 2000 4000 6000200

210

220

230Fuel rate

time [s]

[kg/

s]

measurement datasimulation data

Steam generationand storage

C1

C2

pStmFpSt,SP

xSystem

xε

mT


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.

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.

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

0 10000 20000 30000645

655

665

675

Supplied steam mass flow mT

[kg/

s]

time [s]

0 2000 4000 6000250

254

258

262Live steam pressure

Time [s]

[bar

]

Real plantBenchmark controllerSet point


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.

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.

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:

0 1000 2000 3000 4000 5000 6000-0.04

-0.02

0

0.02

0.04Gradient of measurement data

0 1000 2000 3000 4000 5000 6000-0.04

-0.02

0

0.02

0.04

time [s]

Gradient of simulation data (benchmark controller)

ISERP

ISEBV

Best performance

Lessperformance

ISE0 = 0


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).

h - hard coal fired power plant b - lignite fired power plant c - combined cycle

Nominal Power Output [MW]

≤300 301...600 601...900

I II III

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’.

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

0 50 100 150 20055

60

65

70

75

80

85

90

95

t [min]

Pow

er O

utpu

t [%

]


36 %



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.

qcmref

act

norm

qcmmean ref,

qcmqcm

IND

INDINDIND

σ−

= (7.1)

The meaning of the symbols that are used in formula (7.1) are described in the Table 23.

Symbol Meaning qcmnorm

IND normalized indicator of an indicator q for control loop c in operational mode m qcmact

IND actual value of an indicator q for control loop c in operational mode m qcm

meanref,IND average value of the reference indicator INDref of an indicator q for control loop c in

operational mode m qcm

refINDσ standard deviation of the reference indicator INDref of an indicator q for control loop c in

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).

Figure 101: Distribution of exemplary PTPT-values

Live Steam Temperature, T=60 min.

CLP Indicator Best

Theoretical Performance

Minimum Mean Maximum Standard Deviation Unit Best Practice

Performance Average Practice

Performance Worst Practice Performance

PTPT 0 3.81 7.89 22.56 3.26 [K] load full,T

normPTPIND -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.

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.

Figure 103: Measuring of performance by means of performance scale - example 1

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 300

20

40

60

80

100

Dis

tribu

tion

of P

TPT [%

]

PTPT [K]


Best TheoreticalPerformance

Best Practice Performance

Average Practice Performance

Worst Practice Performance

-2.42 -1.25 0 4.5





-2.42 -1.25 0 4.5

-2.0


If the normalized indicator of an exemplary PTPT-value amounts to -0.5 as shown in Figure 103, then the evaluated performance will be better than the average practice performance and worse than the best practice performance (see Figure 104).

Figure 104: Measuring of performance by means of performance scale - example 2





-2.42 -1.25 0 4.5

-0.5

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 the I&C system

State after the renewal of the I&C system

Improvement of 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.

Figure 108: Components causing unplanned unavailability of fossil-fired power plants in 2009 [41]

28.53 %

2.59 %

7.16 %

0.66 %

9.16 %

4.19 %3.09 %

2.18 %

1.85 %1.62 %

20.47 %

2.83 %1.72 %

1.65 %7.65 %1.5 %

1.13 %

1.23 %

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 %HD - Ash, slag and particulate removal 1.65 %HL - Combustion air system 1.5 %HB - Support structure,enclosure, steam generator interior 1.13 %H - Other components of conventional heat generation 7.65 %MA - Steam turbine 7.16 %MK - Generator 1.23 %M - Main machine sets (other components) 0.66 %LA - Feed water system 4.19 %LB - Steam system 3.09 %P - Cooling water systems 2.18 %E - Fuel supply and residues disposal 1.62 %C - Instrumentation and control equipment 1.85 %U - Structures 0.07 %A - Grid and distribution systems 2.59 %No KKS - Other 28.53 %


Figure 109: Components causing unplanned power plant unavailability of operator 1 in 2010

Figure 110: Components causing unplanned 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.

Boiler 19 %Boiler Tube Leaks 24 %Generator 10 %Steam Turbine 9 %Balance of Plant 3 %Pollution Control Equipment 1 %Transformer/Switchgear 9 %Gas Turbine 2 %Fuel Firing 2 %Other 21 %

19 %

9 %

21 %

9 %

10 %3 %

24 %

2 %

1 %

2 % Boiler 40.5 %Steam Turbine 29 %Generator 29 %Actuators 1.5 %

40.5 %

29 %

29 %

1.5 %

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"

http://upload.wikimedia.org/wikipedia/de/9/96/AUMA_Riester.svg�

http://www.hitachi-power.com/de/�

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.

definition and verification of the control loop performance for …report-p... · 2012. 11. 12. ·...

Documents