Institutionen för systemteknik

Department of Electrical Engineering

Examensarbete

Construction of a Simulator for the Siemens Gas

Turbine SGT-600
Examensarbete utfört i reglerteknik

av

Lisa Nordström

LITH-ISY-EX--05/3713--SE

Linköping 2005-11-02

TEKNISKA HÖGSKOLAN
LINKÖPINGS UNIVERSITET

Department of Electrical Engineer- Linköpings tekniska högskola

ing Institutionen för systemteknik
Linköping University 581 83 Linköping
S-581 83 Linköping, Sweden
Construction of a Simulator for the Siemens Gas Turbine SGT-600
Examensarbete utfört i reglerteknik
vid Linköpings tekniska högskola
av

Lisa Nordström
LITH-ISY-EX--05/3713--SE

Handledare: Markus Gerdin

Examinator: Torkel Glad ...
Linköping: 2005-11-02 …
Abstract
This thesis covers the development of a simulator for the Siemens Gas Tur-
bine SGT-600. An explanation on how a gas turbine works is also given, as
well as the principles behind the control system used by Siemens to control
the turbine.

For Siemens Industrial Turbomachinery to be able to test its control sys-

tem before delivering a gas turbine to the customer, a simulator is needed.
The control system needs to be adjusted for every unique gas turbine, since
there are several options for the customer to choose between when ordering
the turbine. A control system standard is under development, which also
needs to be tested in a simulator.

The framework for the simulator, i.e. the hardware and software that
form the simulator system, was predefined to suit this specific purpose. The
Siemens software SIMIT is used for developing the model. SIMIT is a real
time simulation tool where models are constructed using blocks, similar to
MATLAB Simulink.

A gas turbine is basically a heat engine that produces mechanical energy

or electricity. The main task of the control system is to control the fuel flow
to the combustion chamber and by that keeping the machine at desired
speed.

The gas turbine model was developed using measurement data from a
site in Hungary, where a gas turbine of the type SGT-600 is in service. The
model is based on simplified relations between the signals. By analyzing
measurement data and learning about the functionality of a gas turbine it
was found out that the speed of the gas generator affected most other sig-
nals, like temperatures and pressures. The gas generator speed was found to
be dependent on the heat flow, which is determined by the openings of the
gas control valves.

As a result of this thesis a working simulator for the gas turbine SGT-600
has been developed. The simulator can be used for testing the control sys-
tem standard and for testing the control system when adapting it to a spe-
cific delivery. It is also suitable for educational purposes, for example to
instruct customers.
Preface
This thesis was written at the department of Controls Engeineering at Sie-
mens Industrial Turbomachinery in Finspång, as a final thesis to complete
the studies for Master of Science in applied physics and electrical engineer-
ing international, at Linköping University.

Thanks to Tomas Strömberg, instructor at Siemens, Markus Gerdin, in-

structor at the university, Torkel Glad, examiner, Maja Johansson, oppo-
nent, Emil Haraldsson, linguistic inspector and moral support and to An-
dreas Lindholm and Stefan Gunnarsson for invaluable help.
Table of contents

1 Introduction.................................................................1
1.1 Aim.................................................................................. 1
1.2 Background..................................................................... 1
1.3 Problems to be solved .................................................... 1
1.4 Scope.............................................................................. 2
1.5 Structure of the thesis..................................................... 2

2 How a Gas Turbine Works..........................................4

2.1 Overview......................................................................... 4
2.2 Compressor .................................................................... 5
2.3 Combustion chamber and ignition .................................. 8
2.4 Compressor Turbine and Power Turbine ..................... 10
2.5 Starting System ............................................................ 12
2.6 SGT-600 ....................................................................... 13

3 The Control System ..................................................14

3.1 Configuration overview ................................................. 14
3.2 PLC............................................................................... 14
3.3 PCS 7 Software ............................................................ 15
3.4 PROFIBUS ................................................................... 16
3.5 DP-slaves ..................................................................... 16
3.6 Fail-Safe ....................................................................... 17
3.7 The Control Program .................................................... 17

4 Simulator Setup ........................................................21

4.1 Overview....................................................................... 21
4.2 SIMIT ............................................................................ 22
4.3 Simba pro PCI card ...................................................... 22
4.4 Discussion on the chosen solution ............................... 22
5 Construction .............................................................24
5.1 Selection of model type ................................................ 24
5.2 Extraction and analysis of measurement data.............. 24
5.3 Building the model in SIMIT.......................................... 26
5.4 Communication between the simulator and the control
program ................................................................................. 28

6 Testing.......................................................................31
6.1 WinCC........................................................................... 31
6.2 Sequences .................................................................... 31
6.3 Testing the simulator .................................................... 33

7 Conclusion ................................................................36
7.1 Result............................................................................ 36
7.2 Discussion..................................................................... 37
7.3 Further Development .................................................... 38
References............................................................................ 39

Figure- and chart-index

Figure 1………………………………………………………………….…..4
Figure 2………………………………………………………………….…..5
Figure 3………………………………………………………………….…..6
Figure 4………………………………………………………………….…..7
Figure 5………………………………………………………………….…..8
Figure 6………………………………………………………………….…..9
Figure 7………………………………………………………………….…..9
Figure 8…………………………………………………………………….10
Figure 9……………………………………………………………….…....11
Figure 10......…………………………………………………………….…12
Figure 11...…...………………………………………………………….…14
Figure 12...…..………………………………………………………….….15
Figure 13...……………...……………………………………………….…16
Figure 14...………..…………………………………………………….….18
Figure 15...…..………………………………………………………….….21
Figure 16...……………………………………..……………………….….26
Figure 17...………………………………………………………..…….….27
Figure 18...……………………………………………..……………….….29
Figure 19...……………………………………………..……………….….33
1 Introduction
This chapter will explain the purpose of this thesis and the reason why it
was initiated. The scope will be concretized by declaring what is not cov-
ered and the structure will be commented on to provide guidance to the
reader.

1.1 Aim
The purpose of this master thesis is to develop a simulator for the gas tur-
bine SGT-600. The framework for the simulator has been fixed in two ear-
lier final theses, i.e. the hardware and software has been tested and evalu-
ated and adjusted for the application.

The simulator is supposed to be used for testing the control system before
the delivery of a gas turbine. This will shorten developing time and reduce
installing time at site, thus reducing costs. A control system standard is un-
der development, which also needs to be tested in a simulator. The simulator
can also be used for educational purposes, like training of new personnel
and demonstration for the customer.

It should be possible to start the simulator from the operator station so that it
automatically runs through the start sequences and reaches normal running
state. From there different running states should be possible to simulate.
Ideally the operator should be able to switch between the simulator and a
real process and not notice any difference.

1.2 Background
The Finspång site, developing and producing gas and steam turbines, has
since 2003 been owned by Siemens and is since 2004 called Siemens Indus-
trial Turbomachinery AB.

Siemens is currently migrating from ABB’s control system to Siemens

own, called S7. For every new gas turbine that Siemens sells, the control
system has to be adapted to the specific options included in the delivery.
The control system needs to be tested against a simulator. Previously a
rough model in the ABB system ADVANT had been used. Since the control
system is being converted to S7, there was a need for a new simulator. In
two earlier final theses a simulator for the gas turbine SGT-700 has been
developed. Emphasis was on bringing forth the most suitable simulator
setup. Siemens also needed a simulator for another gas turbine type, SGT-
600, which is a predecessor to the gas turbine SGT-700.

1.3 Problems to be solved

The task to develop a simulator for the gas turbine SGT-600 can be di-
vided into the following subtasks:
• Select an appropriate model type

• Find and analyze relevant measurement data.

1
 • Acquire knowledge of the principles behind how a gas
turbine works and how it is controlled

• Decide parameter values to model the relations be-

tween the signals

• Get the simulator setup system to work with the new

control program for SGT-600

• Implement the model in SIMIT

• Test the model and evaluate the results

1.4 Scope
The gas turbine model should be of the type SGT-600, although it is desir-
able to improve the existing model of gas turbine type SGT-700.

The gas turbine model should be of Mechanical Drive type, and not
Power Generation type. This was the natural choice since the control system
code developed for the SGT-600 type is of Mechanical Drive type.

The simulator setup (the hardware and software that is used in the simu-
lator system) should be the same as in the previous final theses. This is be-
cause the time frame given would not allow time to be spent on testing new
software and hardware. The previous final theses writers had already gone
through several alternative solutions before deciding on the chosen one.

The compressor load does not have to be implemented in this thesis. The
part of the control program that controls the compressor is loaded into a
separate PLC. Since only two PLC:s were available for this thesis, the com-
pressor part of the control program will not be used. For that reason and due
to time reasons the modeling of the compressor had to be left out. An addi-
tional reason for not modeling the compressor was that another final thesis
covering the compressor was being planed.

The model will be of a gas turbine running on gas fuel only. It was desir-
able to model dual fuel (gas and liquid fuel), but since the control system
code for the liquid fuel subsystem was not yet developed, there was no point
in including it in the model, since it would not have been possible to test.

1.5 Structure of the thesis

This thesis has not been strictly divided into one theoretical part and one
implementation part. Instead the theory is provided where it is needed. This
was done to make the thesis more readable and interesting.

First the basic functionality of a gas turbine is explained, to give the

reader an idea of what is to be simulated. Then the control system and the
simulator system are explained, together with the hardware and software
used. This is followed by a description of the construction process and an

2
illustration of the testing of the model. The thesis is concluded with a dis-
cussion about the results and suggestions on further development.

3
2 How a Gas Turbine Works
This chapter will give a basic understanding of how a gas turbine works.
The theory is divided according to the main parts of the gas turbine. Some
specifics on the SGT-600 type are added.

2.1 Overview
A gas turbine is a heat engine that converts chemical energy from the fuel
into heat energy, which is converted into mechanical energy. The efficiency
is between 25 and 45 percent. [1]

The main parts of the gas turbine are the gas generator and the power
turbine. The gas generator consists of compressor, combustion chamber and
compressor turbine (see figure 1 and 2). The purpose of the gas generator is
to generate a flow of pressurized hot gas, driving the power turbine. The
purpose of the power turbine is to convert the pressurized hot gas flow from
the gas generator to mechanical energy, driving a load. [2]

Figure 1 Gas Turbine SGT-600; Bleed Valve 1 (BV1), Bleed Valve 2 (BV2), Compressor,
Combustion Chamber, Compressor Turbine (CT), Power Turbine (PT)

Compressor, combustion chamber and turbine are encapsulated in cylin-

drical casing, where the flow of air and gas is moving straight through. The
compressor compresses air for the combustion chamber, where fuel is mixed
with air and combusted. The hot gases pass through the compressor turbine,
which is driving the gas generator and then expand through the power tur-
bine. In an open gas turbine cycle, which is the completely dominating
form, the air and combustion gases are let out in the atmosphere, whereas in
a closed gas turbine cycle the gas is cooled and led back to the compressor.
[1]

4
Figure 2 Overview of the gas turbine

If the gas turbine is to be used for power generation (PG), there is an AC

generator connected to the power turbine. If it is to be used for mechanical
drive (MD) there will usually be a compressor connected to the power tur-
bine, which is used for compressing natural gas in a pipeline. Compressors
are used at certain intervals in the pipeline to keep the gas flowing at a
steady rate. Mechanical drive applications also include pumping up oil and
driving ferries.

2.2 Compressor
The purpose of the compressor is to compress air for the combustion. An
axial flow compressor consists of one or more rotor assemblies that are
mounted between bearings in the casing. The compressor is a multi-stage
unit, where the pressure is increased by each stage. Each stage consists of
one vertical layer of rotating blades (see figure 3) and one of stator vanes.
The stator vanes decrease the air velocity, increasing the pressure. They also
aim the airflow at a correct angle to the next section of rotor blades. Seal-
ings between the stages prevent the air from leaking. From the front to the
rear the cross section area of the airflow is decreasing, so that the axial ve-
locity remains constant as the volume decreases due to the compression. [1]

5
Figure 3 Compressor

Ambient air is taken through an inlet duct and passes a filter before
reaching the compressor. The filter is necessary to prevent objects to enter
the compressor and to minimize erosion and corrosion. The airflow to the
compressor is controlled by two variable vane stages at the inlet of the com-
pressor, called the inlet guide vane. The SGT-600 gas turbine has ten stages,
divided into three parts: two low pressure sections consisting of the first five
stages and a high pressure section consisting of the last five stages. [1]

The compressor has two cavities, one low pressure cavity between stage
two and three and a high pressure cavity between stage five and six. The
cavities are connected to bleed valves in the casing, the low pressure cavity
to bleed valve one, which is a binary valve and the high pressure cavity to
bleed valve two, which is controllable (see figure 4). [1]

6
Figure 4 Bleed valves

The bleed valves are, together with the variable guide vanes, used for
preventing stall and surging. Surging is a phenomenon that can occur during
start and stop, when the volume air flow is low. The decrease of the cross
section area of the flow in the compressor is optimized for full load (full
speed), while at low speed the smaller compression ratio would actually re-
quire a less reduction of the cross section area. If the compressor rotor ac-
celerates too quickly, the airflow velocity will be too small in relation to the
blade velocity. Since the pressure rise does not correspond to the decreasing
volume the small rear end of the compressor will not be able to complete the
compression. This leads to a reverse airflow, which causes oscillations in
the compressor. Surging is when the airflow through the whole compressor
is broken down, stall when only some stages are affected. The oscillation
can damage the compressor since it creates stress on the blades. [1]

The bleed valves open during start and stop to bypass some of the air to
avoid surging. By controlling the variable guide vanes the airflow to the rear
stages of the compressor can be decreased. At cooling down the bleed
valves are closed so that all air is used for cooling. [1]

Air from the low pressure cavity is used as seal air to prevent oil leakage
in bearing 1, 3 and 4 and air from the high pressure cavity is used for cool-
ing the power turbine discs. [1]

7
2.3 Combustion chamber and ignition
In the combustion chamber the fuel injected through the burners is burnt
with air supplied by the compressor. The combustion chamber is of annular
form with 18 burners in the front end (see figure 5 and 6). The burners are
connected to a section of the turbine inlet. The forward end of the burning
chamber casing is connected to the compressor via a diffuser. The air from
the compressor has a velocity of about 100 meters per second. The airflow
must therefore be decelerated in order not to blow out the flame. In the
combustion chamber a region of low axial velocity and re-circulating flow
has to be created. Efficient combustion is necessary to obtain high thermal
efficiency and to minimize the exhaust gas emissions. Flame temperature
must be 1000-2000 C for efficient combustion. The combustor walls have to
be cooled, since they cannot stand the high temperatures. Out of the airflow
from the compressor only about 25% is supplied to the combustion zone at
full load, the rest is used for cooling the combustor walls and to dilute the
hot gases to a temperature low enough not to damage the turbine parts. [1]

Figure 5 Combustion chamber with burner and flame detector

8
Figure 6 combustion chamber

During the start procedure, gas is led through a small pipe to an ignition
burner, air is let in by opening of ignition air valves and an igniter ignites
the mix of air and gas. Then gas is led through another small pipe and enters
at the cone tip of burner six and a pilot flame is ignited. After that the main
and the primary gas control valves start to open, letting gas from a larger
pipe enter all the burners, igniting the main flame (See Figure 8). The pri-
mary gas control valve opens first and is used for controlling the amount of
gas until a certain load is reached. Then the main valve starts to open as the
primary valve starts to close, resulting in an even flame. The reason why the
gas supply to the combustion chamber is controlled by two valves is that it
gives the most stable flame form. Gas from the primary valve enters the
burners at the cone tip, resulting in a long thin stable flame, while gas from
the main valve enters at the cone sides, resulting in a wide unstable flame
(see figure 7).

Figure 7 Double cone burner

9
Figure 8 Operator Picture of the Fuel System showing the gas paths for the Ignition, Pilot
and Main Flames

There are two flame detectors in the combustion chamber, one to detect
the ignition and pilot flames during start and one to detect the main flame
during start and normal operation.

There is a variable dilution system that decreases the airflow through the
burners by providing a controlled bypass of air to the combustor exit. The
purpose is to ensure complete combustion and therefore minimize the CO
emissions and increase the efficiency on part load. Some of the air is by-
passed from around the combustion chamber into an inner manifold, passed
through six control valves to an outer manifold and fed back into the end of
the combustion chamber. The valves are connected by a drive ring con-
nected to a motor. Without the system, flame temperature would drop with
increasing load, since the air-fuel ratio increases. A too low flame tempera-
ture results in an incomplete combustion with formation of CO. [1]

2.4 Compressor Turbine and Power Turbine

The two-stage compressor turbine provides power to drive the compressor
and the power turbine gives useful mechanical output, driving a load. The
turbines extract energy from the hot gases from the combustion chamber by
expanding the gas to lower pressure and temperature. [1]

Each turbine stage consists of a row of stationary guide vanes, mounted

to the turbine casing, followed by a row of moving blades, fitted to turbine
discs (see Figure 9 and 10). Hot gas is expanded in the convergent passages
between the guide vanes. Pressure energy is converted into kinetic energy
and the gas is accelerated. When the gas reaches the turbine blades it is
given a spin/swirl in the direction of the blades. The gas is forced to deflect
and is further expanded, since the passages are convergent. During this
process energy is absorbed, causing the turbine to rotate and provide power

10
for driving the turbine shaft. Since gas expansion, in contrast to compres-
sion, is a spontaneous process, fewer stages are needed to expand the gas to
atmospheric pressure. [1]

Figure 9 Compressor turbine and power turbine, showing guide vanes and turbine blades

Sealings prevent gas leakage between the stages and to shafts and bear-
ings. Sealing air bled off from the compressor is lead off along the turbine
discs, to cool them and prevent heat transfer to shafts and bearings. [1]

The gas generator rotor and the power turbine are carried by bearings.
Oil is continuously supplied to the bearings during operation. [1]

In full operation the speed of the power turbine is about 7700 rpm. For
power generation the AC-generator is connected via a gearbox to reduce the
generator speed to 1500 rpm. In a two-shafted gas turbine, like the SGT-
600, the power turbine is not mechanically interconnected to the gas genera-
tor. The speed of the gas generator is determined by the actual power de-
mand in combination with ambient conditions, such as temperature and hu-
midity. This allows a wider operating range in comparison with a single-
shafted model. [1]

11
Figure 10 Power Turbine

2.5 Starting System

The compressor is set in motion by a start motor, connected to the compres-
sor through a coupling. The starting system speeds up the gas generator to a
speed necessary for purging (ventilation of the gas turbine and the exhaust
system to get rid of uncombusted gas) and ignition. That speed is approxi-
mately 2300 rpm. After the purging the main fuel supply is opened to the
combustion chamber and the gas turbine is ignited. The starting motor helps
the gas turbine accelerate beyond self-sustaining speed (approximately 5400
rpm). As it reaches self sustaining speed the clutch disengages and the start-
ing motor is switched off. [1]

When the gas turbine is to be stopped the load is decreased slowly. When
it is reduced to about 400 kW the fuel supply is closed and the rotors will
coast down. The starting system is activated and given a speed reference of
approximately 230 rpm. When the gas generator reaches that speed the
clutch will engage and the gas generator will be driven by the starting motor
until it has cooled down, which takes ten hours. In case of power failure
emergency batteries will drive the starting motor. The reason why the tur-
bine has to rotate during cool down is that the rotors, which are supported
by the bearings only at each end, will deform if they are brought to stand
still while they are still hot. [1]

12
2.6 SGT-600
SGT-600, is a compact, low-weight heavy-duty industrial gas turbine with
an electrical output of 24.8 MW. [3]

It is suitable for both mechanical drive applications and power genera-

tion. The applications also include combined cycle operation, cogeneration
and marine propulsion. Combined cycle operation is when the gas turbine is
connected to a steam turbine, where the waste gases from the gas turbine are
used to heat a boiler that provides the steam turbine with steam. Cogenera-
tion means that the gas turbine is equipped with a waste heat recovery unit
(a boiler fan). The waste heat can be used for production of industrial proc-
ess steam or for district heating for example. [3]

Siemens Industrial Turbomachinery in Finspång sells about 60 Gas Tur-

bines every year. About half of them are of the type SGT-600. The reason
for the high percentage of this type is that it is suitable for driving a com-
pressor (mechanical drive), which is the most common application on the
market.

13
3 The Control System
This chapter will give a brief explanation of the hardware and software used
by Siemens to control a gas turbine and provide understanding of how the
control system works.

3.1 Configuration overview

The gas turbine process is controlled by the control program, which is lo-
cated in CPU:s that are part of programmable logic controllers (PLC:s). The
PLC:s are connected with PROFIBUS to I/O modules that send and receive
signals from sensors and to actuators on the gas turbine. The PLC:s are also
connected to an operator station (a PC station with human-machine inter-
face) via industrial Ethernet. From the operator station the process is oper-
ated and the different subsystems are monitored. (See figure 11 for an over-
view of the configuration.)

Figure 11 Overview of the gas turbine setup

The automation system (control system) registers the process variables,

processes the data according to the instructions in the user program, issues
control instructions and set points to the process, supplies the operator sta-
tion with the data for visualization and registers actions on the operator sta-
tion and forwards them to the process. [4]

3.2 PLC
PLC:s (Programmable Logic Controllers) are computers used for controlling
the process. The control program code is loaded into the CPU: s of the
PLC:s.

The PLC:s used are called S7-400. The most important components of
the S7-400 are racks, power supply modules (PS), Central Processing Units
(CPU:s), memory cards and Communication Processor (CP) for Ethernet
and PROFIBUS interface. Racks provide mechanical and electrical connec-
tions between the S7-modules. Power Supply Modules convert the line volt-
age (120/230 VAC or 24 VDC) to the 5 VDC and 24 VDC operating volt-
ages required to power the S7-400. CPU:s execute the user program. Mem-
ory cards store the user program and parameters. Communication processors
enable data exchange between programmable controllers and/or computers
by means of point-to-point connection. [5]

14
3.3 PCS 7 Software
PCS 7 (Process Control System) is a software suite used for creating the
control program, configuring the hardware of the system and creating the
user interface (see chapter 6.1 for information about the part that is used for
creating operator pictures).

PCS 7 projects are created with the Engineering System (ES). The Engi-
neering System includes the applications SIMATIC manager, HW Config,
CFC and SFC Editor.

SIMATIC Manager is the central application that provides access to all

other applications that are used to create and modify a PCS 7 project. From
the SIMATIC manager HW Config and the CFC and SFC charts are
reached for example. The PLC:s are also managed from here. The operating
mode can be changed, settings adjusted and the events can be reviewed in
the diagnostic buffer for example. [6]

Hardware Config is the configuration of the hardware of a system. Con-

figuration refers to the arranging of racks, modules and distributed I/O racks
in a station window (see figure 12). Racks are represented by tables where a
specific number of modules can be inserted. Parameters for the modules are
set, downloaded to the CPU and transferred to the respective modules. [7]

Figure 12 Hardware Config

CFC (Continuous Function Chart) is a graphic editor used to create the

software structure of the CPU. Ready-made blocks are dragged to function
charts, parameters are assigned to them and they are interconnected (see
figure 13). When the required functions are created the code is compiled and
downloaded to the CPU:s. [8]

15
Figure 13 Example of a CFC chart from the control program

SFC (Sequential Function Chart) allows graphic configuration of sequen-

tial control systems. The functions created with CFC are controlled by oper-
ating and state changes and executed selectively. In a sequential control sys-
tem the process is broken down to consecutive steps. Each step includes ac-
tions to be executed. Step transitions between the steps include conditions
that have to be fulfilled to enable passing of control from one step to another
(see figure 19). [9]

3.4 PROFIBUS
Conventional signal transmission between sensors/actuators in the field and
input/output modules of the control system is implemented via parallel
point-to-point connections with copper cables. Fieldbus systems permit digi-
tal communication between the automation system and the field devices on
a single serial bus cable. This results in large savings on installation costs
due to reduction in cabling and input/output hardware and a significantly
greater amount of information can be transferred. [10]

PROFIBUS (process field bus) is the world leader among fieldbuses. It

can be used in all sectors of the production and process industries for high-
speed communication with high measurement accuracy. [10]

PROFIBUS with DP system (Decentralized Peripherals) permits fast

communication with intelligent, distributed I/O devices. It provides high
data transmission rates and short response times. [10]

3.5 DP-slaves
DP-slaves are hardware process devices (seen in figure 11) that handle the
communication between the PLC:s and the sensors, actuators and other
types of measurement points in the plant. They are called DP-slaves because
they use the PROFIBUS-DP protocol to communicate with one or two mas-
ters (PLC:s). The DP-slaves are connected via PROFIBUS to PLC:s. Each
slave holds a variable number of I/O:s. [11]

16
3.6 Fail-Safe
Fail-safe automation systems are used when a fault code could endanger
human life, damage the plant or the environment. They detect errors in the
process and automatically bring the plant to a safe state when a fault occurs.
The safety mechanisms include for example that the configured safety func-
tions are processed twice in different processor sections of the CPU and er-
rors detected in a comparison of the results. Another example is that pro-
gramming errors such as division by zero and value overflow are intercepted
by special fail-safe CFC blocks. [15]

The current control program is partly in a fail-safe configuration. This af-

fects the construction of the simulator in the way that the software used
must support fail-safe, which will be discussed later in this thesis.

3.7 The Control Program

The control program is divided into three parts: Main (including the Turbine
Governor), Protection (failsafe configuration of critical signals) and Com-
pressor (controls the load).

The turbine governor controls the fuel flow to the gas turbine so that the
machine is kept at desired speed and does not run in forbidden operating
areas and so that flame-out is avoided. It further controls the split between
the primary and the main gas valves and it also controls bleed valve 2, the
position of the inlet guide vanes and the combustion chamber bypass. [2]

There are eleven control functions involved in controlling the amount of

fuel fed to the gas turbine (STC, NGGL, SC, MPC, GDC, T7L, GAC, LLD,
PAC, MPPRC and MMPRC). STC and NGGL are used at start, SC is used
at normal operation and the rest are used as limit functions. They all give a
desired heat flow to the combustion chamber as output. The main inputs to
the controllers are the measured speed of the gas generator and of the power
turbine. There are two speed pickups for gas generator speed measurement
and two for power turbine. The pickups are measuring the presence of cogs
on a cogwheel on the turbine rotor. If over speed is sensed at one of the two
pickups, the turbine is tripped. (For more information about trips see the last
passage of this chapter). The maximum value from the two pickups is used
in the governor. [2]

Only one controller is in operation at a time. A minimum selector selects

the one of the control functions (STC, MPC, FLC, GAC, T7L, NGGL, LLD,
PAC, MPPRC and MMPRC) that has the lowest desired heat flow as output.
The output from that controller and the output from GDC are then compared
in a maximum selector, so that the GDC will be selected to be in charge if it
requires a higher heat flow than the previously selected controller. That
means that the fuel flow is controlled by the channel requiring the smallest
valve opening, except if the resulting fuel flow would not be enough to keep
the flame alive. In that case the GDC will be in charge (see figure 14). [2]

17
Figure 14 Turbine Governor as shown in the operator picture in WinCC, screen shot from
startup

The starting control (STC) keeps the acceleration of the gas generator at
a limited rate during startup, thereby preventing thermal stress of the tur-
bine. Before ignition, the engine is running at purge speed (2300 rpm). It is
ramped to 2700 rpm by the start motor. A “start kick” adds some extra fuel
during ignition, to get ignition to all burners. The fuel flow is during start
corrected for ambient temperature, i.e. increasing temperature reduces the
amount of fuel. The fuel flow starts to increase with “STC ramp rate 1” and
the start motor is released at 5000 rpm. The fuel ramp is changed to “STC
ramp rate 2”. If the exhaust temperature reaches a certain point, the fuel
flow is kept constant until the temperature has decreased. At 5600 rpm the
gas generator speed limiter, NGGL, takes over the fuel control. [12]

The gas generator speed limiter (NGGL) takes over from STC at 5600
rpm and stays in controls until the power turbine reaches minimal speed and
SC takes over. NGGL also controls that the maximum allowed speed of the
gas generator will not be exceeded during operation, thus preventing the en-
gine from severe damage. [2]

The speed controller (SC) is used for mechanical drive applications

(whereas for power generation the frequency and load controller (FLC) is
used instead). SC takes over from NGGL shortly before minimal speed of
the power turbine and operates until full load is reached and T7L or NGGL
takes over. The set point of the controller (the speed of the gas generator) is
set by the operator or by the compressor performance controllers.

18
 The maximum servo position control (MPC) is not used during normal
operation. It is used for the operator to manually limit the maximum amount
of fuel (in MJ/s) fed to the combustion chambers. It is also used as backup
control if feedback error occurs. Then the actual desired heat flow becomes
the set point for the MPC controller. [2]

The gas generator deceleration control (GDC) keeps the flame alive, by
keeping the amount of fuel at minimum demand. This is very important if
load rejections occur and call for fuel below the limit to sustain flame. At
load rejection all valves are taken to a minimum, but the main gas control
valve is brought to a level that sustains the flame. The GDC set point is de-
pendent on the actual normalized gas generator speed. [2]

The exhaust temperature limiter (T7L) limits the exhaust temperature and
therefore the maximal load, since the exhaust temperature increases with
increasing load. The set point is a function of ambient air temperature, am-
bient humidity, compressor delivery pressure, exhaust gas pressure and
compressor inlet pressure. [2]

The gas generator acceleration control (GAC) limits the acceleration of

the gas generator (the derivative of the gas generator speed) during loading
and thus preventing the turbine from surging and from transient over tem-
peratures in the gas generator. [2]

The power turbine rotor acceleration control (PAC) controls that the
maximum speed and the acceleration of the power turbine are not exceeded,
to avoid engine damage. [12]

The loss of load detector (LLD) watches the change of speed of the
power turbine. If it exceeds a certain value the LLD will order full closing
of the fuel valves, which will activate the GDC. [2]

The maximum primary pressure ratio controller (MPPRC) and the

maximum main pressure ratio controller (MMPRC) prevent that the differ-
ential pressures over the gas fuel valves get too low. This is done since too
low differential pressures can cause the control of the gas flow through the
control valve to become unstable. If the pressure after the valve exceeds 87
% of the pressure before the valve the controller will activate and prevent
the valve to open any further, thus preventing the differential pressure to get
any lower.

When the controller to be in charge has been selected in the turbine gov-
ernor, its output (the desired heat flow) must be converted into gas control
valve orders. First the primary fuel ratio is calculated, that is the share of the
desired heat flow that is created by the primary gas fuel valve. Then the de-
sired gas control valve positions for both valves (primary and main) are cal-
culated using the effective area (which is calculated using the desired heat
flow and the gas fuel temperature among others) and the pressures over the
valves. The deviation of the actual valve positions from the desired are used
in PID-regulators, whose outputs are the gas fuel valve orders.

19
 The valve positions of bleed valve 1 and 2 and the guide van positions
are functions of the normalized gas generator speed N norm :
288
N norm = N ×
t 2 + 273
where N is the gas generator speed and t2 is an average of three measure-
ments of the compressor inlet temperature. [12]

The emission control system should provide stable combustion at mini-

mum CO and NOx emissions in the entire load range. The emission govern-
ing should also limit combustor pulsations. The system includes governing
of the combustor bypass system, governing of bleed valve 2 and governing
of pilot to fuel ratio (PFR). The combustor bypass system is used on part
load and bypasses combustor air to the burners to increase the flame tem-
perature. This reduces CO emissions, although NOx emissions are in-
creased. Governing of bleed valve 2 is used on low part loads and recycles
compressed air back to the compressor inlet. This increases the combustion
air temperature which reduces CO emissions. The governing of pilot to total
fuel ratio helps reducing NOx emissions and is used especially on full load.
All systems use calculated flame temperature, which can be calculated ei-
ther using the compressor pressure or the temperature increase in the com-
pressor. [12]

The protection part of the control program causes the gas turbine to trip if
it enters a dangerous state. A gas turbine trip interrupts the fuel flow to the
gas generator. The gas generator speed decreases until it reaches about 230
rpm. Then the starting system will take over and drive the gas generator un-
til it has cooled down. The causes of a gas turbine trip can be that there is no
main flame during operation, there is no pilot flame during start up, any of
the gas generator bearing temperatures is high high (i.e. at a dangerously
high level), the vibration level at any of the bearing housings is high high,
the axial position of the shaft is high high, the bleed valve position is incor-
rect, the shaft speed is high high or the exhaust temperature is high high.
Some trips are immediate and others are delayed. An immediate trip causes
immediate unloading, whereas delayed trips implies unloading during 30 or
90 seconds. [1]

20
4 Simulator Setup
This chapter will explain the simulator system and the hardware and soft-
ware that are used. An explanation of why the actual solution was chosen
and its pros and cons in comparison with other possible solutions will be
given.

4.1 Overview
The framework for the simulator, i.e. the software and hardware to be used,
was decided during two former final theses. The decision on which solution
to use was based on investment costs, the demand of fast enough hardware
to enable simulation of a gas turbine and the prerequisite to simulate the
PROFIBUS communication between the simulator and the PLC:s (holding
the control program). It was also important that the simulator should be easy
to modify and develop for future applications.

The chosen solution includes the use of the Siemens software SIMIT in a
normal PC as simulator program and a SIMBA I/O card to communicate
with the PROFIBUS network, connected to the PLC:s.

In the simulator system, just as in the normal process, there will be an

operator station connected with Ethernet to the PLC:s, into which the con-
trol program is loaded. In the gas turbine setup there are DP-slaves con-
nected via PROFIBUS to the PLC:s. The DP-slaves handle I/O:s, connected
to measurement points in the gas turbine, managing the communication to
the control program in the PLC:s. In the simulation system the SIMBA I/O
card, connected via PROFIBUS to the PLC:s, is simulating the I/O:s. It will
send and receive signals to and from the simulator program and forward
them to the control program, which will receive them and act as if they
came from the real gas turbine. (See figure 15 for an overview of the simu-
lator setup and the gas turbine setup).

Figure 15 Overview of the simulator and the process

21
4.2 SIMIT
SIMIT is a modular system of blocks for modeling and simulation similar to
MATLAB simulink. It is a real time simulation system, where data is proc-
essed and simulated in real time. It is therefore not possible to run tests us-
ing data collected from measurements from a real gas turbine, as can be
done in simulink. This makes testing more difficult. However, it is possible
to save output data from the simulation. The complexity of the components
in the SIMIT library is somewhat limited, but new components can be cre-
ated using the component editor.

4.3 Simba pro PCI card

The SIMBA pro PCI card is fitted into a PCI slot in a PC. One card has two
connectors that can be connected to a PROFIBUS network. This means that
it can be connected to two PLC:s. If more PLC:s are needed for the control
program, another SIMBA pro PCI card will have to be purchased.

SIMBA pro itself can be used to create simulations of simple systems, in

which case SIMIT is not necessary. For the gas turbine application, SIMIT,
with more advanced components, is more suitable. [11]

4.4 Discussion on the chosen solution

The former final thesis writers revised several different solutions before de-
ciding to use the one with SIMIT and SIMBA pro. One was to have an in-
ternal simulator in the same PLC:s as the control program. The program-
ming language would then be the same as for the control program. This
means that no additional programming language has to be learned by the
users, as has to be done if SIMIT is used. A rough simulator of this type was
already in use at Siemens at the time. One disadvantage with an internal
simulator is that the simulator is generating extra code to the control pro-
gram that has to fit into the memory of the PLC:s. Another important disad-
vantage is that the PROFIBUS communication cannot be tested if the simu-
lator is internal. The disadvantages lead to the choice of an external simula-
tor. [11]

When it was decided to use an external simulator two different configu-

rations were revised. One was the chosen, with SIMIT and SIMBA pro. The
other was to use SIMULINK and a PROFIBUS DP I/O card. The later solu-
tion was discarded because it was considered too complicated to get the
communication between SIMULINK and the PROFIBUS card to work and
that it does not support failsafe (see chapter 3.6) neither in SIMULINK nor
in the PROFIBUS card. [11]

There are thus several advantages with the chosen solution. The control
program does not have to be changed to run it with the simulator. The buss
communication can be tested since PROFIBUS is used just as in the real
process. It is easy to get the communication between the control program in
the PLC:s and the simulator program to work. The SIMBA I/O card sup-
ports failsafe. SIMIT did not support failsafe at the time when the work with

22
this final thesis started. An update that would do so was due to be released
in March 2005. [11]

A further discussion on the SIMIT simulator, advantages and disadvan-

tages and problems encountered working with it can be found in the chap-
ters 5.3, 5.4 and 7.2.

23
5 Construction
This chapter will describe how the simulator was constructed using the
hardware, software and theory described in earlier chapters. The interaction
between the simulator and the control system will be further explained and
problems related to that communication will be discussed.

5.1 Selection of model type

The first decision to make was whether the model should be based on physi-
cal relations or on measurement data. A model based on physical relations
was considered to be too complicated and not feasible to achieve for one
person considered the given time frame. It was also assumed, after discus-
sions with the control system constructors at Siemens, that a model based on
measurement data would be good enough for the given purpose. The rea-
sons above settled the choice in favor of the model based on measurement
data.

A decision on how to model the relations between the signals had to be

made later in the construction process (See Chapter 6.4).

5.2 Extraction and analysis of measurement data

To enable the development of a model of the gas turbine SGT-600, meas-
urement data from a machine of that type was required. Since there was no
such data from the Finspång testing site (the SGT-600 turbines had been
tested only with liquid fuel there) data had to be gathered from another site.
For this purpose the Conditioning Monitor System (the CMS system)
proved to be very useful.

The Conditioning Monitor System (CMS) is a tool for collecting, pre-

senting, analyzing and storing process data for extended periods. It enables
continuous monitoring and long-time storage of operation parameters of a
turbine or a power plant. Real-time and historical data can be presented for
example as time trends or x-y plots. There are a large number of standard
reports and plots specific for the actual gas turbine type available. It is also
possible to make user specific trends and plots. There are tools to export
data to Excel files. [13]

The only gas turbine of SGT-600 type available was TVK in Hungary.
That meant that the model was going to be based on data collected from that
specific gas turbine.

There was about ten months of logged data available form TVK, so cer-
tain time intervals had to be chosen to make the amount of data manageable.
To get a good picture of the gas turbines functionality some normal starts,
normal stops, trips and changes of load were chosen from the logged data.
This selection of interesting events was made using Trends in CMS, where a
number of selected signals in a chosen time interval can be watched as
graphs (see Figure 16). To be able to find the events mentioned above in the
data, the active load signal was analyzed in particular, since the load rises
from zero during a start, drops to zero with a time delay of about five min-

24
utes during a normal stop and drops to zero immediately during a trip. There
were plenty of those events in the data, although no delayed trips of 30 sec-
onds or 90 seconds were found.

A selection of which signals to use had to be made to reduce the amount

of data. It was assumed to be sufficient to analyze the digital signals directly
in CMS, while the analog signals had to be transferred to an excel-format to
enable further analyzing in MATLAB. All the analog signals that are sent
between the control program and the I/O:s of the gas turbine were needed,
which resulted in more than 100 signals.

In the CMS system data is logged every second, but not if the signal is
constant or linearly changing. This results in data logged with non-regular
intervals. Because of that the chosen data had to be interpolated to be of any
use for analyzing in MATLAB. This was done using an Excel-program de-
veloped by the service-department at Siemens, which is responsible for the
CMS system.

To make the construction of the model practicable, simplified relations

between the signals had to be found, i.e. relations where one out-signal de-
pends on only one in-signal. This was done by first getting a basic under-
standing of how a gas turbine works (se chapter 1) and then watching se-
lected signals using Trends in CMS. It was found that the speed of the gas
generator affected most of the other measured signals, such as temperatures
and pressures. In reality those signals depend on a number of factors, like
position of the bleed valves and inlet guide vane, activation of the combus-
tion bypass system and ambient conditions. It was nevertheless observed
that the behavior of most signals could be described fairly well using the
speed of the gas generator. The speed of the gas generator was found to be
dependent on the heat flow, which is dependent on the openings of the gas
control valves. (See Figure 16)

25
Figure 16 Graphs from the CMS system showing the openings of the main gas control
valve (red), the primary gas control valve (blue) and the heat flow (green).

5.3 Building the model in SIMIT

The first approach to model the relations was to use transfer functions and
identify parameters in MATLAB System Identification Toolbox. Since there
was no ready made transfer function component in SIMIT, a new compo-
nent had to be developed using the Component Type Editor. After several
attempts to model the relations using the System Identification Toolbox it
was found that since all relations were more or less non-linear the result was
not satisfactory. It was decided not to use transfer functions with identified
parameters as the main alternative due to time restrictions. To get a func-
tioning model, with sufficient accuracy, was considered to be the main task.
Another possibility would have been to try linearising the relations before
identification, but this was, as said earlier, not within the given time restric-
tions.

Instead of using transfer functions with identified parameters it was de-

cided to use registers to simply map one input-signal to one output-signal.
For this purpose there are components in SIMIT called polygons, which are
registers with maximum twenty entries that use interpolation between the
points. This approach had been taken by the former final thesis writers, who
had also developed a MATLAB-function that, given the in- and output-
signal, picked out the twenty points that best described the relation between
the signals. The chosen points could then be inserted in the SIMIT poly-
gons. Sometimes it was sufficient to pick out the points manually from a
plot of the output-signal depending on the input-signal, using the Ginput
functionality in MATLAB. To get a dynamic behavior, the polygons were
followed by PT-components, which delay and smooth the signal.
T
yn = yn −1 + s ( xn − yn −1 )
T1

26
The chosen time delay, T1 , corresponds to the time it takes for a step re-
sponse to reach 0.63. The approach was assumed to result in a model good
enough for the given purpose, since a working model for the SGT-700 type
had been constructed using the same method.

Binary signals were mainly modeled as answers on orders. For example

the Start Motor on order signal from the control system would result in the
setting of the Start Motor on indicator signal from the simulator. Buttons
were included to enable manual setting of the signals, thus enabling simula-
tion of faulty signals (See Figure 17). Some signals had to be modeled in a
more complex way, for example the Main Flame indicator signal and the
Pilot/Main Flame indicator signal (indicating ignition, pilot and main
flame). The logic for those signals was developed using information from
the operator picture for the fuel system (seen in Figure 8) and the system
description.

Figure 17 Start motor indicated on or off as answer on order from the control system. Sig-
nals are represented by symbolic names. MBJ stands for Starting System, YU11 is a digital
on order, YU01 a digital off order and XP11 a digital on indication.

The model was going to be of Mechanical Drive type, which means that
a compressor is attached to the power turbine. It was not in the scope of this
thesis to model the compressor (see chapter 1.4). It was nevertheless neces-
sary to have a rough model of the load to be able to model the power turbine
speed. An available MATLAB model of the compressor was used and trans-
lated into SIMIT code.

The signals used in the SIMIT model were grouped in the same way as in
the control system, corresponding to the subsystems that a gas turbine con-
sists of. Specific SIMIT diagrams were used to gather all signals and to
convert them into global signals. The global signals could then be used on
multiple different diagrams in the model. This was done to make the han-
dling and changing of the model easier. The gathering of signals in one
place makes it possible to further develop the model to make it possible to
turn on and off specific signal groups and thereby adapt the model to the
options chosen by the customer for a specific delivery.

Some measured signals from the gas turbine are not transmitted to the
control system via PROFIBUS. This applies to the axial displacement (of

27
gas generator and power turbine) and bearing vibration signals, which are
instead transmitted via modbus. Since they are not transmitted via
PROFIBUS they are not included in the I/O-cards in the hardware configu-
ration and are consequently not present in the signal list imported to SIMIT.
They can therefore not be simulated in the SIMIT-simulator. The signals are
nevertheless needed by the control system and must therefore be simulated.
This was solved by internal simulation of these signals (in the PLC:s). The
signals were modeled on CFC-charts that were added to the control program
code. The solution is not ideal, since the purpose is to be able to test the
control program in the simulator without changing it. The other possible so-
lution would have been to add the signals to I/O-cards in the hardware con-
figuration to be able to simulate them in SIMIT. This would have requested
changes in the hardware configuration as well as in the control program
code and was seen a less suitable solution.

During the work with the construction of the model it turned out that
some of the measured signals from TVK were apparently not correct. Other
signals were not even included in the data. The solution to this problem was
to use data from the gas turbine type 10C, which had been used to develop
the model of that turbine type. That made it possible to model the signals,
even though not in a completely correct way since the data was from an-
other turbine type.

5.4 Communication between the simulator and the

control program
The signals that will be used in SIMIT have to be imported from SIMATIC.
The output-signals from SIMATIC will be input-signals in SIMIT and vice
versa. A new signal list has to be imported for every new SIMIT-project. If
the signals are changed in the control program the signal list in SIMIT has
to be modified or reimported. In SIMIT a signal list can be imported from
SIMATIC to a gateway, from where they can be dragged-and-dropped onto
the diagrams. The signals are identified through their addresses on the I/O
cards, but they are represented by symbolic names, which make the SIMIT-
model easier to grasp (see figure 17).

It proved to be a non-trivial task to import the signal list to SIMIT. A

complete list should include addresses, symbolic names and for analog sig-
nals ranges and types (see figure 18). In the SIMIT-version used in the be-
ginning a SIMBA-gateway could be created, to which one could import ei-
ther a list with the addresses and the symbolic names or a list with the ad-
dresses, types and ranges, but not a complete list with everything. This was
solved by the construction of a JAVA-program that combined the two lists
into one complete. At the end of the twenty weeks dedicated to this thesis a
new version of SIMIT arrived. This new version was necessary to use for
reasons shown later in this chapter. In the new SIMIT-version the SIMBA-
gateway had been exchanged for a ProfibusDP-gateway. It was no longer
possible to import the types and ranges of the signals to the gateway. They
must now be added manually to the signal list. The import of symbolic
names did not work properly. The SIMIT support in Germany has been
alerted to the problem and is working to solve it.

28
Figure 18 ProfibusDP gateway with signals imported from SIMATIC

When the work with this final thesis started in February 2005, the current
version of SIMIT did not support failsafe. Since failsafe signals are used in
the control program it is necessary that the simulator software can handle
failsafe. A new version of SIMIT that would support failsafe was going to
be released in March 2005 but was delayed until July 2005. When the simu-
lator was run (using the SIMIT version that did not support failsafe) to-
gether with the control system the failsafe signals sent from SIMIT were
received as invalid values in SIMATIC. This caused trips to activate which
prevented the gas turbine from being started, which made it impossible to
test the model. Attempts were made to work around the problem but the
only result was that the signals could be sent properly for a short time, be-
fore they turned invalid. The model could therefore not be tested before the
new version of SIMIT arrived. After installing the new failsafe-supporting
version of SIMIT it seemed to work fine and the failsafe signals were re-
ceived properly. The remaining problem was that when restarting the simu-
lation after having made changes to and compiled the code, the failsafe sig-
nals were invalid again. The CPU:s had to be restarted several times before
all the signals would turn valid. It seems that SIMIT still needs some im-
provement. The new version made it possible (although with a lot of trou-
ble) to test the model, but if the simulator should be used for production
purposes, a new more stable version of SIMIT is most certainly required.
The SIMIT support has been alerted to the problem and they have localized
an error in the IM-card. The suppliers are working on a new update.

Another problem concerning the communication that was discovered

during the work with the simulator was that some of the I/O-cards in the
hardware configuration did not work, i.e. the signals sent via these cards
were not transmitted. Attempts were made to troubleshoot the cards and re-
move them from the hardware configuration and then replacing them. Since

29
the attempts were ineffective, another solution had to be found. The mal-
functioning cards were moved to other I/O-modules in the hardware con-
figuration. This solved the problem in the way that the signals are now
transmitted properly, which is essential to get a functioning simulator.

30
6 Testing
This chapter will describe the environment used for testing and running the
simulator. The sequences used to start the gas turbine process (or the
simulator) will be described and the testing process will be accounted for.

6.1 WinCC
WinCC is used for creating operator pictures of the gas turbine process. In
this thesis those operator pictures are used to start and stop the simulated gas
turbine and to watch the simulated process. (The gas generator speed and
the power turbine speed can be watched as trends, as well as temperatures
and other signals. Incoming alarms and events are logged in lists. Subsys-
tems of the process can be watched. See chapter 6.3 for more information on
how WinCC was used during the testing process.)

WinCC (Windows Control Center) belongs to the OS (Operator Station)

Engineering part of the PCS 7 software package. It is used for creating
process pictures (configuring PSC 7 OS). OS clients are PC stations that en-
able control and monitoring of an automation process. They are connected
to OS servers, which function as HMI connections to the automation sys-
tem. The OS client has its own WinCC project and visualizes the process
data generated on an OS server. [15]

6.2 Sequences
Sequences are used to start, run and stop the gas turbine. The main se-
quences are the Unit Sequence, the Turbine Sequence and the Gas Fuel Se-
quence. There are also sequences for the compressor and for liquid fuel, but
they will not be explained further since they are not used to run the simula-
tor (See Chapter 1.4).

The Unit Sequence starts the subsystems that need to be running before
the turbine starts, like lubrication oil system and ventilation system. The
safety system is reset and checked. It then sends an order to the Turbine Se-
quence to start the gas turbine and waits for the indication that the turbine is
in operation. When the order is received it indicates that the unit is in ser-
vice and continues running until a stop order is received from the operator
station. When it gets the order to stop, the turbine is unloaded and the gas
turbine is stopped and set in standby position, from where it can be started
again. [14]

The Turbine Sequence (seen in figure 18) is started by order from the
Unit Sequence. It starts with purging of the gas generator and sends an order
to The Gas Fuel Sequence to start. When purging time has elapsed it ignites
the pilot flame and then the main flame. The turbine is accelerated by the
start motor until the gas generator reaches 5200 rpm. The turbine is then ac-
celerated further without start motor. When the power turbine speed exeeds
95% of minimum speed, an indication is sent to the Unit Sequence that the
turbine is in operation. When a stop order is received from the Unit Se-
quence an order is sent to the Gas Fuel Sequence to stop, the flame is put

31
out and the cool down procedure starts. An indication is sent to the Unit Se-
quence that the turbine has been stopped. [14]

The Gas Fuel Sequence is started by the turbine sequence. It starts with
checking the fuel control valves by opening them, closing them and then
setting them in start position. If the valve positions deviate too much from
the control signals during the test, the start is aborted. After the purge time
has elapsed the ventilation valves are checked. Then leak tests are per-
formed on the Shut Off Valves before they are opened. The Shut Off Valves
are used for shutting off the fuel supply when the turbine is stopped. The
leak tests are executed by closing the valve in question and measuring the
pressure after the valve. When the pressure is measured again after a certain
time it is not allowed to differ more than a certain limit value from the first
measured value. When the gas fuel sequence receives an order from the
Turbine Sequence to stop, the Shut Off Valves and the Gas Fuel Isolation
Valve are closed and a ventilation valve is opened. (See Figure 8 for an
overview of the valves.) [14]

32
Figure 19 Turbine Sequence in an SFC chart

6.3 Testing the simulator

Testing was performed by starting the sequences from the operator pictures
in WinCC. In WinCC overviews of the sequences can be watched, which
makes it possible to see in which step the sequence is at the moment. It is
also possible to see the conditions that have to be fulfilled to move on to the
next step.

33
 Before any testing could be done all conditions had to be removed that
were connected to the compressor-part of the control program. (That is the
part of the control system that controls the load, which in normal operation
is situated in a third PLC. In this simulator it is not used, as explained in
chapter 1.4.) This had to be done since the gas turbine otherwise would have
stopped at one step waiting for an indication from a CPU that did not exist.

The aim of the testing process was to be able to run through the se-
quences to reach the state “Turbine in operation”. Before that was fulfilled
the gas turbine would stop at one step, not able to reach the next, due to dif-
ferent errors, such as errors in the logic of the model or errors that originated
in malfunction I/O-cards in the hardware configuration (explained in chapter
6.4). It was then necessary to see which condition was not fulfilled to be
able to identify the subsystem or the signal that was not modeled correctly.
It was usually necessary to look at the CFC-charts of the subsystem in ques-
tion, while running the process, to see the changing of state of the signals
and systematically try to track the problem backwards chart by chart.

The testing got complicated and was slowed down because of the failsafe
problem (discussed in chapter 6.4), since it was necessary to stop the simu-
lation to be able to make changes in the model. It could then take long after
restarting the simulation before all failsafe signals were properly transmit-
ted.

After having corrected all found errors in the model and moved some
I/O-cards in the hardware configuration, it was possible to get the simulator
running, i.e. it would start and accelerate, but not reach “Turbine in opera-
tion”. The remaining problem was that the speed of the gas generator and
the speed of the power turbine changed to fast, which caused the control
system to shut down the gas turbine.

PT-blocks were inserted to delay and smooth the speed signals. The de-
lays that were needed to get it slow enough were so big that the system got
unstable, i.e. the speed of the gas generator and the speed of the power tur-
bine were fluctuating radically.

In the operator pictures in WinCC the turbine regulator can be watched,

i.e. the actual set points and which regulator is in charge at the moment (see
figure 13). The instability problem was found out to originate in the com-
pressor (load) model. This could be seen as the fluctuation of the speed sig-
nals started when the Speed Controller (that controls the speed of the power
turbine) took over during acceleration.

The compressor model was based on an existing MATLAB-model (see

chapter 5.3) that had not been constructed to be run in real time. It was too
fast, i.e. the speed of the power turbine changed too fast. This caused the
Load Loss Detector to activate. To slow it down, PT-blocks had been in-
serted, as mentioned above. Attempts were made to adjust the time delays
so that it would be long enough for the speed to change at a reasonable pace
and short enough not to cause instability.

34
 It was found out that it was not sufficient to use time delays in the com-
pressor model. Limitations on how fast the power turbine speed could
change were inserted. This together with tuning of the time delays resulted
in a reasonable stable model. It was however necessary to make changes in
the control program to increase the limit of the power turbine speed when
the Load Loss Detector should activate. The model still has some minor in-
stability problems in a certain range of the power turbine speed (further dis-
cussed in chapter 7.1).

35
7 Conclusion
In this chapter the result of the testing of the simulator will be presented.
The limitations, advantages and possible usage will be discussed and sug-
gestions for further development will be given.

7.1 Result
A working model for the gas turbine SGT-600 has been developed, based on
the one previously developed for SGT-700.

The simulator can be started from WinCC and runs through the se-
quences much like a real gas turbine. The preparation systems are started,
purging is performed, the gas is ignited and the gas generator starts to accel-
erate. After acceleration to a certain speed the Speed Controller takes over
the control. A set point for the power turbine speed is set by the operator,
which causes the speed to change and stabilize at the requested level. If the
power turbine speed set point is set to a level high enough to result in a dan-
gerously high exhaust gas temperature, the T7 limiter will take over the con-
trol and stabilize the exhaust gas temperature at a safe level. At shutting
down the speed of the gas generator and the exhaust gas temperature are
slowly decreasing. The temperature delays are much shorter than in reality,
since there is no point in watching the simulator cooling off for hours.

The model has some minor instability problems in a certain range of the
power turbine speed. The problem origins in the compressor model, as ex-
plained in chapter 6.3. To model the compressor (load) was however not in
the scope of this thesis. Since a better model of the compressor is under de-
velopment in another thesis, it was decided not to put more effort into fine-
tuning the old model.

The model is realistic in the way that it is based on measurement data

from a real gas turbine. It has nevertheless been adjusted to work with the
control system. Time delays have been adjusted to get a stable model and do
not always correspond to real measured time delays. It can seem strange to
adapt the model to the control system, since the purpose of the model is to
test the control system. But, since the control system had been tested (and
the control parameters adjusted) against a real gas turbine it was assumed
that if the model was tuned to fit the control system it would be close
enough to the real turbine.

The combination of polygons and PT-blocks seem to work sufficiently to

model the relations between the signals. The polygons handle non-linear
relations and the PT-blocks add dynamics to the system. The main limita-
tion of the model is instead the simplified relations between the signals, i.e.
the output signals depend in reality on many more input signals than mod-
eled in this thesis. To include all affecting input signals, it would be neces-
sary to develop a mathematical model based on physical relations. This was
however out of the scope of this thesis.

36
7.2 Discussion
The simulator can be used for testing the standard of the control system or
to test the control system when adapting it to a specific delivery. To further
serve the latter purpose it would be good to develop option packages to en-
able quick adjustment of the simulator to the unique composition of each
gas turbine.

It is also specially suited for educational purposes, e.g. to teach the cus-
tomers about gas turbine functionality, mainly because of the operator pic-
tures (that have no equivalence in the SIMATIC-simulator, i.e. the simulator
where the simulation takes place internally in the PLC:s). The operator pic-
tures in SIMIT make it easy to control and watch the simulator and also to
simulate different operating scenarios. If this should be done with the SI-
MATIC-simulator one has to change parameters in the code, which is not
user friendly. The staff of the customer training department asked for a
simulator that would be easy to carry around. This is not the case with this
simulator, since the two PLC:s are quite heavy. The problem can however
not be solved at the moment, since PLC:s must be used. All other simulator
setups would have the same problem.

It is at the time of this writing not quite clear if and how the simulator
will be used at SIEMENS Industrial Turbomachinery. It is clear that a simu-
lator is needed, but not if it should be this one in SIMIT, the internal SI-
MATIC-simulator or yet another type. It is up to the heads of the depart-
ments concerned to decide which way to go. This final thesis has contrib-
uted with evaluation of the SIMIT software which will make the decision
easier. The functionality of the model itself is similar to the one in SI-
MATIC.

The advantages of the SIMIT-simulator compared to the SIMATIC-

simulator are:
• It is possible to test the PROFIBUS communication.

• It does not require extra memory of the CPU. This can

be extra relevant in the future, since there are plans to
fit the whole control program into one CPU.

• It is possible to make operator pictures in SIMIT,

which can be used to control the simulator. This would
be very useful if the simulator should be used for edu-
cational purposes.
The disadvantages are:
• The staff has to learn to use SIMIT, which is at the
moment not used.

• The investment cost for SIMIT-licenses is high.

• SIMIT needs further development, which can cause

problems when using it (as seen in this thesis).

37
 • Even though most blocks are the same in SIMIT as in
SIMATIC, user-made blocks in SIMATIC have to be
made in SIMIT as well, which causes extra work.

7.3 Further Development

The problem with the malfunctioning I/O-cards mentioned in chapter 6.4
must be solved in a proper way, without moving the I/O-cards. Since the
hardware configuration must not be changed, it is essential to get the I/O-
cards to work in the I/O-modules where they are supposed to be. This
should best be done by the people who have set up the hardware configura-
tion.

A more realistic model of the compressor needs to be developed, to re-

place the rough model used in this simulator. (This is being done in another
thesis at the department of Oil and Gas.) A generator model also needs to be
developed, to enable simulation of Power Generation.

Option packages need to be developed, to enable adaptation of the simu-

lator to specific deliveries. This has been prepared in this thesis by grouping
and gathering of the signals on specific charts.

To extend the model to be valid for dual fuel, a model for the liquid fuel
system needs to be developed. This can however not be done before the
control system code for the liquid fuel system is ready.

It is necessary to update SIMIT to get better failsafe functionality and to

be able to import the signal list properly. SIMIT support in Germany has
been informed of the problems described in this thesis and is working to
solve them.

Operator pictures needs to be developed in SIMIT if the simulator is to

be used for educational purposes.

Simulators for other gas turbine types can be developed in the same way
as this one. The simulator for SGT-700 needs to be modified to work with
the new turbine governor and the new version of SIMIT. It might be useful
to merge the two simulators into one, with a button for the user to switch
between them.

38
References

1) Siemens internal documentation, GT10 Basic Training

2) Siemens internal documentation, System Description GT10B2
3) Siemens, (2005),
http://www.powergeneration.siemens.com/en/oilgas/drives/gt/sgt600/index.
cfm (Accessed 2005-08-11)
4) Siemens, (2005), PCS 7 V6.1 Engineering System,
http://cache.automation.siemens.com/dnl/jk1NTM3MQAA_21407752_HB/
PHB_PCS7_e.pdf (Accessed 2005-06-15)
5) Siemens, (2004), Automation System S7-400 Hardware and Installation,
http://cache.automation.siemens.com/dnl/TQ1MTg5OQAA_1117849_HB/4
24ish_e.pdf (Accessed 2005-06-15)
6) Siemens, (2005), PCS 7 V6.1 Getting Started - Part 1,
http://cache.automation.siemens.com/dnl/TYxNjcwNwAA_21405992_HB/
ps7gs1b_e.pdf (Accessed 2005-06-15)
7) Siemens, (2004), Configuring Hardware and Communication Connec-
tions STEP 7,
http://cache.automation.siemens.com/dnl/DM0NTMzMQAA_18652631_H
B/S7hwV53_e.pdf (Accessed 2005-06-15)
8) Siemens, (2005), CFC for S7,
http://cache.automation.siemens.com/dnl/TU1NTA2NwAA_21401430_HB/
CFC_For_S7_e.pdf (Accessed 2005-06-15)
9) Siemens, (2005), SFC for S7,
http://cache.automation.siemens.com/dnl/zg0NTY3OQAA_21400488_HB/s
7sfcs7b_e.pdf (Accessed 2005-06-15)
10) Siemens, (2004), PROFIBUS Product Brief
11) Lindholm and Klang, (2005), Modelling and simulation of a gas turbine,
LITH-ITN-ED-EX--05/009--SE
12) Siemens internal documentation, (2004), GT10B2 engine control speci-
fication
13) Siemens internal documentation, (2004), CMS User Guide, Version 2.2
14) Alstom internal documentation, (2003), GT10 Mechanical Drive Stan-
dard Sequence diagram
15) Siemens internal SIMATIC documentation

Figure 1 is from
http://www.powergeneration.siemens.com/en/oilgas/drives/gt/sgt600/index.
cfm (Accessed 2005-08-11, modified by the author)
Figure 2, 3, 4, 5, 7, 9, 10 are from [1]
Figure 11, 15 are from an internal Siemens document (modified by the au-
thor)
Figure 8, 12, 13, 14, 16, 17, 18, 19 are screen shots from the simulator or the
control system
Figure 6 is from
http://www.powergeneration.siemens.com/en/oilgas/drives/gt/sgt600/emissi
ontech/index.cfm (Accessed 2005-08-12)

39