Professional Documents
Culture Documents
Modelling and Simulation of A Gas Turbine: Henrik Klang Andreas Lindholm
Modelling and Simulation of A Gas Turbine: Henrik Klang Andreas Lindholm
Modelling and Simulation of A Gas Turbine: Henrik Klang Andreas Lindholm
LITH-ITN-ED-EX--05/009--SE
Henrik Klang
Andreas Lindholm
Norrköping 2005-04-22
Datum
Avdelning, Institution Date
Division, Department
_ ________________
Titel
Title Modelling and simulation of a gas turbine
Författare
Author Henrik Klang, Andreas Lindholm
Sammanfattning
Abstract In this thesis, a gas turbine simulator for the Siemens GT10C was developed and implemented.
It concerns everything from the theory behind the simulator; both the hardware and software involved,
to how the actual simulator was built using these tools. The theory concerns itself with basic automatic
control concepts, as well as basic turbine theory.
The simulator setup is being discussed concerning both technical and economic issues. A robust
hardware solution is then selected, using the basic requirements, which the simulator then is built
around.
The tools used are the Siemens SIMATIC automatic control system and the Siemens SIMIT real-time
simulator using a SIMBA Pro PCI card to interface with the PLC:s in the SIMATIC system. Matlab are
also used to a lesser extent to build the simulator behavior in SIMIT.
In the end, a fully featured simulator is presented that can be used for various purposes such as training
operators, trying out new concepts and testing the automatic control system used to control the turbine.
Further development that could be done, by other engineers, in the future, is also discussed.
Nyckelord
Keyword Automatic control, fieldbus, gas turbine, modelling, S7, Siemens, SIMATIC, SIMBA, SIMIT, PLC,
PROFIBUS, simulation
Upphovsrätt
Copyright
The publishers will keep this document online on the Internet - or its possible
replacement - for a considerable time from the date of publication barring
exceptional circumstances.
The online availability of the document implies a permanent permission for
anyone to read, to download, to print out single copies for your own use and to
use it unchanged for any non-commercial research and educational purpose.
Subsequent transfers of copyright cannot revoke this permission. All other uses
of the document are conditional on the consent of the copyright owner. The
publisher has taken technical and administrative measures to assure authenticity,
security and accessibility.
According to intellectual property law the author has the right to be
mentioned when his/her work is accessed as described above and to be protected
against infringement.
For additional information about the Linköping University Electronic Press
and its procedures for publication and for assurance of document integrity,
please refer to its WWW home page: http://www.ep.liu.se/
Preface
This thesis is a result of about 20 weeks of analysis and implementation at
Siemens Industrial Turbomachinery AB, Finspång, Sweden.
It is a compulsory part of the education for the authors to receive their
Master of Science degree in Electronics Design Engineering from the Department
of Technology and Natural Sciences (ITN) at Linköping Institute of Technology.
The work was carried out in between June and November of 2004. The exam-
iner at Linköping Institute of Technology was Måns Östring and the instructor
at Siemens Industrial Turbomachinery AB, Thomas Strömberg.
The thesis was done in co-operation with another thesis worker, Daniel Wi-
gren, which also is a student at Linköping Institute of Technology studying to
receive a degree as a Master of Science in Applied Physics and Electrical Engi-
neering.
Thank you!
To everyone who deserves it.
Mandatory quotes from (in)famous persons iii
Abstract
In this thesis, a gas turbine simulator for the Siemens GT10C was developed
and implemented.
It concerns everything from the theory behind the simulator; both the hard-
ware and software involved, to how the actual simulator was built using these
tools. The theory concerns itself with basic automatic control concepts, as well
as basic turbine theory.
The simulator setup is being discussed concerning both technical and eco-
nomic issues. A robust hardware solution is then selected, using the basic re-
quirements, which the simulator then is built around.
The tools used are the Siemens SIMATIC automatic control system and the
Siemens SIMIT real-time simulator using a SIMBA Pro PCI card to interface
with the PLC:s in the SIMATIC system. Matlab are also used to a lesser extent
to build the simulator behavior in SIMIT.
In the end, a fully featured simulator is presented that can be used for various
purposes such as training operators, trying out new concepts and testing the
automatic control system used to control the turbine.
Further development that could be done, by other engineers, in the future,
is also discussed.
Contents
Abstract iv
1 INTRODUCTION 1
1.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Questions at hand . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.4 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.5 History of the Finspång development site . . . . . . . . . . . . . 2
1.6 The structure of this thesis . . . . . . . . . . . . . . . . . . . . . 2
2 THEORY 4
2.1 Overview of fieldbus technology . . . . . . . . . . . . . . . . . . . 4
2.2 PROFIBUS° R
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 PROFIBUS-DP° R
. . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 MPI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5 The Siemens SIMATIC totally integrated automation system . . 7
2.6 SIMATIC hardware . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.7 Engineering and operator stations . . . . . . . . . . . . . . . . . 7
2.7.1 S7 PLC’s . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.7.2 DP-slaves explained . . . . . . . . . . . . . . . . . . . . . 9
2.8 SIMATIC software . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.8.1 SIMATIC manager . . . . . . . . . . . . . . . . . . . . . . 10
2.8.2 Hardware config . . . . . . . . . . . . . . . . . . . . . . . 11
2.8.3 SFC editor . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.8.4 CFC editor . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.8.5 NetPro . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.8.6 Fail-safe versus non fail-safe . . . . . . . . . . . . . . . . . 12
2.8.7 WinCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.9 The Siemens SIMIT simulation system . . . . . . . . . . . . . . . 13
2.9.1 SIMBA Pro . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.9.2 Using SIMIT . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.10 The Siemens GT10C gas turbine . . . . . . . . . . . . . . . . . . 15
2.10.1 Compressor . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.10.2 Combustion chamber . . . . . . . . . . . . . . . . . . . . . 17
2.10.3 Power turbine . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.11 Turbine regulators . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.11.1 STC - starting control . . . . . . . . . . . . . . . . . . . . 19
2.11.2 NGGL - gas generator speed limiter . . . . . . . . . . . . 19
2.11.3 SC - speed controller . . . . . . . . . . . . . . . . . . . . . 20
2.11.4 T7L - exhaust average temperature limiter . . . . . . . . 20
2.11.5 T7Li - exhaust inner temperature limiter . . . . . . . . . 20
2.11.6 MPC - maximum servo position control . . . . . . . . . . 20
2.11.7 GAC - gas generator acceleration control . . . . . . . . . 20
2.11.8 GDC - gas generator deceleration control . . . . . . . . . 21
2.11.9 PAC - power turbine acceleration control . . . . . . . . . 21
2.11.10 LLD - loss of load detection . . . . . . . . . . . . . . . . . 21
2.12 Automated start of turbine . . . . . . . . . . . . . . . . . . . . . 21
2.12.1 Unit sequence . . . . . . . . . . . . . . . . . . . . . . . . . 21
vi CONTENTS
5 RESULT 41
5.1 Conclusions and discussion . . . . . . . . . . . . . . . . . . . . . 41
5.2 Future development . . . . . . . . . . . . . . . . . . . . . . . . . 42
References 43
LIST OF FIGURES vii
List of Figures
2.1 Old fieldbus network. . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Modern fieldbus network. . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Two Siemens SIMATIC S7 PLC boxes (back) connected to DP-
slaves (front). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Basic configuration with a DP-rack in the hardware config editor 11
2.5 Two typical CFC-blocks, normal (left) vs. fail-safe (right). . . . . 12
2.6 SIMIT and SIMBA Pro interaction scheme (OS and ES comput-
ers connected to the S7 PLC are not included). . . . . . . . . . . 14
2.7 SIMBA Pro list of configured DP-slaves. . . . . . . . . . . . . . . 14
2.8 Example of a SIMIT diagram/modell. . . . . . . . . . . . . . . . 15
2.9 The GT10C gas turbine is a turbine with two rotors, the com-
pressor rotor and the power turbine rotor. This type is known
as a double shaft turbine. The three main parts of the turbine is
enclosed in red. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.10 The compressor valves locations of the turbine. . . . . . . . . . . 17
3.11 Solution I, PLC to PLC communication (simplified schematic). . 27
3.12 Solution III & IV, PLC to PC (C/C++/SimuLink) communica-
tion (simplified schematic). . . . . . . . . . . . . . . . . . . . . . 29
3.13 Final hardware setup with the turbine PLC’s containing the con-
trol program and the PC containing the turbine simulation pro-
gram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.14 The HMI of a turbine system (operator screen). . . . . . . . . . . 33
4.15 Grouping of diagram and operating screens in SIMIT. The dif-
ferent alphanumeric codes are an internal Siemens scheme for
naming different systems in the turbine. . . . . . . . . . . . . . . 34
4.16 Simple development process flowchart of the simulator model. . . 36
4.17 Data from performance tests used when developing the model. . 37
4.18 The relation between the curve taken from SCADA Pro (blue)
and the polygon curve done with Matlab (red). . . . . . . . . . . 37
4.19 The SIMIT model to calculate the heat flow value. . . . . . . . . 39
viii LIST OF TABLES
List of Tables
2.1 Comparison between PROFIBUS and older fieldbus standards . . 5
2.2 Speed versus pressure ratio in the compressor. . . . . . . . . . . . 17
2.3 Valve characteristics. . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4 The unit sequence which starts up turbine. . . . . . . . . . . . . 22
2.5 Gas fuel sequence. . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6 Turbine sequence. . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.7 Quick recap of investment costs for each solution. Notes: 1 A
more powerful PLC might be needed. 2 If implemented. . . . . . . 30
ABBREVIATIONS ix
Abbreviations
AI Analogue In
API Application Programming Interface
AO Analogue Out
CC Control Center
CFC Continuous Function Chart
CP Communication Peripheral
CPU Central Processing Unit
DI Digital In
DIP Dual-Inline Package
DO Digital Out
DOS Disk Operating System
DP Decentralized Peripheral
ES Engineering Station
FAT Factory Acceptance Test
GDC Gas generator Deceleration Control
GUI Graphical User Interface
HMI Human Machine Interface
I[/]O Input/Output
I 2C Inter-Integrated Circuit
IGV Inlet Guide Vane
IPX Internetwork Packet Exchange
MAC Media Access Control
MPI Multipoint Interface
NIC Network Interface Card
OLE Object Linking and Embedding
OPC OLE for Process Control
OS Operator Station
OSI Open System Interconnection [Reference Model]
PCI Peripheral Component Interconnect
PCS Process Control System
PID Proportional Integration Derivation
PLC Programmable Logic Controller
PROFIBUS PROcess FIeld BUS
PS[U] Power SUpply
PT Power Turbine
RAM Random Access Memory
SCADA Supervisory Control and Data Acquisition
SCL Structured Control Language
SFC Sequential Function Chart
TCP/IP Transmission Control Protocol/Internet Protocol
UR2 Universal Rack Type 2
1 INTRODUCTION 1
1 INTRODUCTION
1.1 Purpose
The primary purpose of this thesis is to develop a simulator and model for
a specific gas turbine, the GT10C; which uses the Siemens SIMATIC system
(“SIMATIC”, also referred to as “S7” where 7 is the latest version of SIMATIC).
When this is done the model of the gas turbine will be implemented in the system
which has been devised.
Secondary purpose are to include more additions to the simulator, such as
simulating the next coming stages in the turbine system; compressors, and other
loads. More wanted additions is to be able to simulate other turbine models in
the same family.
The first goal of the thesis is to analyze what kind of system that would be
needed to build and implement the simulator itself, both hard- and soft wise.
When this is done, the actual simulator will be implemented in the developed
system, and finally, at the end, tested against the actual turbine automation
and control system located in multiple S7 PLC boxes.
This thesis is being carried out at Siemens Industrial Turbomachinery AB
(“Siemens”) located in Finspång, Sweden which is a part of the Siemens group.
1.2 Background
Siemens Industrial Turbomachinery AB is about to adopt the S7 system, thus
replacing the currently used one developed by ABB.
The current project at hand (as of June 2004) is to install a GT10C gas
turbine in Eischleben, Germany. This is the first project, in Finspång, to use
the S7 system from Siemens, thus the development of a simulator would be
helpful in the entire development cycle of a turbine and its systems.
The simulator will be used for system testing, training new operators and
simulating new additions to the automatic control system, before the system
is tested in an actual turbine on site. This will help reduce, both development
cost, and time to market, for new additions to the system. Moreover, it will
shorten the time spent training operators for a specific turbine thus increasing
total profit per installed turbine and project.
• How should the communication work in between the simulator and the
PLC. Is there any ready-made solutions to use?
Secondary,
1.4 Limitations
The initial limitations will be to devise a simulator for only one turbine, the
GT10C. Although extending the simulator for other turbine types such as the
GTX100 would be favorable.
At first the model of the turbine shall be static, that is, no dynamic signals
shall be modeled. Static signals in this report refers to such ones that does not
use any feedback when calculating the output signals of the system.
If, and when, the static modeling is done, a half dynamic model should be
developed. In this report a half dynamic model is a model where analogue and
digital signals can be triggered by other signals, but not as a closed loop, such
as a PID (for further discussion of these concepts please refer to Section 4).
Furthermore, an external unit (load) of the turbine, shall be modeled if there
is any time.
Last, the chapter RESULT discusses the results, and what could have been
done differently. Also future improvement that would be possible is discussed
here.
4
2 THEORY
To be able to understand the next coming chapters some explanations of the
terms and concepts, in this thesis, will be discussed below. A simple overview
of the programs used will also be presented to give the reader a more in depth
view of the problems the authors faced and measures taken to solve these.
This chapter also explains, on a very low level, how a turbine works, and
especially the turbine which the simulator in this thesis is developed for.
2.2 PROFIBUS°
R
Location A
PLC
Location B Location C
Location A
PLC
Fieldbus Cable
Location B Location C
PROFIBUS is the leader of fieldbus communication with over 20 per cent of the
world market. Approximately 500.000 plants and factories use PROFIBUS,
with over 5 million nodes2 installed.
2.3 PROFIBUS-DP°
R
2 A node is a PROFIBUS station with a unique PROFIBUS address such as PLC’s, sensors
and actuators.
2.4 MPI 7
2.4 MPI
MPI[4], Multipoint Interface, is a closed Siemens standard. It is used as a
programming interface for S7 PLC’s (see Section 2.7 for information on S7
PLC’s), but can also be used as a interface to operator panels simultaneously.
Each MPI interface, or node, is identified by an address.
2.7.1 S7 PLC’s
The S7 PLC’s differ in function, speed and memory capacity. The configura-
tion needed is dependent upon the application. Processes such as conveyor belt
systems with few stations could settle with a simpler S7 PLC’s, while more com-
plex systems such as turbines and paper machines could need several powerful
S7 PLC’s working together as a cluster or separate, dividing the handling of
DP-slaves and the calculations needed to run the plant process.
The most basic S7 PLC hardware configuration consists of the following:
The PS, CPU and communication process modules are placed in a UR2 rack.
The UR2 rack is both a mechanical and electrical “backplane”. The rack lets
the different components communicate with each other as well as provides a me-
chanically stable platform for use in harsh industrial environments. Depending
on the type of the rack it can mount multiple “PLC cells”, where each group
consists of a power supply, CPU and communication processor.
The PS supplies power to the rack through the backplane. The PS also
consists of two batteries (rechargeable), which are used if the regular power
from the grid is lost, thus creating redundancy. The S7 PLC program itself
is stored on a flash memory card, which holds the information, even during a
power loss. Although, the configuration data which holds information of what
kind of components that is mounted on the rack, is stored in a volatile RAM
thus making PS redundancy important via the batteries.
Another source of redundancy can be created by adding a extra PS to the
rack. For these purposes, and depending on the saftey needed, there are different
types of PS such as redundant and non-redundant. Obviously, creating a non-
redundant solution is a solution with low initial costs, but could be proven a
fatal decision afterwards, due to production loss caused by downtime of the
PLC-system.
The CP modules is in many ways like a normal network interface card
(“NIC”) found in any modern PC, thus having connections for ethernet wiring.
Since Industrial Ethernet (a subset of Ethernet) is very much like normal Eth-
ernet the wiring can be connected to switches and hubs like any other computer
network. The engineer can then program the PLC through the CP cards in-
terfaces as well as through the MPI interface as mentioned earlier. Also, the
ethernet interfaces are often connected to the ES and OS stations where the op-
erator can control the plant through the S7 PLC. There are CP modules which
holds fiber optic connectors thus handling fiber optic nets if the S7 PLC needs
to process time critical data and process objects.
At the heart of the plant control is the CPU module(s). It consists of a CPU,
and a memory module. The memory module size needed depends on the size of
2.8 SIMATIC software 9
the plant control program it needs to store. Also, depending on the model type
of the CPU module it can be exchanged for larger modules (varying between
1-16MB).
The CPU module also holds a DP interface and a combined DP/MPI in-
terface. The MPI can be used to programme the CPU modules flash card via
an expansion card in a PC. Usually, though, the memory card is programmed
through the CP modules Ethernet connection (that is via a network).
It is through the DP interface that the DP-slaves are connected. That is, the
DP interface handles the process objects communication (see next sub chapter
for a more throughout description). As with the other kinds hardware modules
detailed above there are different types of CPU’s available, varying in speed and
capabilities such as redundancy and fail-safe (see Section 2.8.5 for an explanation
of fail-safe).
Figure 2.3: Two Siemens SIMATIC S7 PLC boxes (back) connected to DP-
slaves (front).
least, software built on old concepts which the developers have tried to fusion
with new ones such as drag and drop and GUI. Therfore, SIMATIC is relatively
hard to work with initially, creating a steep learning curve for beginners.
A summary of the programs explained are included in the list,
• Continuous function chart editor - Drag and drop editor where func-
tion blocks are interconnected to create the PLC functions (i.e. PID reg-
ulators, logic blocks et.c.).
• PG\PC interface - Sets the interface used when programming the CPU
module.
Figure 2.4: Basic configuration with a DP-rack in the hardware config editor
Figure 2.5: Two typical CFC-blocks, normal (left) vs. fail-safe (right).
2.8.5 NetPro
This program is used to set up the communication in between the PLC’s and the
operator station(s). In addition to hardware config where the communication is
initially set up, this program modifies the necessary “blocks” to ensure that the
communication in between the PLC and OS works correctly concerning message
and alarms sent between the process, PLC’s and OS.
2.8.7 WinCC
WinCC, or Windows Control Center, is the program where the operator pictures
are built in. The connections to the signals from the CFC-editor is connected
to GUI components in WinCC such as buttons, diagrams, pictures and sliders.
WinCC can be run on multiple PC’s known as operator stations (“OS”)
which makes it easier to have an overview of the gas turbine process and to
control it through the GUI.
The SIMIT suite can be extended with a SIMBA Pro PCI card (“SIMBA”). The
card itself is fitted into a PCI slot in a PC and accessed via a special software
which SIMIT uses to relay its information flow through to the PCI card itself
(through the use of a hardware driver). The schematic view of SIMIT combined
with the SIMBA Pro PCI card is seen in Figure 2.6.
SIMBA Pro has two connectors. Each can be connected to a PROFIBUS
network, which means the card can handle two networks at the same time.
SIMBA Pro does not need SIMIT to function as it is a self-reliant system.
SIMIT should be seen as an add on containing more advanced graphical GUI
and better analysis tools. SIMBA Pro holds several tools and components to
be able to simulate simpler processes.
The values of the I/O ports of the DP-slave can also be set individually if
needed. Since SIMBA Pro is built for S7 communication through PROFIBUS
the DP-slave configuration can be easily imported into SIMBA Pro from the
SIMATIC hardware config program. A list of a imported DP-slave configuration
can be seen in Figure 2.7.
SIMBA Pro also handles fail-safe DP-slaves by adding extra signals for the
fail-safe to be turned on or off for each signal in the hardware config configured
in the fail-safe mode.
14
SIMIT
Figure 2.6: SIMIT and SIMBA Pro interaction scheme (OS and ES computers
connected to the S7 PLC are not included).
Each signal has a hardware address and an alias assigned to it (this is done
during the development cycle of the PLC program in SIMATIC). The alias is a
name used when connecting I/O ports in SIMIT to simplify the process since
only a hardware address would be confusing and rather cumbersome.
A very simple example is shown in Figure 2.8 where two input signals, IN-
PUT SIGNAL 1 and INPUT SIGNAL 2, is added together. The result is
sent through a sin block, and finally output to the OUTPUT SIGNAL port.
More complex concepts such as feedback (PID-regulators) are also available to
the engineer.
Figure 2.9: The GT10C gas turbine is a turbine with two rotors, the compressor
rotor and the power turbine rotor. This type is known as a double shaft turbine.
The three main parts of the turbine is enclosed in red.
2.10.1 Compressor
The function of the compressor is to compress the surrounding air. When the
air is compressed the temperature and pressure increase. To avoid surge3 in
the compressor, two bleed valves, are used as well as an inlet guide vane (IGV).
These elements can be seen in Figure 2.10(BV1 and BV2).
The IGV controls the amount of air that flows into the compressor. The bleed
valves are used to level out the pressure in the compressor to avoid fluctuation in
the pressure, which is an unwanted effect. BV1 is just an ordinary on/off valve
and BV2 is a control4 valve. BV2 is connected to the air inlet through piping,
this to be able to control the temperature in the combustion chamber. At a
higher flame temperature the emissions are lower. The CO and N Ox values
decrease and the turbine is in that way more environment friendly.
BV1 is used only during the start of the compressor and closes when the
speed Nnorm of the compressor reaches 6800rpm. BV1 is fully closed between
6950rpm and when stopping the turbine it remains closed until 400rpm.
r
288
Nnorm = N · (2.1)
t2 + 273
The relation Eq. 2.1 between the valves position and speed is later used in
the SIMIT modell to verify a stable operation mode of the compressor. That is,
if the simulated values is out of range, the turbine control system would trip5
since the model is not working as expected.
To ensure stable operations and required engine performance, the compressor
guide vanes must be controlled according to Table 2.2. P3 is the compressor
discharge pressure (M P a) is calculated as an average of three different pressure
transmitters.
3 When a surge occurs the compressor stalls, that is, stops to work.
4A control valve lets the PLC control the exact position of the valve, that is, how much it
should open.
5 A trip of a turbine is when the control program shuts the turbine down because of errors
deceleration control (GDC) goes active. The GDC is activated if the load6 on
the turbine is lost due to an malfunction.
When the high pressurized warm air expands through the compressor the
kinetic energy is transformed to mechanical energy by the turbine blades and
increase the speed of the compressor, thus if the air and heat flow through the
turbine is not controlled the compressor would accelerate rapidly. The heat
flow value can be derived from the fuel gas valves characteristics when a certain
percentage opening generates a certain heat flow. The characteristics of the gas
fuel valves can be seen in Table 2.3. The valve characteristics is used in the
control program of the turbine (that is, in the SIMATIC system) and also in
the SIMIT model to obtain the heat flow value. The heat flow now expands and
continues through the power turbine.
is not entirely accurate since it does not drive the water jet directly, rather it uses something
called a mechanical drive in between).
2.11 Turbine regulators 19
before, connected to water jet propulsion systems like the Stena Carisma ferry
over Öresund.
power turbine speed variation. The register is valid for all matching options;
the points in between are linear interpolated.
This sequence initiates the purging of the turbine. The turbine needs to do
a purging before start to ensure that there is no gas trapped inside it. The
sequence is also used for accelerating the turbine with help of the start motor
after ignition. After 5400rpm the start motor is disengaged and the turbine is
now accelerating by itself, by controlling the gas fuel valves. This sequence also
monitor when the ignition starts and ignites the fuel gas by looking at pilot flame
and main flame indicators. The turbine sequence includes nine steps. Table 2.6
displays a complete list of all the steps included in the turbine sequence.
In the start of the gas turbine the unit sequence is activated first, the sequence
executes each step when the transition criteria are fulfilled. When the sequence
enter step 10, ventilation and lube oil systems are activated and running, the
compressor is ready for start. The trip system is also reset and operative. Now
the turbine sequence starts at step 1. This step also starts the gas fuel sequence
which runs through the valve check and leakage test. At the same time the
compressor turbine has purged the turbine for 75 seconds at 2300rpm and also
increased the speed to 2600rpm which is the speed of ignition. Pilot ignition
is initiated and the compressor turbine accelerates with help of the start motor
until 5400rpm where the start motor is disengaged, in the mean time the main
ignition also has been activated. The turbine exhaust temperature increases
rapidly to 400 degrees centigrade. The gas fuel sequence has now reached step 9
and will maintain this step until a stop is initiated. Now the turbine accelerates
by itself only using the gas fuel valves, the turbine sequence now reaches step
5 and the turbine is in service. When the heat flow value is over 17.77M J/s
the power turbine starts rotating and accelerates controlled by the turbines
ten controllers to a speed of 3250rpm. The unit is now in service and the unit
sequence has reached step 13. It is now up to the operator to set the appropriate
speed of the power turbine or use the set point set from the compressor PLC.
24
2.13 Summary
A short summary, and quick facts list follows. It is not meant to be comprehen-
sive; rather it should be used as a reminder on what the chapter is about.
Since the THEORY part is very important to get a firm grasp of the de-
tails explained and discussed the reader should feel that he or she is somewhat
familiar with the words and concepts in this summary before proceeding.
• DP slaves controls the process objects (such as motors and sensors) and
sends the information through the DP interface (cables) to the S7 PLC.
• Each signal in the PCS7 system have a unique address’ depending on the
DP-slave address.
• Each unique address has a signal name alias to make it easier looking up
and working with signals rather then using the address’.
• The engineer can program the CPU from both the MPI and Ethernet
interface depending on the situation.
• The SIMBA Pro card holds two PROFIBUS interfaces which can be di-
rectly connected to the S7 CPU’s DP interfaces.
• The SIMIT simulation system is a real time system and the GUI is very
similar to Mathworks Simulink.
• SIMIT uses SIMBA Pro and the PCI-card through a software interface to
communicate to a S7 PLC.
3 DECIDING THE SIMULATOR SETUP 25
This thesis revolves around the GT10C gas turbine, but since there is a need
to extend it to other models, one of perquisites would be to make the simulator
easy to modify and re-engineer when needed. Each delivered turbine comes
with “options”, that is, specific parts which the customer decides. This makes
each turbine unique thus the simulator should be able to be modified to handle
all these different “options” as they are ordered by the customer. Therfore it
should be easy to understand and change the simulation system itself, although
the basic knowledge of automatic control and the SIMATIC system should be
known.
If the PROFIBUS communication between the PLC and the simulator should
be tested, the simulator cannot be in the same PLC as the control program. As
it is now a test can be done using the program in the PLC to try out different
parts of the control system. But since this is only local, the communication in
the cables will not be tested, thus, propagation time, delay and other factors
will be left out of the test making the results inconclusive compared to a real
test of the control program against a turbine.
If the PROFIBUS communication could be included, somehow, all the im-
portant PROFIBUS factors mentioned above could be taken into account to
check if the control program is feasible and also making the simulator more
accurate.
Making sure that the hardware used for the simulator is fast enough to
handle all the signals is also an important factor. Although, it is very difficult
to conclude if the simulator setup finally chosen will be able to handle the load
done by the control program, therfore, one have to do an estimation if it is
possible.
Another important feature of the simulator is to be able to use the original
OS pictures which the engineers at site uses to control an actual turbine with.
That is, it is not just enough to simulate the signals the control program excepts.
It’s also important to be able to use these operating pictures to “control” the
simulator as one does with a turbine. Thus, if both the turbine and the simulator
26
is seen as two black boxes the engineers should be able to “switch” these and
the person operating the turbine should notice no difference at all.
Initially there were discussions between four different solutions (as the su-
pervisor of this thesis suggested). These were devised before the actual thesis
work begun, thus it was not known if they were possible to implement or not;
which is the first thing this thesis will discuss.
The fifth solution was developed at a later stage after tips by SIMATIC
consultants at Siemens.
The cost of another PLC is a major disadvantage. Since the PLC’s are de-
signed to work in rugged industrial conditions they are more expensive. A PLC
costs somewhere around 100KSEK and would thus result in large expenses.
One could argue, however, that the PLC being used to develop the simulator
software latter could be used in a turbine project where it is actually needed,
that is; reused. Thus, just pushing back the investment cost to an earlier date
than needed for an actual turbine project (which would lie forward in time).
If the simulator is to be sold to a third party (as a product) the expense of
the entire setup cost would be to large if a dedicated PLC is to be used for the
simulation.
After some more research into this solution the authors of this thesis came to
the conclusion that it is not feasible. This is mainly due to one single reason; one
have to change the control program itself. Adding and removing features from
the control program would be dangerous and could cause unnecessary confusion.
Obviously, simulating the turbine without changing the control program is the
only solution accepted (that is, if there is no other possible ones, but since there
are this condition can be set).
To bypass this, a possible fix could be to somehow connect the S7 PLC’s
to a PC and program a software which would act as a “switch” and change
the packets being sent and reroute them to the simulator and the other way
around. This is certainly possible but would take extensive time to implement,
surpassing well over 20 weeks, which is the time frame that the simulator should
be developed within.
Thus this solution was discarded at an early stage. See Figure 3.11 for an
schematic of this solution.
3.1 Different solutions 27
Cable
PSU
PSU
DP
CP
DP
CP
S7 S7
• Would the solution mirror an actual turbine in real life since no commu-
nication on physical mediums is used.
Obviously, the problems with this solution does not depend on the PC speed,
rather it depends on if it is feasible to build a simulator program in a program-
ming language such as C++ from scratch.
Certainly, from the authors perspective, designing a basic program which
includes the elements needed, would be possible. PID regulators, different kinds
of flows, action and reaction elements and so on are so basic that they would
be easily implemented in a object oriented language such as C++ or even Java.
Although, this would take time, designing a software is, with all the unknown
factors involved in this project is hazardous. When a certain point is reached
a not so easy obstacle could show itself (that is an obstacle that could take to
long time to breach), since the exact function of the simulation program is hard
to conceive in just a few months, this has to be a evolutionary process, and the
time at hand would not be sufficient.
The work needed of examining the basic requirements, programming the
simulator, and then needing to implement the elements and connecting these in
the simulator could prove to take to long. When at the same time the authors
had to learn the inner workings of a turbine and other issues revolving around
this.
Although, since some other solutions which the authors were working on
simultaneously were not going too well, the decision was to see if a simple sim-
ulation framework could be programmed. The choice fell on the programming
language Java due to its short development cycle for prototyping simple pro-
grams and its object oriented style (which is ideal for a simulation program
where objects in, for example, a industrial process in involved).
A pre-study began, programming a basic interface for a simulation engine.
At the same time a interface had to be done which communicated via a PCI-
card to the PLC itself. For this reason, a CP5614A2 card was bought from
Siemens. This card can simulate a PROFIBUS slave unit and thus be used by
the program to implement a process object.
After a few weeks of examining the API included with the card, the con-
clusion was that the task of actually implementing the slave simulation would
be impossible because the PCI-card only handled one PROFIBUS slave at a
time, and the simulation program would need at least hundred and more to
work. Another PCI-card could probably have been bought which could handle
more slaves simultaneously but since that would have taken even more time,
and because another solution which seemed to work was being done at the same
time (Solution V – SIMIT), the idea was discarded.
The difference between the C++ solution and the Simulink would be to use
Simulink as a simulation framework instead of C++/Java. The problem with
building a interface to the PLC would still be here, i.e. one have to use a PCI-
card as in the C++/Java-solution. Moreover, one have to learn the SimuLink
C++ interface and adapt it to the PCI-cards API-interface.
This solution was actually dropped at an early stage since it would be to
cumbersome to learn both the Simulink API and a PCI CP-card API.
3.1 Different solutions 29
Cable
PC+SimuLink or C/C++
PSU
DP
CP
S7
3.1.6 Conclussions
The final choice is far from obvious. As mentioned in the THEORY chapter,
SIMIT lacks in some areas. The preferred solution would have been using either
a more powerful simulation program such as Simulink or writing a fully new
7 It should be noted, to the confused reader, that, even though this thesis was done at
Siemens, and Siemens themselves are the creators of SIMATIC, internal economic politics
(which says that each department should uphold its own costs) states that the software have
to be payed in full to other departments and companies in the Siemens sphere. I.e. even
though this thesis was done at a Siemens company they still have to pay for Siemens software.
30
program in C/C++/Java. But considering cost versus time the decision was
nevertheless easy to make.
As time progressed it would show that SIMIT was not able to handle the
demands needed. It lacked fail-safe slave support. This is why solution III
(PLC–C/C++) was more carefully examined than most of the other solutions
as a desperate attempt to solve the fail-safe issues.
This shows the problem when using commercial programs: it is hard to
influence the developers to add the support needed for a particular project,
even though they are based within the same company.
The fail-safe problem was solved, if not fully, in an acceptable way, which
made it possible to continue with the project.
As seen in Figure 3.13 the final solution contain two PLC’s, one for the
main program which takes care of most of the systems, while the turbine govern
program PLC contains the regulator of the turbine (i.e. the PID-controllers).
This makes the above decisions of the chosen simulator setup not totaly accurate
since it is done for a setup with just one PLC for the entire turbine control
program; this is no different from using two or even three PLC’s; the reasoning
is still the same.
Solution I II III IV V
Extra PC Yes Yes Yes
Additional
commercial
software Yes Yes
Extra PLC Yes Maybe1
Extra PLC
memory Yes
Interface card Possibly Yes Yes Yes
Tests
Redundancy? Partly No Yes2 Yes2 Yes
Chg. in PLC
program? Maybe Yes
Can it be done
in 20 weeks Uncertain Probably Uncertain Uncertain Probably
Total cost
(KSEK) ∼ 100 ∼ 120 ∼ 15 ∼ 50 ∼ 80
Table 3.7: Quick recap of investment costs for each solution. Notes: 1 A more
powerful PLC might be needed. 2 If implemented.
3.1 Different solutions 31
1
0
SIMIT+Simba Pro
CP
Turbine Govern
PSU
DP
PROFIBUS
S7-414
PROFIBUS
Ethernet
CP
PSU
DP
Main
S7-417
ES
OS
Figure 3.13: Final hardware setup with the turbine PLC’s containing the control
program and the PC containing the turbine simulation program.
32
• Diagram screens.
• Operator screens.
First off, one need to consider that the simulator should be easy to maintain
and improve. Thus the HMI needs to be easy to overlook and understand.
For this purpose a common graphical user interface for all the systems was
designed to make maintainable and improvement tasks easier, as well as using
it.
Each picture contain a version number, the system name and the authors
of the system. This makes co-operation between more than one engineer easier
since version handling is a major subtask that could take much of the develop-
ment time if not done correctly. And since SIMIT does not contain any easy
concurrent engineering system this is a must. In Figure 4.14 the user interface
of an operator screen can be seen. For a diagram screen see Figure 4.19.
Another aspect of a well designed user interface is that each signal group
should be well defined and easy to overlook in a quick way for an engineer. This
actually proved to be a problem; how the signals should be grouped. This was
decided on an individual basis for each system. For example, should alarm signal
toggling be implemented in the same system as it is grouped or in a operator
screen where all alarm signals are located. Here, the decision fell on the first
proposal, since each gas turbine system is well defined by Siemens engineers
and going outside the system grouping would prove non intuitive. This is if an
34
engineer which is familiar with the gas turbine system classification scheme at
Siemens would work with the simulator.
Each signal group is defined the same way both in the operator as well as
the diagram screens to get a more intuitive connection between the two. The
goal was to make the entire diagram scheme visible on the screen at the same
time; although this proved to be an impossible task. Thus, the diagram was
divided into pages (a feature which SIMIT support).
In Figure 4.15 a list of how diagram and operator screens are grouped, in
SIMIT, can be seen.
Figure 4.15: Grouping of diagram and operating screens in SIMIT. The different
alphanumeric codes are an internal Siemens scheme for naming different systems
in the turbine.
• Behavior.
• Interface communication.
4.5 A simulation diagram example (calculating heat flow) 35
Start
Analyze data in
SCADA Pro
Process data in
MATLAB
Analyze system
description
documents
(Updated)
control program
Develop model
from
specifications
Sim . operator
screens
Implement model
in SIMIT
Sim . diagram
screens
Test model
against turbine
PLC
Real time
response
Fills the
Behaviour No
demands?
Interface
communication Yes
Updated
control Yes
program?
No
End
Figure 4.17: Data from performance tests used when developing the model.
Figure 4.18: The relation between the curve taken from SCADA Pro (blue) and
the polygon curve done with Matlab (red).
38
Figure 4.19: The SIMIT model to calculate the heat flow value.
40
the heat flow value increases as well as the speed of the power turbine. And as
the speed increase the pressures, flows and temperatures also increase.
By using this simulator it is very easy to see how the regulators behave at
start up and switching in between. When reaching the maximum load another
phenomenon has to be taken in account. When the regulators works closely
together the exhaust temperature gets very high at max load. Then the T7L
limiter (Section 2.11.4) in operation, makes sure that the temperature is not
increasing to dangerous levels while NGGL (Section 2.11.2) tries to reach the
set point set by the operator.
Here, the model was too fast in the beginning and the system was unstable.
By adding a first order time delay the model was stabilized.
5 RESULT 41
5 RESULT
5.1 Conclusions and discussion
First of all, all action points and questions at hand was not fulfilled. The authors
had to leave out the implementation of other turbine models; one proved to be
more than enough to be able to handle within the decided time frame. This
could be something for future engineers to work on (see the next section).
To summarize things the positive advantage of the solution developed is,
• SIMIT can not simulate signals of higher resolution than 10ms which could
lead to problems when powering up and down the turbine.
Finally, a test against the control program using the model in SIMIT was
performed. Obviously, at first several problems was encountered, thus the model
had to be adjusted. After some more development a successful start of the
turbine was performed using the simulator.
The involved personnel, on the department where this thesis was developed,
was satisfied with the developed model since it simulates a turbine more accurate
than the current.
Finally, a presentation of the model was done for the Siemens personnel from
the research and development departments to verify that the model worked in
a satisfactory way and to show the SIMIT software which has never been used
at this Siemens facility. This resulted in a discussion on how the model could
be used with for example customer training.
The HMI was also seen as simple which would not require to much “know
how” about control systems which would make it simpler for sales personnel to
demonstrate it to potential customers.
As of February 2005 Siemens has decided to improve the simulator by de-
veloping it on other turbine models. This has resulted in the employment of
another thesis worker which are to examine if this is possible.
Siemens has understood the importance of having such an environment as
this to simulate different test cases and system behavior. Also the customers
have started to demand, more and more, that the systems being sold should
42
be stable and bug free. With this simulator setup the chances of this are much
higher. This, in turn, will lead to lower costs when installing the turbines on
site.
References
[1] Abb in australia, September 2004. URL http://www.au.abb.
com/global/auabb/auabb500.nsf!OpenDatabase\char’38db% =/
global/auabb/auabb504.nsf\char’38v=DB6\char’38e=us\char’38c=
5C2BD9E042EE7A41% C1256D86001BEF70.
[2] Fieldbus technology, October 2004. URL http://murray.newcastle.edu.
au/users/students/1999/c9518176/fieldbuste% ch.html.
[3] Profibus - profibus & profinet., June 2004. URL http://www.profibus.
com/profibus.html.
[4] Siemens AG. S7-400 and m7-400 programmable controllers
- hardware and installation.
, September 2004. URL http://www.siemens.co.jp/simatic/japan/as/
plc/data/400/424ish\_e.pdf.