Power Plants and Boilers Models for Operator Training ... · Power Plant and Boiler Models for...
Transcript of Power Plants and Boilers Models for Operator Training ... · Power Plant and Boiler Models for...
Power Plant and Boiler Models for Operator Training Simulators
P. Neuman
NEUREG, Plc., 193 00 Prague, Czech Republic
(Tel: +420-777-648-906; e-mail: [email protected]).
Abstract: A specific Operator Training Simulator for Power Plant’s Operational Staff (Boilermen,
Turbine Drivers, Engineers) is described in detail. The objective Power Plants Dynamic Models for
operator´s training in Czech Republic are used also in the first standard utilized Operator Training
Simulator (OTS) of coal-fired Czech Power Plants Opatovice. The Power Plants consist of six boilers
with steam capacity 250 t/h of each, common steam collector and of six turbogenerators with power 60
MWe each. Total heat capacity is 800 MWh. The process model is developed on the basis of
mathematical-physical analysis (”first principles” method) of the individual technological subsystems.
Dynamic model of the OTS consists of complex thermal power unit, i.e. of two steam boilers working
into one section of the common steam collector and of two steam turbines-generators. The process model
describes the standard and abnormal operation regimes in the range of 0 % - 100 % of Maximum
Continuous Rate, with implementation of individual disturbances by the instructor. The own-built library
called “EnergySIM” was developed. on the basis of MATLAB-SIMULINK packages.
Keywords: Power Plants, Steam Boilers, Pulverized Boiler, Fluidized Bed Boiler, Dynamic models,
Engineering and Operator Training Simulators
1. INTRODUCTION
The Technology Object Oriented Modeling describes each
part of the model as an object with certain behaviour
(Maffezzoni, 1992). To comply with this concept, an own-
built library, called “EnergySIM”, has been partially
developed (based on MATLAB - SIMULINK). The main
assets of „EnergySIM“ are the following ones:
- It is a modular system where the final model is obtained by
assembling instances of general-purpose models taken from
the library.
- The system is open in the sense, a user is free to expand the
default libraries with her own models.
- The library “EnergySIM” includes the models and modules
of many typical components that can be found in
thermoelectric power plants.
2. SIMULATOR CONFIGURATION
The OTS is of “partially stimulated type” and it is created
from modules of own NEUREG library called “EnergySIM”.
This library could be used for modelling all types of power
plants (Conventional Power Plants, Combined Cycle Power
Plants, etc.) based on the following modules: Drum Boilers,
Steam Turbines, Superheaters, Reheaters, Once-Through
Boilers, Fluidized Bed Boilers and Gas Turbines. The
“distributed control system” is emulated in MATLAB-
SIMULINK (Neuman, et al., 2003), too.
The basic information for creating structure of an OTS is
represented in a process diagram. It is demonstrated on an
example of a feeding water subsystem. This part is depicted
in Fig.1, that represents the Process Instrumentation Diagram
(P&ID). Other subsystems modelled in the same way are the
following ones: Feeding Water, Air Supply Loop, Coal
Feeders & Mills, Flue Gas Loop, Super Heaters, Common
Steam Collector, Steam Turbines & Generators (Neuman, et
al., 2002). In the P&ID, all the objects are depicted that are
either manipulated by control or where some measurement is
performed. It is typical for an OTS that all the information,
available in monitoring and control system, must be present
also in the models used in OTS. It is in this case only, that the
operator can manipulate all the valves, fans, switches etc. as
if he would do it in a real plant. This specific requirement
brings a need to have technology object-oriented models of
all such elements, that can be seen in the process diagram
(Neuman, et al., 2004).
3. MODELING OF SUBSYSTEMS
The principles for modelling are described in reference
(Ordys, et al.,1994). According to these principles, the library
“EnergySIM” includes the following modules: Gas and
Steam Turbine Units, Two Phase Units (HRSG, boiler-
evaporator, condenser, deaerator tank), Single Phase Units
(economiser, superheater), special Drum and Once-through
Evaporator Units, Water Storage, Media Flow Resistance
Unit, Heat Flow Resistance Unit, Burner Units, Combustion
Chamber Units, etc. Each module is derived from the three
conservation physical laws (“first principles”) and it needs a
few parameters. As the description of all modules would be
out of scope of this paper, only some basic modules are
shortly described.
3.1 Feeding Water Model
Preprints of the 18th IFAC World CongressMilano (Italy) August 28 - September 2, 2011
Copyright by theInternational Federation of Automatic Control (IFAC)
8259
The feeding water scheme, depicted in Fig.1, correspons to
P&ID diagram in Fig.2. and models the following processes:
Water is supplied through a feeding pump, which has been
modelled by its flow/pressure characteristics. The water is
pre-heated in two high pressure heaters (VTO) in series
using steam as heating medium. Cold water can also be
mixed in parallel with hot water from VTO. Then, the water
is passed through series of two primary (EKO I) and four
secondary economizers (EKO II). As the secondary
economizers are of special construction, they must be
modelled in two parts. Heated water after primary
economizer is also supplied as spray to the super heaters to
control the steam temperature. The pressurized water is
nearly at the saturation point and then, it is supplied to a
drum boiler where it is evaporated to steam. The evaporating
processes in the drum are difficult to describe and therefore,
three alternative models are offered for Engineering
Simulators (Neuman, 1997) or Operator Training
Simulators:1 - a simple second order model (without “swell
and shrink” phenomenon), 2 - a modified fourth order
“Aström” model (Astrom and Bell 2000), keeping the
dynamic behaviour of the drum boiler sufficiently simple for
modelling, but with reasonable precision, or 3 - a higher
order model necessary for realistic modeling of particular
operating conditions, like low-loads, fast start-up, or
abnormal shutdown (Neuman, et al., 2002).
The feeding water flow rate to the drum is controlled by the
feeding valve (NV) according to the water level in the drum
boiler, feeding water flow and steam flow from the boiler,
and pressure difference across is the feeding valve is
maintained by a differential valve (DV) – In Fig.1, it is
situated in the blocks V3_DV_NV1, V3_DV_NV2”,
whereas in Fig.2, these valves V01A / AG61,AG60 are
behind VTO II in series with NV. All the other valves realise
the switching functionality necessary during the swaping
between hot and cold-water branches. They are modelled but
not mentioned in this description. The P&ID diagram of this
feeding water subsystem is depicted in Fig.2, where each type
of described elements has its own object oriented block
realization.
Fig.1 SIMULINK scheme of the feeding water subsystem
(Drum Boiler)
Fig.2. The original P&ID Diagram Feeding Water
Preprints of the 18th IFAC World CongressMilano (Italy) August 28 - September 2, 2011
8260
3.2 Steam Turbine Model
3.2.1 Simple IEEE Steam Turbine Model
The simple IEEE dynamic model of steam turbine, presented
in Fig.3, is derived from the fact (see Eq. (1) that power of
turbine NT depends generally on product of three variables:
steam flow rate MT , (enthalpy) isentropic heat gradient H0
and internal thermodynamic efficiency ηTD . Those models
are used in Power Systems Model, e.g., in so called Network
Simulators (Neuman, et al., 1999).
NT = MT . H0 . ηTD (1)
Variable H0 depends on steam parameters (temperature,
pressure). In IEEE model, steam is supposed with constant
parameters (i.e. constant enthalpy). This approach
corresponds to standard operation of the block at nominal
parameters. However, for OTS purposes , this assumption is
completely unrealistic for modelling the power turbine due to
large changes in power during the start up and shutdown.
Variable ηTD depends on turbine speed in two ways: a) partly
through the friction and ventilation losses, b) partly through
the differences in speed triangle which describes steam inlet
into the blades of turbine. The effect caused by suggested
small changes of turbine speed in the range of +10 %
(enough for IEEE engineers) is also neglected, because
successive changes in efficiency are very small, practically
in unit percentage.
Dynamics of steam during flow through the individual parts
of turbine is described on the basis of steam expansion in
closed volumes. This dynamics is modelled as 1-st order
transfer functions. Time constants depend on volume of
relevant part of turbine, initial value of steam pressure ,
steam flow rate and also there exists a dependence of
specific steam volume on steam pressure. It means that time
constants are not constants any longer but that they depend
on the “working point” and its changing with the load.
However, these time constants are constant in the IEEE
models. The coefficients kLP and kHP respect the ratio of
energy production in the Lowpressure Part (LP) and
Highpressure Part (HP) of turbine.
Governor Output
Turbine Output
Mid and Lowpressure Part
Highpressure Part
R T
+ V T
1
v min
v max
-
p
1
G min
G max
Control Valves
Intercept Valves
+ T p 1
1
HP
k HP
+ T p 1 LP
k LP
P T M T
IV k
0
1
IV T
1
v Imin
v Imax
+ -
Fast Valving
p
1
0
1
v CStop
v IStop
T
1
R
+
+
+
N T
Reheater
BOILER
Fig.3. Block scheme of the simple IEEE steam turbine
model
The IEEE steam turbine model shows the turbine with steam
reheater, which is usually used for turbines in Czech
Republic. These turbines have more parts and the steam flow
rate is coming back into boiler reheater after expansion in HP
parts. For calculation of IEEE turbine dynamics, the above
mentioned simplifications are used, which are suitable only
for RT model of interconnected power systems, but not
acceptable for power unit training simulator. Therefore, the
specific realistic model of condensing steam turbines without
reheaters was developed.
3.2.2 The realistic Steam Turbine Model
Generally, in the case of the steam turbine, the mechanical
power is obtained by removing energy from the superheated
steam stream by expanding it to a lower pressure. In our case
steam enters the Highpressure Part HP Lowpressure Part LP
chest via the main admission control valves. After passing
through the HP stage the steam passes through the Low
Pressure (LP) stage before returning to the condenser.The
main modelling assumptions are as follows:
- Superheated steam is treated as an ideal gas.
- HP, LP, turbine stages are converted to equivalent nozzles
through which one-dimensional uniform polytropic steam
expansion takes place. - Energy storage volumes are lumped
- Inlet kinetic energy of steam to each stage can be neglected.
- Mass flow dynamics between input and output is modelled
as a 1st order lag.
The steam turbine model is sub-divided into two sections. A
section is defined here as consisting of a lumped steam
storage volume carrying dynamics, followed by a complete
turbine stage that is modelled by steady-state relationships.
Note that a complete turbine stage (HP, LP) comprises of a
number of impulse and/or reaction stages in series. High
pressure section, as an example, is described by following
equations:
- three differential equations of steam mass balance – eq. (2),
section mass flow dynamics, and heat balance – eq. (3),
- twelve algebraic equations, e.g. nozzle equation solved for
rate of nozzle pressure drop rhp. – eq. (5). There are five
inputs, seven parameters, one constant, and five outputs.
The following equations (2), (3), (4), (5), are selected as an
example from the whole complex set of equations. These
equations are implemented in SIMULINK (Neuman, et al.,
2003).
Mass balance equation has form:
ohpi wwdt
dV 0 (2)
Equation of conservation of energy (heat balance) has form:
000 hwhwhdt
dV ohpii (3)
Equation of perfect gas has form:
000
00
TRp
Tc
hhT in
p
in
(4)
Nozzle equation has form:
Preprints of the 18th IFAC World CongressMilano (Italy) August 28 - September 2, 2011
8261
m
m
pA
wrr
hphp
ohpm
m
hpm
hp 2
1
00
212
(5)
where:
V - chest storage volume, 0T - outlet steam temperature,
pc .- specific heat in constant pressure, 0 - outlet steam
density from the chest, 0h - outlet steam specific enthalpy
(from the steam chest), ih - inlet steam specific enthalpy,
iw - inlet steam flow, ohpw - outlet steam flow, hpr - ratio
of nozzle pressure drop, m - exponent of polytropic curve
for superheated steam, n - polytrope exponent for wet steam
3.3 Heat Exchanger Models
It is documented on principle scheme of general heat
exchanger– see Fig.4. The more detailed description of I/O
specification and parametrization is available in commercial
documents and customer manuals (Neuman, et al., 2004).
Chamber 1 RMF
RMF
RHF
in
in out
out
Chamber 2
Fig.4. Principle scheme of Heat Exchanger layout (where:
HF stands for heat flow, MF is mass flow)
The principles and parameters of economiser modules ECO
are following ones: V is volume, m is mass, Ai is cross
sectional area, Cp is specific heat, “alpha_i” and “alpha_o”,
are coefficients and variables of heat transfer calculation - see
Economizer menu, in Fig.5. There can be also seen the
following parameters: ho is enthalpy, Tm is mean
temperature of wall, T is temperature of wall surface.
The final goal is to develop reasonably complex non-linear
models that capture the key dynamical properties over a wide
operation range. Building a model from this library is very
easy and during modelling one can feel that he is constructing
an AutoCAD drawing of the model. In this way, every one is
able to build a model of his own choice very easily and
without deep knowledge of modelling. The modelling of all
the modules is based on the three-conservation laws, Mass,
Energy and Momentum balances (Dolezal and Varcop 1970).
To make the modelling a user-friendly process, a dialogue
box, as shown in Fig.5, will appear when user clicks the
mouse over the model (object). Initial conditions and
parameters of the block can be entered through this display
window. The chambers are divided mainly in two groups.
The first one in which the “water and steam flows”, and the
second one in which “air and flue gases flow”. Data in form
of a vector of ten variables are obtained from each unit; in
which first six (flow, enthalpy, pressure, temperature, quality,
total head) are common to all chambers and other four
depend on chamber (e.g. in boiler they are level, volume of
water, volume of steam and steam quality at the riser exit). In
case of flue gases, data in form of a vector of five variables
(flow, enthalpy, pressure, temperature, heat) are obtained.
Fig.5. Economizer menu for parameter setting
The primary unit of the library is given by the name unit
specified by the name which is used for its invocation and
has its corresponding menu e.g. Economiser in Fig.5
For example, boiler chamber, valve, pump etc. are units.
Units are combined in a sub-model. E.g. feeding water,
where there is a heat exchanger as shown in Fig.4, is a sub-
model. In chamber 1, water enters and leaves as steam
passing through a flow resistance, while in chamber 2, hot
flow gases enter and leave with heat loss through a flow
resistance. These two chambers are connected through a heat
resistance. Similarly, ECO is a “user application” of the
submodel “heat exchanger”.
4. EMULATION OF CONVENTIONAL BASIC LEVEL
CONTROL
The Distributed Control System (DCS) consists of manual
control from operator control panel (MOSAIC), control from
SCADA/HMI mimics and control from standard industrial
PID controllers. An example of the fully emulation scheme of
real DCS can be seen in Fig.6, where two cascade PID
controllers (blocks REG.1, REG.2) and logic control modules
(eg. Blocks 1.2, 1.3, 5.3) for level control in the drum have
been modelled (emulated) in MATLAB-SIMULINK.
Switching between manual control and PID control is given
either by operator (from SCADA/HMI or from MOSAIC) or
by a protection logic, where e.g. Feeding Water Valve (NV)
can be in PID control mode only when Difference Pressure
Valve (DV) is also in PID control mode – see Fig.2, Fig.1.
This switching is shown on bottom half of the Fig.6.
Preprints of the 18th IFAC World CongressMilano (Italy) August 28 - September 2, 2011
8262
Fig.6. Control system for water level control in drum boiler
All the PID controllers (blocks REG.1, REG.2 – in Fig.6) are
realized by Z-tranformation of the continuous PID controller.
Equation for the continuous PID is following one:
ddeT
r
dt
tdeTrtertu
I
D0
00 (6)
It is tranformed by trapezoidal method into discrete version
of controller (Balate, 2004):
0
1
1
2
2 qzqzqu (7)
where T is the sampling period
I
D
I
D
D
T
TrTTrrq
T
TrTTrrq
TTrq
0
000
0
001
02
5.0/
5.0/2
/
(8)
Dynamic simulation model can also be utilized in designing
the control system for operated processes. However the
control scheme and control algorithm have to be the same as
in real operation distributed control system, because OTS
must be fully realistic.
The development and verification of some advanced control
algorithms for the engineering dynamic model is also
possible but only in the frame of Engineering Simulators.
E.g., the extended Luenberger’s state controller-observer of
superheated steam temperature was developed for coal
pulverized steam boiler of Power Plant Shen Tou 500 MW,
in District Shan Si, China (Neuman, et al., 1988).
5. SIMULATION AND OPERATIONAL RESULTS
The OTS simulation transients displayed in SIMULINK and
SCADA/HMI are in following figures 7, 8, 9, 10.
Fig.7. Drum boiler water level transient in SIMULINK (time
scaling: 1 unit = 0,1 sec)
Fig.8. Drum water level transient in SCADA/HMI - OTS
The operational transients from real time process control is
very similar to the above mentioned simulation transients
(see Fig. 11), only a little slower. It is in convenience with
declared accuracy of a dynamic model of power unit.
E.g. , when controlling the level in the drum, the change of
steady state value dHdes = 2 cm has the simulated transition
time Tpr = 120 s, with change trend dH = 1 cm/13 s. The
values from the real process are Tprp = 160 s, and change
trend dHp = 1 cm/18 s.
Preprints of the 18th IFAC World CongressMilano (Italy) August 28 - September 2, 2011
8263
The similar simulation accuracy is achieved for controlling
the temperature of overheated steam (the simulated change
trend is dT = 1 °C/10 s), for the change of steady state value
dTdes = 5 °C. The same variable measured on the real process
is dTp = 1 °C/16 s.
Fig.9. Steam temperature transient in SIMULINK
Fig.10. Steam temperature transient in SCADA/HMI - OTS
Fig.11. Drum water level transient in real process
6. CONCLUSIONS
Operator Training Simulators require the high realism of the
models. Therefore the parametrization and verification of
model´s features were performed only with participation of
the very experienced operators of Czech Power Plants
Opatovice.
In the contemporary performed work, the circulating
fluidized bed (CFB) boiler model is developed (Majanne and
Köykkä, 2009). The simulator is also built with MATLAB
SIMULINK software. The model is also based on the mass,
energy, and momentum balances together with physical
behaviour of heat transfer, reaction kinetics etc. Dynamic
modelling and simulations have been used to examine the
operation of power plant with circulating fluidized bed boiler.
Hitherto simulations give a clear representation of CFB.
One of the future research subjects in this field is the
advanced intelligent control system design (Vasek, et al.,
2010) for a pilot operator training simulator OTS with
circulating FBB, which is in preparation process.
REFERENCES
Astrom, K. J. and Bell, R. D. (2000). Drum-boiler dynamics.
Automatica, Vol. 36, pp. 363 – 378.
Balate, J. (2004 ). Automatic Control (in Czech). 2nd
edition
BEN, Praha.
Dolezal, R. and Varcop, L. (1970). Process Dynamics –
Automatic Control of Steam Generation Plant.
Elsevier, London.
Maffezzoni, C. (1992). Issues in modeling and simulation of
power plants. In Proceedings of IFAC symposium
oncontrol of power plants and power systems, Vol. 1.,
pp 19-27, Munchen, Germany.
Majanne, Y. and Köykkä, P. (2009). Dynamic Model of a
Circulating Fluidized Bed Boiler. In Proceedings of
IFAC Symposium on Power Plants and Power
Systems Control, Tampere, Finland.
Neuman, P., Stecha, J. and Havlena, V. (1988), State
controller with observer design for superheater
temperature control. Preprints of 4th
IFAC Symposium
on Computer Aided Design in Control Systems
CADCS´88, Beijing, China.
Neuman, P. (1997), Engineering Simulator for Fossil Power
Plant. Preprint IFAC/CIGRE Symp. on Control of
Power Systems and Power Plants, Beijing, China.
Neuman, P., Máslo, K., Šulc, B. and Jarolímek, A. (1999).
Power System and Power Plant Dynamic Simulation.
In: Preprints of 14 th
IFAC World Congress, Volume
O, paper O-7c-04-02, pp. 179-184, Beijing.
Neuman, P., Pokorny, M., Varcop, L., Weiglhofer, W. and
Javed A.J. (2002). Engineering and Operator Training
Simulator of Coal-fired Steam Boiler. Proc. 10th Int.
Conference MATLAB´02, Prague, Czech Republic.
Neuman, P., Pokorny, M., Varcop, L. and Weiglhofer, W.
(2003). Operator Training Simulator of Coal-fired
Power and Heating Plants. Proc. 11th Int. Conference
MATLAB´03, Prague, Czech Republic.
Neuman, P., Pokorny, M., Varcop, L. and Weiglhofer, W.
(2004). Operator Training and Engineering Simulator
of Fossil-fired Power and Heating Plants. Proc. 6th
Int. Conference CONTROL OF POWER
SYSTEMS´04, Strbske Pleso, Slovak Republic.
Vasek, V., Neuman, P., Balate, J. and Vargovcik, L. (2010).
Intelligent Control System for Municipal District
Heating Networks. Annual Reports for Ministry of
Education Czech Republic, Nr. 2C06007, Prague,
Czech Republic.
Ordys, A.W., Pike, A.W., Johnson, M.A., Katebi, R.M. and
Grimble, M. J. (1994). Modelling and Simulation of
Power Generation Plants. London 1994, Springer-
Verlag.
Preprints of the 18th IFAC World CongressMilano (Italy) August 28 - September 2, 2011
8264