Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Applied Energy 85 (2008) 1173–1189

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Modeling and investigation start-up procedures of a combined


cycle power plant
Falah Alobaid a,*, Ralf Postler a, Jochen Ströhle a, Bernd Epple a, Kim Hyun-Gee b
a
TU Darmstadt, Energiesysteme und Energietechnik, Petersenstrasse 30, 64287 Darmstadt, Germany
b
DOOSAN Heavy Industries and Construction, Boiler System Development Team, Changwon, South Korea

a r t i c l e i n f o a b s t r a c t

Article history: This study describes the particular development and investigation of a static and dynamic
Received 25 January 2008 simulation model and its application to improve the start-up process of a combined cycle
Received in revised form 3 March 2008 power plant. Generally, the power plant system and control design mean highly complex
Accepted 7 March 2008
interactivities. The dynamic simulation models using powerful computers are effective
Available online 25 April 2008
tools for studying and understanding the operating characteristics of power plants to meet
and improve the design, control strategy and operational requirements. The heat recovery
Keywords:
steam generator (HRSG) is modeled by using commercial simulation software named
Heat recovery steam generator
Control circuit
advanced process simulation software (APROS). The HRSG model includes an advanced
Dynamic simulation control philosophy and turbine bypass systems to have a high level of accuracy, especially
Start-up improvement during hard transients. The comparison between the simulation results and measured data
is documented. The received results proved and embodied that the simulation is both very
reliable to estimate the real HRSG dynamic behaviour and capable to predict the opera-
tional processes. Through a parametric study, the start-up time will be reduced while keep-
ing the life-time consumption of critically stressed components under control.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction

As the electricity consumption has risen quickly over the past few decades, the number of thermal power plants has in-
creased worldwide. A heat recovery steam generator (HRSG) produces steam by using the heat from exhaust gas of the gas
turbine and feeds it to steam turbine. This combination produces electricity more efficiently than either the gas turbine or
steam turbine alone which causes a very good ratio of transformed electrical power per CO2 emission. The combined cycle
power plants are characterized as the 21st century power generation by their high efficiency and possibility to operate on
different load conditions by reason of the variation in consumer load. An increasingly important utilization for the combined
cycle power plants is the compensation of varying electricity feed-in from renewable energy sources like wind power. A
combination between wind power and other energy sources such as combined cycles with improved load flexibility should
be used in order to mitigate the economic effects of wind variability.
With the availability of high powerful processors and advanced numerical solutions there is a great opportunity to devel-
op high performance simulators for modeling energy systems in order to consider various aspects of the system. The HRSG
dynamic simulation has not been performed as frequently as the steady-state studies. One example for dynamic behaviour is
the fast start-up process of the HRSG as a major interest for designers. It is a complex task including several limitations that
have to be fulfilled simultaneously. An important restriction hereby is the maximum allowed thermal stress of the HRSG
components caused by temperature gradients, like the wall temperature gradient of drum or superheater outlet header.

* Corresponding author. Tel.: +49 (0) 6151 16 3791; fax: +49 (0) 6151 16 5685.
E-mail address: falah.alobaid@est.tu-darmstadt.de (F. Alobaid).

0306-2619/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2008.03.003
1174 F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189

Nomenclature

Cp specific heat capacity (kJ/kg K)


D diameter (m)
DH hydraulic diameter (m)
F force (N)
f friction factor for phase (–)
gz standard gravity (9.81 m/s2)
h specific enthalpy (kJ/kg)
k Stodola coefficient (–)
m mass (kg)
m_ mass flow rate (kg/s)
p pressure (MPa)
Q heat flow (kW)
T temperature (°C)
t time (s)
u velocity (m/s)
Z elevation (m)
a volume fraction (–)
C mass transfer (kg/m3 s)
g isentropic expansion efficiency (–)
v specific volume (m3/kg)
q density (kg/m3)
r thermal stress (N/m2)

Subscripts
a annular flow
b bubbly flow
d droplet flow
E rate of entrainment
en flow with entrainment
fl form loss
g gas phase
i interface between phases
k liquid or gas phases
l liquid phase
ne flow without entrainment
ns non-stratified flow
pu pump
R rate of stratification
sat saturated
va valve
W wall

Abbreviations
APROS advanced process simulation software
DC device control
FG flue gas
GT gas turbine
HRSG heat recovery steam generator
HP high pressure
HP_BFP high pressure boiler feed pump
HP_Drum high pressure drum
HP_FW high pressure feed water
HP_SH high pressure superheater
IP intermediate pressure
IP_BFP intermediate pressure boiler feed pump
IP_Drum intermediate pressure drum
IP_FW intermediate pressure feed water
IP_SH intermediate pressure superheater
LP low pressure
LP_BFP low pressure boiler feed pump
F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189 1175

LP_FW low pressure feed water


LP_Drum low pressure drum
LP_SH low pressure superheater
PI proportional-integral controller
RH reheater
ST steam turbine

All in all the dynamic simulation is a critical procedure when operating systems are close to the limits either of the process,
the materials, the emissions or the economics. This needs strong requirements on both the quality of models and their
numerical solution with respect to accuracy and efficiency.
Recently, several studies [1–5] have been made on the prediction of the single pressure HRSG transient performance or
the dual pressure HRSG [6]. Further simulation studies deal with modeling of fast start-up of the triple pressure HRSG by
using different simulators. For example, the combination between dynamic simulation and nonlinear programming is pro-
posed by Shirakawa et al. [7]. Another study has been modeled by using Modelica simulator [8].
In this work, a computing model of an existing HRSG is build up with advanced process simulation software (APROS)
[9]. The HRSG model verification and evaluation with design data for steady state is performed. After reaching the appro-
priate accuracy in steady state, dynamic simulation and comparison with measured data are settled. The comparison be-
tween the measured and simulated data for each pressure circuit is performed in detail, which has not been considered in
the previous studies. Once having a validated model, a parameter studies with start-up time reduction have been
conducted.

2. Description of the real power plant

The combined cycle power plant under investigation consists of a horizontally based gas turbine connected to a vertical
gas path. The exhaust gas temperature of the gas turbine is 628 °C. Hence no additional flue gas heating with a channel bur-
ner application is necessary. The water/steam circuits of the HRSG consist of three pressure systems with a forced circulation
in the evaporator path and a reheater section after the high pressure turbine. The process diagram of the combined cycle
including the gas turbine and the three pressure circuits of the HRSG is shown in Fig. 1.
Table 1 includes the characteristic technical data of the real HRSG for gas and water/steam sides.

Fig. 1. HRSG flow diagram.


1176 F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189

Table 1
Technical data of the real power plant

Turbine steam inlet conditions Flue gas inlet conditions Flue gas outlet temperature
HP RH/IP LP
p = 97.7 bar p = 21.4 bar p = 4.1 bar _ ¼ 587:3 kg=s, T = 628.3 °C
m T = 81.6 °C
T = 567 °C T = 567 °C T = 293 °C
m_ ¼ 77:4 kg=s m_ ¼ 83:15 kg=s m_ ¼ 9:8 kg=s

3. Description of the HRSG model

The HRSG model is build up in commercial software package APROS developed by VTT, Finland. APROS allows full scale
modeling and dynamic simulation of different processes in real time. The process related component data are archived and
maintained in different model libraries and are based on non-linear equation systems that simulate the real-life process sta-
tus with a very high accuracy throughout the entire operating range of the plant [9].

3.1. HRSG flow model

The HRSG model includes different thermal hydraulic modules, i.e. the homogenous and the six-equation model [9] as
shortly described below.

3.1.1. Homogenous flow model


나비에스
This flow model is suited for one phase flow heat exchangers either water or steam such RH system, turbine 스톡스
sections,
방정식에서 of
economisers, etc. The homogenous solution system of APROS is based on the one-dimensional conservation equations 점도 = 0
mass, momentum and energy. -> 오일러 방정식
Density / time + Density x Velocity / Elevation
oq oqu
Mass balance : þ ¼0 ð1Þ
ot oz
Force
oqu oqu2 op
Momentum balance : þ þ ¼ qg z þ F w ð2Þ
ot oz oz
oqh oquh op
Energy balance : þ ¼ þ Qw ð3Þ
ot oz ot
The terms Fw and Qw denote friction force and heat flow through walls. The equations are discretized and non-linear terms
are linearized. Void Fraction

3.1.2. Six-equation model


This flow model is suited for two phase flows water/steam heat exchangers, where there is slip between the phases such
evaporator and condenser systems. The use of the two-phase flows six-equation model requires knowledge of mass, momen-
tum, and energy transfer between the phases. This transfer can be expressed from the flow parameters and their derivatives.
The six-equation solution system of APROS is based on the one-dimensional conservation equations of mass, momentum and
energy. When the equations are applied for the liquid and gas phases, a total six partial differential equations are used.
oðak qk Þ oðak qk uk Þ
Mass balance : þ ¼ Ck ð4Þ
ot oz
oðak qk uk Þ oðak qk u2k Þ op
Momentum balance : þ þ ak ¼ Ck uik þ ak qk g z þ F wk þ F ik þ F va þ F fl þ DP pu ð5Þ
ot oz oz
oðak qk hk Þ oðak qk uk hk Þ op
Energy balance : þ ¼ ak þ Ck hik þ Q ik þ Q wk þ F ik uik ð6Þ
ot oz ot
In these equations, the subscript k is either l = liquid or g = gas. The subscripts i and w refer to interface between two phases
and wall, respectively. The term C is the mass exchange rate between phases (positive value means evaporation and negative
value condensation). The terms F and Q denote friction force and heat flow. The last three terms of the momentum equation
are valve friction, friction from form loss and pump head. In the energy equation the term h is the total enthalpy including
the kinetic energy.
The interfacial heat transfer Qik in Eq. (6) is calculated separately for the liquid and gas phases.
Gas : Q ig ¼ kig ðhg  hg;sat Þ ð7Þ
Liquid : Q il ¼ kil ðhl  hl;sat Þ ð8Þ
The evaporation or condensing mass flow rate C is calculated by forming the energy balance for the interface.
Q il  Q ig
C ¼ Cg ¼ Cl ¼ ð9Þ
hg;sat  hl;sat

Separate heat transfer correlations are used for evaporation hk > hk,sat and condensation hk < hk,sat cases.
F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189 1177

For the wall heat transfer Qwk in Eq. (6), there are separate heat transfer correlations for three heat transfer zones wetted
wall, dry wall and a transition zone between wetted wall and dry wall. If the wetted wall heat transfer is selected, only the
liquid phase is assumed to be in contact with the wall of the flow channel. In the dry wall zone only the gas phase touches the
wall. The critical heat flux and minimum film boiling temperature are needed for the selection of the heat transfer zone. The
heat transfer correlations usually calculate heat fluxes in [W/m2]. The heat flux is converted into volumetric heat flow [W/
m3] by multiplying with the heat transfer area (surface area of the wall) and dividing with the volume of the calculation
node.
The interfacial friction Fik in Eqs. (5) and (6) (the friction between the liquid and gas phases) is strongly dependent on the
flow regime that is prevailing in the flow. Different interfacial friction correlations are used for the different flow regimes.
The modeled flow regimes are stratified flow and non-stratified flow consisting of bubbly, annular and droplet flow. The final
value for the interfacial friction is then obtained as a weighted average of the different correlations. Void fraction, rate of
stratification and rate of entrainment are used as weighting coefficients. The interfacial friction is calculated using different
friction correlations for different flow regimes and rate of stratification and entrainment R, E respectively.
F ik ¼ RF ist þ ð1  RÞF ins ð10Þ
The interfacial friction Fins in non-stratified flow is:
F ins ¼ ð1  EÞ½ð1  aÞF ib þ aF ia  þ EF id ð11Þ
The interfacial friction Fist in stratified flow is:
0:01½1 þ 75ð1  aÞqg DujDuj
F ist ¼ ð12Þ
DH
The forces Fia,Fib and Fid correspond to the interfacial friction in annular, bubbly and droplet flows respectively. The friction
Fwk between one phase (liquid or gas) and the wall of the flow channel in Eq. (5) is calculated with the formula:
2f k qk uk juk j
F wk ¼ ð13Þ
D
The quantity fk is the friction pressure loss coefficient or friction factor for phase k.

3.1.3. Simulation of steam turbines


The steam turbine simulation is calculated from momentum and energy equations (2 and 3) because the homogenous
flow model is used. The pressure and enthalpy drops are added as source terms in the both equations. The pressure drop
in steam turbine is calculated as:
1
Dp ¼ kqw2 ð14Þ
2
whereas k is the loss coefficient in APROS
k ¼ f ðKÞ ð15Þ
The Stodola coefficient K is used to calculate the pressure loss coefficient over the turbine section.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
K¼m _ p1 v1 =ðp21  p22 Þ ð16Þ

where m_ is the mass flow rate, p1 is the inlet pressure of the turbine, p2 is the outlet pressure of the turbine and v1 is the
specific volume before the turbine.
The enthalpy drop over the steam turbine is calculated as:

DHs ¼ ½h1s  ðh1s  h0 Þðp2 =p1 Þg=4:27  h0 m ð17Þ


In Eq. (17) the subscript ‘s’ refers to steam, 1 to the state before the turbine, and 2 to the state after the turbine. h0 is the
reference enthalpy (1950 kJ/kg).

3.2. HRSG model

The APROS model [10,11] includes the water/steam paths in a high level of detail while the gas turbine section is modeled
as an inlet boundary condition for the HRSG model with the characteristic parameters temperature, mass flow, pressure and
flue gas composition. In APROS the flue gas path is modeled from the gas turbine exit to the HRSG exit. On the water/steam
side all the bundle heat exchangers namely superheaters, reheaters, evaporators and economisers have been implemented
with real geometry data. Furthermore for each pressure circuit turbines, condenser, drums, pumps, and valves are modeled
either. Significant work has been done for building in, testing and improving the control circuits that must control all basic
functions in the HRSG model. Applying realistic parameters of controllers is fundamental for getting correct response of
those control circuits in the dynamic behaviour, such as drum level, turbine steam bypass, attemperator controllers, etc.
The final HRSG model represents a very high level of detail and accuracy. As an example the low pressure and turbine nets
1178 F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189

are shown in Figs. 2 and 3 respectively. The low pressure (LP) net is the main circuit in the HRSG and it consists of the fol-
lowing heat exchangers: LP superheater, LP evaporator and LP economiser.
The low pressure feed water led to the LP economiser is supplied from the condenser. A drum level control valve is located
between the LP economiser outlet and the LP drum. The water of the LP drum circulates through the LP evaporator where is
heated by the flue gas and converted into saturated steam in the LP drum. The saturated water in the LP drum is fed to feed
water pumps of the HP and IP circuits. The dry steam exits the LP drum and flows through the LP superheater.
The turbine net consists of three turbine stages HP, IP, and LP. The HP superheated steam enters the HP turbine with
567 °C and 100 bar. After leaving the HP turbine the cold and partly expanded HP steam mixes with the IP superheated
steam and enters the reheater section before entering the IP turbine. At the outlet of the IP turbine the further expanded
steam is mixed with the LP superheated steam and is led into the LP turbine. From the outlet of the LP turbine the fully ex-
panded steam flows into the condenser.

Fig. 2. Low pressure circuit.

Fig. 3. Turbine net.


F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189 1179

Fig. 4. HP drum level control circuit.

As representative examples for control circuits, the HP drum level and IP bypass control circuits are explained. The control
circuit of the HP drum level depicted in Fig. 4 regulates the HP feed water mass flow by controlling the HP feed water valve,
which is located upstream of HP Economisers. The operation mode of HP drum level control circuit is described in the
following:

(a) The difference between the HP feed water mass flow (1) and HP steam mass flow (10 ) dm _ is measured, filtered and
amplified.
(b) The difference between the HP drum level set point (20 ) and actual value (2) dL is measured and amplified.
_ dL) affects the PI controller (4).
(c) Deviation (3) of these two paths (dm;
(d) PI controller (4) commands the DC continuous device (5).
(e) DC continuous (5) device operates the HP_FW valve.
(f) Tracking feature of PI controller makes the control circuit stable and responds milder (6).
(g) Correction from HP drum pressure will affect the HP drum level control circuit in case of shut down/start-up processes (7).

The IP bypass control circuit consists of two control circuits. The first circuit is a binary one which measures the void frac-
tion at the inlet of IP turbine. The second one is an analogue circuit. It controls the bypass mass flow into condenser. The
operation mode is as follows:
Case 1: If the IP void fraction at the inlet of the IP turbine is below one, i.e. a mixture of water and steam, the IP turbine
valve will be closed and the IP mixture will not enter the IP turbine section. Instantaneously the second circuit will be set into
operation so that the IP bypass valve opens in order to allow the available mass flow to stream into the condenser.
Case 2: When the IP void fraction at the inlet of the IP turbine becomes one, i.e. pure steam, the IP turbine valve will be
opened and the IP superheated steam will enter the IP turbine again. The IP bypass valve will close and its control will be put
out of operation automatically.

4. Model validation

The HRSG model analyses with design data for steady state and hard transients for dynamical simulation have been per-
formed with high accuracy, so that finally start-up simulations and comparisons with measurement data can be produced.
1180 F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189

The exhaust mass flow and temperature of the gas turbine are applied as dynamic boundary conditions for start-up. Their
transients during start-up process are shown in Fig. 5. The flue gas composition has been assumed as constant over the entire
load range of the gas turbine.

800 250
Flue gas / Steam turbine power
700
200

Steam turbine power [MW]


600
Temperature[ºC]
Mass flow [kg/s]

500 150

400
100
300

200
50
100
FG_Temp Measured FG_Mass flow Measured Steam turbine power
0 0
0 100 200 300 400 500 600 700
Time [MIN]

Fig. 5. Dynamic boundary conditions for start-up.

100
HP_FW mass flow
80
Mass flow [kg/s]

60

40

20
Measured Simulated
0
0 100 200 300 400 500 600 700
Time [MIN]

100
HP_SH Steam mass flow
80
Mass flow [kg/s]

60

40

20
Measured Simulated
0
0 100 200 300 400 500 600 700
Time [MIN]
Fig. 6. HP_FW and HP_SH steam mass flow.
F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189 1181

In the following sections, the simulation results for the dynamic behaviour of the three pressure circuits, namely high
pressure, intermediate pressure/reheater section and low pressure are presented.

4.1. High pressure circuit

Fig. 6 displays that APROS predictions for the dynamic behaviour of high pressure feed water and high pressure super-
heated steam mass flows have a very good agreement towards measurements. The gradients are nearly identical for both

2
HP Drum level
1.8
Drum level [m]

1.6
Δ L=0.45

1.4

Δ L=0.15
?L
1.2

1
Start-up
Start-up Measured
Measured Simulated
Simulated
0.8
-100 0 100 200 300 400 500 600 700
Time [MIN]

Fig. 7. HP drum level.

700
HP_SH Temperature
600
Temperature [ºC]

500

400

300

200
Measured Simulated
100
0 100 200 300 400 500 600 700
Time [MIN]

12
HP_SH Pressure
10
Pressure [Mpa]

2
Measured Simulated
0
0 100 200 300 400 500 600 700

Time [MIN]

Fig. 8. HP_SH pressure and temperature.


1182 F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189

parameters. The oscillations of the measured high pressure feed water mass flow result from a drum level reduction before
start-up displayed in Fig. 7. The difference between the real HP drum level before and after start-up is DL = 0.45 m. During
start-up, the control circuit of the HP_FW mass flow tries to reach back the normal drum level set point and as a result the
HP_FW mass flow valve will open and close suddenly many times causing unstable behaviour of the measured HP_FW mass
flow.
In the HRSG model, this phenomenon does not occur because of an advanced HP drum level control circuit has been used
with tracking feature (feedback signal). So the control circuit response is significantly smoother. Another aspect is that the
difference between the simulated HP drum level before and after start-up amounts to DL = 0.15 m and so it is three times less
than the measured one.
The slight deviation between the measured/simulated HP steam mass flows in Fig. 6 occurs as the simulated superheated
steam mass flow raises gradually in contrast to the measured one that rises very quick after opening the high pressure steam
valve.
In Fig. 7 the peak is very significant right after begin of the start-up in both HP drum level measured and simulated. The
deviation of the absolute peak value results from the lower drum level of the real power plant. This level reduction is forced
by the real plant control in order to avoid drum feeding beyond the upper level limit during start-up.
Fig. 8 shows the high superheated steam pressure and temperature behaviour of measurement and simulation.
The slight difference between measured/simulated HP_SH temperatures between (0 and 50 min) is caused by the slight
difference between the measured/simulated HP steam mass flows. In the real power plant, the superheated steam mass flow
is kept constant equal to zero until 30 min after start-up. But in the HRSG model, the superheated steam flows through the
heat exchanger and absorbs the heat from flue gas. This makes the HP superheated temperature raises earlier than in mea-
surement. It is not possible to keep the steam mass flow equal to zero until 30 min in the HRSG model because it causes
numerical errors in case no steam flows in the HP circuit.

4.2. Intermediate pressure and reheater circuits

The mass flows of intermediate feed water and steam are depicted in Fig. 9. Outstanding to mention are the very hard
transients of the measured curves which are predicted by APROS in the same quality so that there is reached a very high

14
IP_FW mass flow
12
Mass flow [kg/s]

10

2
Measured Simulated
0
0 100 200 300 400 500 600 700
Time [MIN]

12
IP_SH Steam mass flow
10
Mass flow [kg/s]

2
Measured Simulated
0
0 100 200 300 400 500 600 700
Time [MIN]
Fig. 9. IP_FW and IP_SH steam mass flow.
F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189 1183

level of accuracy. The heavy oscillations of measured intermediate feed water mass flow are caused by the high difference
between the real IP drum level before and after start-up DL = 0.35 m as is mentioned before in HP drum.
The intermediate superheated steam temperature and pressure displayed in Fig. 10 show a good agreement between both
measured and simulated data. A slight difference is the initial value of the IP_SH pressure. The reason for that is the initial IP
drum pressure which is at 5 bar in measurement and less than 1 bar in simulation. So it takes more time to regain pressure
level in the HRSG model. It is not possible to keep the pressure constant before the start-up in the HRSG model like the real

350

IP_SH Temperature
300
Temperature [ºC]

250

200
Measured Simulated
150
0 100 200 300 400 500 600 700
Time [MIN]

3
IP_SH Pressure
2.5
Pressure [Mpa]

1.5

0.5
Measured Simulated
0
0 100 200 300 400 500 600 700
Time [MIN]
Fig. 10. IP_SH temperature and pressure.

120
RH Steam mass flow
100
Mass flow [kg/s]

80

60

40

20
Measured Simulated
0
0 100 200 300 400 500 600 700
Time [MIN]

Fig. 11. RH steam mass flow.


1184 F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189

power plant, because no auxiliary steam unit has been modeled for keeping the pressure constant independent of the tem-
perature. So by decreasing the temperature during standing idle state, the pressure drops down less than 1 bar.
The simulated IP_SH temperature rises faster than the measured one because the simulated IP superheated steam mass
flow increases more moderate and it absorbs the full thermal energy as it is mentioned before in HP superheated steam mass
flow.
The reheated steam mass flow displayed in Fig. 11 agrees very well with the measured values. The transients are nearly
identical whereby a slight constant offset between both curves can be identified in the time period between 100 and
700 min.

4.3. Low pressure circuit

Looking at the transient curves of the low pressure feed water and steam mass flows in Fig. 12, there are a very high level
of accuracy between measured and predicted data. The fast start-up is matched very well by APROS and the hard transients
can be simulated as they occur in the real plant.
As mentioned before, the hard oscillations of measured low pressure feed water mass flow result from sensitivity behav-
iour of LP_FW valve which tries to compensate the level differences in the LP drum. For the LP drum this effect is particular
noticeable because not only the LP path but also the HP and IP paths affect the LP_FW mass flow where they take out the feed
water from the LP drum. During start-up the HRSG model does not show the oscillations of the LP_FW mass flow and LP
drum level because the application of an advanced control circuits, however the average trend of both parameters are
identical.
The transient behaviour of the low pressure superheated steam temperature and low pressure drum level depicted in
Fig. 13 matches well with the measurement data.

140
LP_FW mass flow
120

100
Mass flow [kg/s]

80

60

40

20
Measured Simulated
0
0 100 200 300 400 500 600 700
Time [MIN]

14
LP_SH Steam mass flow
12
Mass flow [kg/s]

10

2
Measured Simulated
0
0 100 200 300 400 500 600 700
Time [MIN]

Fig. 12. LP_FW and IP_SH steam mass flow.


F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189 1185

LP_SH Temperature
325

Temperature [ºC]
275

225

175
Measured Simulated
125
0 100 200 300 400 500 600 700
Time [MIN]

3
LP_Drum level
2.75
Drum level [m]

2.5

2.25

1.75
Measured Simulated
1.5
0 100 200 300 400 500 600 700
Time [MIN]
Fig. 13. LP_SH temperature and LP drum level.

5. Parameter study for start-up

The dynamic validation of the HRSG model displayed high reliability towards the measured data. Whereas the com-
parative results of the heat recovery steam generator model and the real plant prove that the simulation is very reliable
to evaluate the real HRSG dynamic phenomenon. Therefore parameter studies like start-up improvement can be
conducted.
As dynamic boundary conditions for the modified start-up, the exhaust mass flow of the gas turbine and the exhaust tem-
perature of the flue gas are varied. In order to improve the start-up procedure, new dynamic boundary conditions are set
with higher gradients, so that the standard start-up duration will be reduced as shown in Figs. 14 and 15. As a result of
start-up improvement the steam turbine reaches the maximal power in 76 min instead 148 min. This means the HRSG saved
approximately 1 h and a quarter in the start-up procedure.
The start-up improvement process showed that under the described transient for inlet boundary conditions, the water/
steam circuits are up to full load within 76 min. The area between the improved and standard start-up lines is the effective
gain of energy. After calculation of this area considering the electricity price in Germany [12], there is a monetary gain of
3500 €/start-up as follows:

(a) The Electricity price is 4 Cent/kWh = 40 Euro/MWh.


(b) The energy gain area is 5185.83 MW min/start-up = 86.4 MWh/start-up.
(c) Monetary gain = Electricity price Eur/MWh * The energy gain MWh/start-up  3500 MWh/start-up.

Usually the average number of power plant start-ups amount to 150 times and more per year. Taking this figure into ac-
count it can be saved more than half million Euro per year.
Now, this result has to be checked from both the acceptable start-up transients for water/steam and gas turbine sides
respectively. Therefore it is important to compare APROS predictions for critically stressed components like HP_SH and
RH systems with their acceptable transients given by design data.
1186 F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189

700
Flue gas temperature
600

500

Temperature [ºC]
400

300

200

100
Measured Improved
0
0 25 50 75 100 125 150
Time [MIN]

700
Flue gas mass flow
600

500
Mass flow [kg/s]

400

300

200

100
Measured Improved
0
0 25 50 75 100 125 150
Time [MIN]
Fig. 14. Boundary conditions for improved/standard start-up.

5.1. High pressure superheater system analysis

The effect of high temperature gradient in flue gas side during improved start-up should be studied in the HP superheater
system. Particularly the gradient of HP superheated temperature in water/steam side has to be taken into consideration. It is
displayed in Fig. 16. The HP_SH outlet temperature in the improved start-up behaves equal to the HP_SH outlet temperature
in case of standard start-up. A minor relevant difference can be realized after 25 min. Particular hard transits of an improved
start-up are possible and the thermal stress r of HP_SH is not exceeded.
The HP_SH pressure in the improved start-up increases faster than the HP_SH pressure in standard start-up which is de-
picted in Fig. 16. For this parameter, the acceptable start-up gradients from design need further investigations with respect
to material ability.

5.2. Reheater system analysis

As another very critical parameter the reheater outlet temperature has been investigated. In the improved start-up the
hard transient and the fast temperature jump in flue gas side may be effect the thermal stress of RH system but as shown
in Fig. 17 the RH temperature characteristic behaves almost the same in improved/standard start-up. This means the dynam-
ical behaviour is kept equal during the improved start-up.
F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189 1187

160
Steam turbine power
140

Energy gain [MW.MIN]


120
ST Power [MW]
100

80
Energy gain
60 Standard
Improved
40

20

0
0 25 50 75 100 125 150
Time [MIN]

Fig. 15. Steam power comparison.

700
HP_SH Temperature
600
Temperature [ºC]

500

400

300

200
Standard Improved
100
0 25 50 75 100 125 150
Time [MIN]

12
HP_SH Pressure
10
Pressure [Mpa]

2
Standard Improved
0
0 25 50 75 100 125 150
Time [MIN]
Fig. 16. HP_SH temperature and pressure characteristics.
1188 F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189

700
RH Temperature
600

Temperature [ºC]
500

400

300
Standard Improved
200
0 25 50 75 100 125 150
Time [MIN]

2.5
RH Pressure

2
Pressure [Mpa]

1.5

0.5
Improved Optimized
0
0 25 50 75 100 125 150
Time [MIN]

Fig. 17. RH temperature and pressure characteristics.

Further important parameter on the RH water/steam side that should be investigated is the RH pressure during improved
start-up. Their transient behaviour is depicted in Fig. 17. Similar to the HP_SH system, the RH pressure increases faster than
in improved start-up and reach the maximal values earlier than standard start-up.
This parameter study for a large scale HRSG showed a high potential in start-up time of water/steam side. In case of a
positive experimental proof that no material breakdown in water/steam side will happen during improved start-up proce-
dure, and in case the transients from the gas turbine are possible, the time period for start-up can be shorted by a factor of 2
as previously explained.

6. Conclusion

The purpose of this study was the dynamic simulation and start-up improvement of an already existing HRSG.
The static and dynamic investigations of a running HRSG power plant with the advanced process simulation software
(APROS) have been successfully completed. The obtained results show a very high level of accuracy and reliability according
to the design data. Hereby the HRSG model represents the real heat recovery steam generator and it can clearly describe the
dynamic phenomenon of the operational process, in order to have reliable start-up information for a flexible and economic
production of electricity. In addition to this the HRSG model represents an efficient tool for future investigation, improve-
ments, further parameter studies and optimization of the operating conditions. Finally it should be emphasized that the
developed HRSG model offers the possibility to simulate the dynamic behaviour of other real heat recovery steam generator
by the same procedure. The strategy of model build up and the developed control circuits can be transferred to other
applications.

References

[1] Pasha A. Combined cycle power plant start-up effects and constraints of the HRSG. ASME paper 92-GT-376; 1992.
[2] Dechamps PJ. Modelling the transient behavior of combined cycle plants. ASME paper 94-GT-238; 1994.
F. Alobaid et al. / Applied Energy 85 (2008) 1173–1189 1189

[3] Jolly S, Gurevich A, Pasha A. Modeling of start-up behavior of combined cycle HRSGs. ASME paper 94-GT-370; 1994.
[4] Kim TS, Lee DK, Ro ST. Dynamic behaviour analysis of a heat recovery steam generator during start-up. Int J Energy Res 2000;24:137–49.
[5] Bausa J, Tsatsaronis G. Dynamic optimization of start-up and load increasing in power plants – Part II. J Eng Gas Turb Power 2001;123:251–4.
[6] Shin JY, Jeon YJ, Maeng DJ, Kim JS, Ro ST. Analysis of the dynamic characteristics of a combined-cycle power plant. Energy 2002;27:1085–98.
[7] Shirakawa M, Nakamoto M, Hosaka S. Dynamic simulation and optimization of start-up processes in combined cycle power plants. JSME Int J, Ser B
2005;48:122–8.
[8] Casella F, Pretolani F. Fast start-up of a combined cycle power plant a simulation study with Modelica. The Modelica Association; 2006.
[9] APROS Advanced Process Simulation Software. http://www.vtt.fi/index.jsp?lang=en.html.
[10] Khartchenko NV. Umweltschonende Energietechnik. Würzburg: Vogel Verlag und Druck GmbH & Co KG; 2000.
[11] Akel A. Fluid mechanics. Damascus: Damascus University; 1981.
[12] Durchschnittspreise für Strom an der EEX in EURO/MWh. von Januar 2002 bis Mai 2007. See also: http://www.udo-leuschner.de/kurzschluss/93-
104.htm.

You might also like