Applied Thermal Engineering 236 (2024) 121852

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Dynamic process simulation of a 780 MW combined cycle power plant

during shutdown procedure
Falah Alobaid a, *, Jakob Wieck b, Bernd Epple a
a
Technical University of Darmstadt, Institute for Energy Systems and Technology, Otto-Berndt-Straße 2, 64287 Darmstadt, Germany
b
General Electric Switzerland GmbH, Switzerland

A R T I C L E I N F O A B S T R A C T

Keywords: The share of variable renewables in power generation is increasing due to the push to reduce emissions, and the
Combined-cycle power plant flexibility of conventional power plants has become a major concern. There are many ways to increase the
Dynamic process simulation flexibility of existing power plants and implement a more flexible system for the future, such as a fast start-up
Flexible operation
procedure and higher load change rates. In this study, a 2D dynamic process simulation model of a reference
Shutdown procedure
combined cycle power plant is presented to increase the flexibility of cyclic operation by investigating, for the
Validation study
first time in the literature, the shutdown procedure using APROS software. The 780 MWel Tambak Lorok
reference combined cycle power plant consists of the GE 9HA.02 heavy-duty gas turbine and a three-stage
horizontal heat recovery steam generator with reheat. Several steps were considered to improve the accuracy
of the developed dynamic process simulation model, including varying the steam leakage to the atmosphere to
control how much thermal energy remains in the heat recovery steam generator, changing the shape loss co­
efficients of the tubes to achieve higher or lower recirculation, and decoupling the reheater and the intermediate
pressure drum. The developed model is validated with real data from three different combined cycle power
plants in different countries with various capacities. The comparison shows that the developed dynamic simu­
lation model can accurately represent the real behaviour of natural convection in the heat recovery steam
generator. The main outcome was to provide detailed information on the shutdown procedure of large-scale
combined cycle power plants, addressing the aspect of the development of new start-up strategies to reduce
the time required for the start-up procedure.

considered, such as the expansion of high-voltage transmission infra­

structure, the use of large-scale energy storage systems, and the
1. Introduction
deployment of highly flexible power generation units, such as combined
cycle power plants (CCPPs) [7,8]. The nominal process efficiency of a
Power generation represents the largest sector of primary energy
modern combined cycle power plant can reach up to 64% (for example
consumption in 2020, accounting for 40% of the total, and is responsible
Irsching, Germany, Bouchain, France, and Tambak Lorok, Indonesia
for 37.5% of total annual CO2 emissions [1]. The policy response to CO2
[9,10]). Combined cycle power plants can withstand demanding load
emissions was a United Nations agreement in 1997, the Kyoto Protocol,
changes of up to ± 60 %/min, which is suitable for primary frequency
which was extended in 2015 by the Paris Agreement [2]. In the Paris
response [11,12]. The operating minimum load for typical gas turbines
Agreement, a long-term goal was set by the signatories to limit the in­
can be reduced to 40% of the nominal load. This value can be further
crease in global temperature to no more than 2 ◦ C above pre-industrial
reduced to 20% if sequential combustion is used. Furthermore, com­
levels but aimed to keep it to 1.5 ◦ C above pre-industrial levels [3,4]. To
bined cycle power plants can be operated with a low minimum load,
achieve these goals, CO2-neutral power generations such as wind tur­
which is important for flexible operation as this is the lower limit for
bines and solar panels become particularly important. The growing
negative load changes. That also makes them a cost-effective choice, as
share of renewables, some of which are weather-dependent, is leading to
the number of start-up and shut-down procedures can be reduced. The
a new discussion about the flexibility of power generation systems and
literature generally defines three types of shutdown and start-up pro­
the power grid in particular [5,6]. To maintain the security of supply and
cedures for a CCPP, namely hot, warm, and cold start-ups, respectively
improve the flexibility of the future power grid, various solutions can be

* Corresponding author.
E-mail address: falah.alobaid@est.tu-darmstadt.de (F. Alobaid).

https://doi.org/10.1016/j.applthermaleng.2023.121852
Received 30 June 2023; Received in revised form 5 October 2023; Accepted 23 October 2023
Available online 25 October 2023
1359-4311/© 2023 Elsevier Ltd. All rights reserved.
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

Nomenclature wal wall

F force/volume [N/m3] Abbreviations

g standard gravity [9.81 m/s2] Apros process simulation software
h static enthalpy [kJ/kg] ASPEN process simulation software
h0 stagnation enthalpy [kJ/kg] Attemp attemperator
ṁ mass flow rate [kg/s] BFP boiler feedwater pump
p static pressure [Pa] CC combustion chamber
q heat flow/volume [kW/m3] CCPP combined cycle power plants
T temperature [◦ C] Comp compressor
t time [s] ECO economiser
u longitudinal fluid velocity [m/s] EVAP evaporator
ρ density [kg/m3] FG flue gas
Γ mass transfer [kg/(m3 s)] FW feedwater
v specific volume [m3/kg] GE general electric
χ void fraction [-] GT gas turbine
z axial position [m] HP high pressure
HRSG heat recovery steam generator
Subscripts IP intermediate pressure
g gas phase LP low pressure
i component index or interface between phases MATLAB/SIMULINK mathematical software
form form loss RH reheater
k liquid or gas phase SH superheater
l liquid phase ST steam turbine
pu pump WS water/steam
val valve

[13,14]. The selection of the appropriate start-up type depends on the Based on this validated model, the same authors [25] introduced a dy­
temperatures of the steam turbine and the casing metals, which correlate namic simulation model of a once-through HRSG to study the load
with the duration since the shutdown. Using only these three start-up cycling and start-up processes for next-generation high-efficiency com­
modes (hot, warm, and cold start-ups) may result in a more costly and bined cycle power plants. A similar approach was followed by Watanabe
slower start-up because, for example, the temperatures of the compo­ et al. [21] by analysing the dynamic characteristics of a once-through
nents of the combined cycle power plant may be between a warm and a HRSG using the Matlab/Simulink environment. Mertens et al. [26]
hot start-up (often in such a case a warm start is selected, which is investigated the effects of the design of a heat recovery steam generator
probably not the best choice). Therefore, it is important to determine the (once-through steam generator vs. drum steam generator) on the dy­
behaviour of the CCPP during the shutdown procedure for a possible namic behaviour under the boundary condition of the start-up of a gas
faster start-up. turbine. Both models are configured to provide equal steam mass flows
Mathematical simulation models for combined cycle power plants with equal steam parameters (temperature, pressure) at full load,
can contribute to a better understanding of the process and play an allowing a fair balance evaluation of the two technologies. The results
important role in increasing the flexibility and efficiency of the plant show generally similar responses of the high-pressure systems to gas
[15]. Different process modelling concepts can be applied, namely turbine startup, with the exception of an accelerated pressure buildup in
black-box process modelling and white-box process modelling [16]. In the once-through evaporator. Liu and Karimi [27] used the software
black-box process modelling, also known as data-driven models, the ASPEN HYSYS to simulate the transient behaviour of a three-pressure
process parameters are determined by combining training data (exper­ HRSG with reheat during an off-design operation. For model valida­
imental data) and network structure using various regression techniques tion, the numerical results were compared with those of the GateCycle
and artificial intelligence [17–19]. In contrast, the process parameters in analogue model, showing good correspondence. McConnell et al. [28]
white-box models (e.g., dynamic process simulation) are derived from presented a multi-stage HRSG model built using the transient energy
the mass, energy and momentum conservation equations and therefore flow equations program. Model predictions at steady state and during
agree better with measurements. Recently, the use of dynamic process start-up were validated against data from a real plant and showed good
simulation is rapidly increasing to meet the demands of more dynamic agreement. Sabia et al. [29] studied the capability of ASPEN dynamics to
power generation [7,20,21]. In the literature, several efforts have been simulate a hot start-up procedure of a 350 MWel CCPP. The authors
made to study the transient behaviour of CCPPs to improve operational conclude that these simulations can be performed with an acceptable
flexibility during load changes and start-up procedures. For example, relative error, but face the challenge of accounting for thermal inertia,
Casella and Pretolani [22] built up a dynamic process simulation model which is not easy to implement, and believe that ASPEN is better suited
using the software Modelica to assess the fast start-up of a CCPP. The for simulating chemical processes than for simulating steam cycles. A
objective of the study was to shorten the start-up time while regulating comparative study between ASPEN dynamics and APROS as well as
the lifetime use of essential components. Since the study focused on the ASPEN dynamics, APROS and Power Plant Simulator & Designer con­
analysis of thermal stresses in the steam turbine, the low-pressure part of cerning the simulation of a HRSG during the warm start-up was per­
the heat recovery steam generator (HRSG) has been simplified and the formed by Alobaid et al. [30] and Bany Ata et al. [31], who agreed with
behaviour of the heat exchanger has not been examined. Alobaid et al. the view of Sabia et al. [29] regarding the suitability of ASPEN dynamics
[23,24] developed a dynamic simulation model for a drum-type HRSG to simulate the water/steam cycle of the heat recovery steam generator.
using APROS. The model shows satisfactory agreement with real mea­ Recent work on the dynamic simulation of a combined cycle power plant
surements for different loads and during the warm start-up procedure. during load changes and start-up procedures can be found, e.g., in.

2
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

[21,32–34], summarized in a review paper on the steam cycle power addition, the best configuration of the model and the evaluation are
plant [35]. shown. Finally, the main results of this study are highlighted in the
From the above literature review, many studies were found dealing conclusion section.
with the dynamic simulation of load changes and start-up procedures of
CCPPs. Regarding the shutdown procedures, there is only one study by 2. Reference power Plant
Mertens et al. [36], which numerically evaluates the start-up and shut­
down procedures of a 350 MWel CCPP with a three-pressure HRSG. The Tambak Lorok combined cycle power plant is powered by GE’s
Given the limited existing work (only one publication), this study con­ largest and most efficient gas turbine, the GE 9HA.02 gas turbine (GT),
tributes to filling the research gap of dynamic simulation of the com­ along with GE’s D650 steam turbine (ST), a heat recovery steam
bined cycle power plant during shutdown procedures. For the first time generator and plant equipment. The power plant, which started oper­
in the literature, a detailed 2D dynamic process simulation model of the ating in 2020, supplies around 780 MW of electricity to the Indonesian
780 MWel Tambak Lorok combined cycle power plant is introduced to grid, equivalent to the power needs of up to five million Indonesian
enhance the flexibility of cyclic operation. The developed dynamic households. The technology used in this power plant is one of the most
simulation model is generated using APROS software and compared efficient, operating at more than 64% with the lowest emissions in the
with real data from three different power plants with different power industry, fuel flexibility, and fast load changes of 8 to 12% per minute
outputs. The novelty and the objectives of this study can be briefly [37]. The HRSG is horizontally orientated, triple pressure (low-pressure
described as follows: (LP) and intermediate-pressure (IP) drum-type boilers and high-pressure
(HP) once-through-type boiler without supplementary firing) with
1) Investigating the shutdown procedures of the 780 MWel Tambak reheater (RH), in detail:
Lorok combined cycle power plant to address the aspect of switching
to a new start-up strategy, aiming at reducing the time required for - Once-through evaporator path in the high-pressure circuit.
start-up. - Natural circulation evaporator path in the intermediate and low-
2) Investigating the effects of natural convection during the shutdown pressure circuits.
procedures by applying a 2D dynamic process simulation model - Reheater section after the high-pressure turbine.
instead of the 1D dynamic modelling on model accuracy. - Three steam turbine stages (HP, IP, and LP).
3) Creation of a new research forum to discuss the development of new
strategies for the control and operation of CCPPs during the modified This type of HRSG is often referred to as a Benson once-through
start-up and shutdown procedures. HRSG and allows for faster dynamic response of start-ups and load
changes. The flue gas from GT exhaust has a temperature equal to 638 ℃
After the introduction in Section 1 in the previous lines, the and a mass flow rate of 1119 kg/s. Fig. 1 shows a schematic diagram of
remaining sections of this paper are organised as follows: Section 2 ex­ the CCPP. As well as the technical data of flue gas and water/steam side
plains the configuration of the reference power plant. The software used is presented in Table 1.
and its scope is described in Section 3. In Section 4, a description of the
process flow model is presented. Results assumptions, and approaches
for model tuning used are summarised and discussed in section 5. In

Fig. 1. Schematic flow diagram of the reference combined cycle power plant

3
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

Table 1
Technical data of the reference combined cycle power plant
HRSG outlet steam conditions Gas turbine conditions

High pressure Reheater/Intermediate pressure Low pressure Condenser Steam turbine Flue gas Power

p = 162 bar p = 30.3 bar p = 5.3 bar p = 0.05 bar Power = 230 MW TInlet = 638 ℃ 550 MW
T = 600 ℃ T = 605 ℃ T = 315 ℃ T = 32 ℃ Isentropic Efficiency: TOutlet = 70 ℃
m˙=113.8 kg/s m˙=134.4 kg/s m˙=151.9 kg/s m˙=151.9 kg/s High-pressure steam turbine = 0.86 m˙=1119 kg/s
Intermediate pressure steam turbine = 0.9
Low-pressure steam turbine = 0.9

3. Simulation software Herr, Fgra is the gravitational acceleration force per volume, Fwal and qwal
are the wall friction force per volume and the heat flow through walls
In this present study, the APROS process simulation software per volume. The terms u and ρ refer to the longitudinal velocity and
developed by Fortum and the Technical Research Centre of Finland density of the fluid, respectively. The function f describes the pressure
(VTT) [38,39] is applied to model the 780 MWel Tambak Lorok com­ losses due to valve and form friction as well as the hydrostatic pressure
bined cycle power plant. APROS software, which represents Advanced differences and the pressure force of a pump. The total enthalpy h0 is the
PROcess Simulation, includes dedicated component libraries for the static enthalpy including the kinetic energy of the flow. The source term
process, automation, and electrical systems for dynamic modelling of in the mass equation may contain additional mass flows into the system.
thermal power plants, energy, and industrial processes. The modelling is The pressure derivative term appears in the energy equation since the
done by selecting the required process components (e.g., compressor, total enthalpy is used instead of the internal energy U. In the case of two-
gas turbine, heat exchanger) and linking them via material, energy, or phase flow components, the three variables (local pressure, total mass
heat flows to build a model of an existing process and/or a new system flow, and fluid enthalpy) are supplemented by the void fraction, which
for research. All necessary configuration data (e.g., compressor or gas can be calculated by adding a fourth additive constitutive equation to
turbine characteristics as well as the geometry of the heat exchanger) the previous partial differential equations.
can be inserted into the process components, which are connected via The two-fluid flow model, which is also known as the heterogeneous
electrical and automation signals for control purposes. There are many or six-equation flow model, does allow for chemical and thermal
studies in the literature on the use of APROS models to simulate the imbalance (velocity and temperature imbalance between the phases)
dynamic behaviour of thermal power plants during load changes or but assumes mechanical equilibrium (the phases are always under the
start-up procedures. Most of this work included validation analyses of same pressure). It more accurately describes two-phase flow than the
the numerical results obtained and confirmed the accuracy and read­ mixture flow model, but defining the interaction terms between the
ability of the models developed, e.g., for combined cycle power plants phases is challenging. This is because the interaction terms cannot be
[23,40,41], nuclear power plants [42,43], pulverized coal power plants determined from physical laws but are generally obtained from experi­
with CO2 capture [44–46], municipal waste incinerators [47], circu­ ments under various artificial conditions. As a result, the two-fluid flow
lating fluidized bed combustor [48] and concentrated solar power plants model is more vulnerable to numerical instability than the mixture flow
[49]. It should be mentioned that the physics behind dynamic modelling model. The two-fluid flow model expresses separate conservation
with the APROS software (equations for the cycle components such as equations for mass, momentum, and energy for each phase of the water-
steam turbine, heat exchangers and feed water pumps) can be found in a vapour mixture. The mass conservation equation for one phase reads as
detailed review article [7]. In the following, only the essentials are follows:
explained.
The unsteady simulation is based on the transient solution of ∂(χ k ρk ) ∂(χ k ρk uk )
+ = S + Γik (4)
nonlinear differential equation systems together with correlations for ∂t ∂z
friction and heat transfer, which describe the real behaviour of the
thermal power plant over the entire operating range. The software in­ The momentum conservation for phase k is expressed as:
corporates two thermal–hydraulic modules, i.e., a mixing flow model ∂(χ k ρk uk ) ∂(χ k ρk u2k ) ∂p
and a two-fluid flow model. The one-dimensional mixing flow model ∂t
+
∂z
= − χ k + Γik uik + Fgra,k + Fwal,k + Fik + fk (val + form
∂z
(also known as homogeneous or three-equation flow model) presumes + pu)
thermodynamic equilibrium between the phases (water and steam). It is
(5)
represented by three partial differential equations for mass, momentum,
and energy that describe the dynamic behaviour of the characteristic The energy conservation is formulated as:
quantities. For single-phase flow components (e.g., superheaters,
reheaters, and economizers), the three characteristic fluid variables are ∂(χ k ρk h0,k ) ∂(χ k ρk uk h0,k )
+
∂p
= χ k + Γik h0,ik + qik + qwal,k + Fik uik (6)
local pressure, total mass flow, and fluid temperature or fluid enthalpy ∂t ∂z ∂t
for supercooled water or superheated steam. The conservation of mass is
The subscript k is either l = liquid or g = gas. The subscript ik refers to
stated as follows:
the interface between two phases, while the subscript wal, k denotes the
∂ρ ∂(ρu) interface between one phase and the wall. The term Γik represents the
+ =S (1)
∂t ∂z mass exchange rate from phase i to phase k. The function fk shows the
effects of valves, pumps, and friction on the flow. The terms F and q are
The momentum conservation is written as: the average friction force per volume and heat flow per volume,
∂(ρu) ∂(ρu2 ) ∂p respectively. The symbol h0 in the energy equation is the total enthalpy
+ = − + Fgra + Fwal + f (val + form + pu) (2) including kinetic energy. The wall friction Fwal,k , interfacial friction Fik ,
∂t ∂z ∂z
interfacial heat flow qik and wall heat flow qwal,k can be modelled using
The energy conservation is expressed as: empirical correlations, based on the flow regime prevailing in the flow
[50]. This flow model is suitable for water/steam mixtures with domi­
∂(ρh0 ) ∂(ρuh0 ) ∂p
+ = + qwal (3) nating mass and heat transfer between the phases, such as in evaporators
∂t ∂z ∂t

4
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

and condensers. the final superheater to control the temperature at the inlet of the high-
To solve the one-dimensional partial differential equations, the finite pressure turbine. Both attemperators use water from the high-pressure
volume method is applied. The partial differential equations are dis­ feedwater pumps. The feedwater mass flow is controlled by a feed­
cretized spatially and temporally, and all nonlinear terms are made water control valve whose control logic is based on the real power plant
linear. In the spatial discretization, the staggered discretization scheme control structure.
is adopted. The state variables (such as pressure, enthalpy, and density) The most important part of a simulation of the shutdown process of a
are computed at the centre of the mesh cells and the flow-related vari­ CCGT power plant is the HRSG since this is where the recirculation of the
ables (such as velocity) are computed at the boundaries between two hot gases takes place when the GT exhaust gases no longer play the main
mesh cells. For the enthalpy solution, the first-order updraft method is role. The shutdown procedure starts when the GT’s load reaches 0% and
applied. In the mesh cell, the variables are averaged over the entire its speed starts to decrease. Immediately after the shutdown, the bypass
mesh. For time discretization, the implicit method is used, where the valves of the HP and IP circuits are closed to keep the steam in the plant.
linear equation groups for pressure, void fraction, and enthalpy are At this point, the GT exhaust flow is low and temperature stratification
solved sequentially. The density is then updated as a function of the begins throughout the flue gas path. Fig. 2 illustrates the natural con­
solved pressure and enthalpy. vection that happens inside the flue gas path of a horizontal HRSG
during a shutdown procedure. Due to convection, the hot air at the top
4. Model description moves towards the cold end, where the evaporators are located. This hot
gas flows onto the evaporators, causing them to produce steam again
The APROS process simulation software is used to model in detail the and build up a pressure that can be observed at the start of a shutdown
780 MWel Tambak Lorok combined cycle power plant, including the gas based on data from real power plants. Since 1D modelling is not suffi­
turbine, the flue gas path, the three pressure circuits of the subcritical cient to account for this effect, a 2D model of the HRSG was developed to
once-through heat recovery steam generator, and the steam turbine. The simulate the behaviour of the natural convection that occurs during the
flue gas path is modelled from the gas turbine inlet to the heat recovery shutdown procedure in the flue gas path of the heat recovery steam
steam generator exit. On the water/steam side, all bundle heat ex­ generator (see Fig. 3).
changers have been implemented with real geometry data. These To simplify the model, the gas turbine is not considered; the values of
include high-pressure superheaters, high-pressure separators, high- the mass flow rate of flue gas and the exhaust temperature are given by
pressure once-through evaporators, high-pressure economisers, reheat­ curves provided by GE, not by a dynamic mathematical model; the low-
ers, intermediate-pressure superheaters, intermediate-pressure drum, pressure superheater is not considered, and the feed water is given as a
intermediate-pressure evaporators, intermediate-pressure economisers, boundary condition, so the cycle is not closed (the steam coming out of
low-pressure superheaters, low-pressure drum, low-pressure evapora­ the steam turbine does not recirculate as condensate water at the
tors and low-pressure economisers. Furthermore, all connecting piping, beginning of the process). These simplifications have only a minor effect
headers, pumps, steam and gas turbines, valves, stream splitters and on the shutdown results but can increase the simulation speed.
mixers as well as different controllers were implemented in the devel­
oped model on a 1:1 basis as in the real power plant. As an example, only 5. Results
the high-pressure system is described since the other systems are almost
identical to conventional drum HRSGs. The feedwater mass flow, which 5.1. Modelling approaches
flows into the high-pressure circuit, is supplied by the high-pressure
boiler feed pump. The feed water flows into the HP evaporators via To improve the model results, the following approaches were
the high-pressure economisers. In the high-pressure once-through followed:
evaporators, the water is heated by the flue gas and converted into
steam. The steam then leaves the high-pressure evaporators and flows 1) Manipulation of steam leakages to the atmosphere.
into the superheater system, where the steam absorbs additional heat 2) Changing the form loss coefficient of the pipes to obtain faster or
from the flue gas. The superheated steam leaves the heat recovery steam slower convection:
generator and enters the high-pressure turbine section. Interstage and • Changing the first column.
final cool injectors (attemperator) are provided at the inlet and outlet of

Stack

HRSG
FW/LP ECON
HP ECON2

HP ECON1
HP EVAP

IP ECON

LP EVAP
IP EVAP
HP SH

IP SH
RH

G GT

Fig. 2. Natural convection inside a horizontal HRSG during the shutdown procedure

5
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

Fig. 3. Schematic flow diagram of the 2D dynamic process simulation model of the flue gas path of the heat recovery steam generator

• Modifying all other columns with the best configuration found for
the first column, as the mass flow rate in this column is higher and
therefore the results are more sensitive to it.
3) The best configuration on the previous items is selected and decou­
pling of RH and IP drum is performed.

All the obtained results from this model are validated by real data
from three large-scale combined cycle power plants from three different
countries (see Table 2). Unfortunately, there is currently no data avail­
able from the 780 MW Tambak Lorok combined cycle power plant
during the shutdown process. For confidentiality reasons, all results
have been normalized so that the overall curve is apparent, but the
values are given as percentages (from 0 to 100%).

5.1.1. Manipulation of steam leakages

As discussed in section 4, when a shutdown is performed in a CCPP,
the bypass valves to the steam turbine are closed to keep the steam in­ Fig. 4. HP pressure decay (Plant 1, Plant 2, Plant 3, and simulation) for
side the HRSG, allowing a start-up to be faster and cost-efficiently per­ different steam leakages
formed. However, since it is impossible to retain all the steam in the real
plant, steam leakage to the atmosphere is included in the model to atmosphere (0.00 kg/s of steam leakage), the pressure builds up higher
ensure that this issue is accounted for in the simulation of a shutdown and takes a longer period to decay, which is not observed in the real data
procedure. The greater the steam leakage the smaller the building up of from the power plants. The opposite can also be stated, when the losses
pressure and the faster the temperatures are decreased. The steam to the atmosphere are too high, the system loses a lot of thermal energy
leakages of the superheater and the reheater can be changed indepen­ and the building up of pressure is not as high as desired but the decay
dently, but since this operation serves to compensate for the unavoid­ over time is close to the reality. With these two opposites in mind, a
able pressure losses to the environment, they have been left at the same value in the middle (0.04 kg/s of steam leakage) was considered optimal
value to better reflect reality and keep the approach simpler. to move forward on tuning the model. Fig. 5 illustrates that the decay of
Fig. 4 shows the pressure decay in the high-pressure circuit, while the temperature inside the superheater is only mildly affected by the
temperature decay is shown in Fig. 5. The value of the steam leakage is amount of steam released into the atmosphere. The different pattern
closely related to the amount of thermal energy that is kept inside the observed at the end of the simulation, after 48 h, is related to the fact
HRSG, so Fig. 4 reveals that when no steam is released into the that the amount of steam inside the superheater is considerably smaller

Table 2
Technical information of the power plants used as validation of the simulated results
Power Location Number of HRSGs HRSG Type Capacity SH Tem- HP Drum IP Drum LP Drum
Plant perature

Plant 1 France 1 horizontal triple pressure 605 MW 584 ◦ C 157 bar 29.5 bar 5 bar
+ RH
Plant 2 United Kingdom 2 horizontal triple pressure 880 MW 543 ◦ C 151 bar 32 bar 8.4 bar
+ RH
Plant 3 Brazil 3 triple pressure 1,551 MW – – – –

6
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

Table 3
Form loss coefficients used on the first vertical column
Loss coefficient SH and RH Form loss coefficient
1st column leakage other columns

Form loss 1 0.04 kg/s 5

coefficient 1
Form loss 3 0.04 kg/s 5
coefficient 3
Form loss 5 0.04 kg/s 5
coefficient 5
Form loss 10 0.04 kg/s 5
coefficient 10

Fig. 5. HP temperature decay (Plant 1, Plant 2, Plant 3, and simulation) for

different steam leakages

so the simulation starts to oscillate leaving the dots plotted not so linear.
The IP drum pressure decay, shown in Fig. 6, is like the HP pressure
decay. The lower the steam leakage, the higher the building up of
pressure and the longer its decay. The sawtooth pattern is obtained as a
safety valve is programmed to open as the IP drum reaches a specific
pressure and then closes again when the pressure lowers. Nevertheless,
the chosen value (0.04 kg/s of steam leakage) is in concordance with the
real data.

5.1.2. Manipulation of form loss coefficient on 1st column

In the next step of the approach to improve the model accuracy, Fig. 7. HP pressure decay (steam leakage 0.04 kg/s; Plant 1, Plant 2, Plant 3,
several simulations were performed with the previously selected value and simulation) for different loss coefficients on 1st column
for the steam leakage in the atmosphere (0.04 kg/s for superheater and
reheater) by changing the form loss coefficient of the tubes that allow
the recirculation inside the HRSG and, consequently, the heat transfer by
convection. The mass flow rate inside these vertical tubes (see Fig. 3) can
vary depending on their position from the exit of the GT (the closer the
bigger the mass flow rate). Therefore, in this step, only the form loss
coefficient of the first column of vertical tubes was changed, since the
largest mass flow rate flows in them, while all other columns were kept
unchanged (see Table 3). The larger the form loss coefficient in the
tubes, the more difficult it is for a fluid to flow in that tube, so this is
closely related to the recirculation and energy distribution in the HRSG.
Fig. 7 and Fig. 8 show the temperature and temperature decay in the
high-pressure circuit, respectively. Here, the steam leakage is 0.04 kg/s
for the superheater and reheater. The 1st column form loss coefficients
were varied, and the obtained results were compared with the real
power plants. It can be concluded that the form loss coefficient has a
great influence on the recirculation in the HRSG and thus on the shut­ Fig. 8. HP SH temperature decay (steam leakage 0.04 kg/s; Plant 1, Plant 2,
down procedure itself. Fig. 7 describes the link between the pressure Plant 3, and simulation) for different loss coefficients on 1st column
build-up and the recirculation, in this case in terms of the form loss
coefficient. As the values of the form loss coefficient increase, the con­
vection within the HRSG decreases because it is more difficult for the
fluid to flow within these tubes, so the thermal energy is conserved
longer, but the pressure build-up is not as intense and is also delayed.
The coefficient of form loss in the first column closest to the output of the
gas turbine is the one that better reflects the real data in terms of the
timing of the pressure build-up, its magnitude, and its decay. In contrast
to the steam leakage, the form loss coefficient has a strong influence on
the temperature decay at the outlet of the superheater, following the
same concept explained for the pressure decay at the high-pressure
circuit (see Fig. 8). The larger the form loss coefficient, the slower the
recirculation and the temperature decay.
Fig. 9 shows the pressure decay at the IP drum compared to real data
from two power plants. The safety valve placed at the IP drum does not
allow the changing in the form loss coefficients to alter the magnitude of
the building up of pressure, but these variations change the period of the
Fig. 6. IP drum pressure decay - Plant 2, Plant 3, and simulation) for different
steam leakages decay. The bigger the form loss coefficient, consequently the lower the

7
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

Fig. 9. IP drum pressure decay (steam leakage 0.04 kg/s; Plant 1, Plant 2, Plant
3, and simulation) for different loss coefficients on 1st column Fig. 10. HP pressure decay (steam leakage at 0.04 kg/s; form loss coefficient
first column at 1; Plant 1, Plant 2, Plant 3 and simulation) for different loss
coefficients on the other columns
recirculation, and the more delayed the building up of pressure and its
decay. Also, for this case, the best configuration is with form loss coef­
ficient equals 1. same time as the steam turbine bypass valves is connected between the
IP drum and the RH section so that they are independent of each other.
5.1.3. Manipulation of form loss coefficient of other columns Then, a mass flow rate is introduced only in the RH as a boundary
As the model is further improved and now used in combination with condition to keep the mass inflow as it occurs in a real plant. Different
the two configurations previously obtained, a comparison is made when values for the steam leakage going from the RH to the atmosphere and
the form loss coefficient of other columns is also changed. Since the mass the mass flow rate entering through the boundary condition were tested
flow rate in these tubes is very small, about 100 times smaller, these (see Table 5) (see Figs. 12–13).
changes should not affect the results very much. However, the study was
carried out to prove that the influence is low, which confirms the above
5.2. Evaluation
hypotheses. The values used in this section and the constant conditions
from the previous procedures can be seen in Table 4.
The final configuration can be seen in Table 6, which explains how
In this case, only one figure is shown because the purpose of this
the model was tuned to obtain better results. Considering the satisfac­
section is to prove the hypothesis that the change in the form loss co­
tory results obtained with the decoupling approach, it was kept in the
efficients of the other tube columns does not affect the results. The HP
final configuration. All the figures of the best results compared to the
pressure decay was chosen to be present and is shown in Fig. 10. As can
real data from the power plants are shown in the following figures.
be seen, the hypothesis could be confirmed. Changing the form loss
Fig. 14 shows that the model can reproduce well the pressure decay
coefficient of the other columns does not affect the result, since the air
observed in the real power plants, and the magnitude of the values also
mass flow rate in these tubes is almost negligible compared to the mass
matches, which is to be expected since the power plants have similar
flow rate in the tubes of the first column. Therefore, the value for the
specifications. The temperature decay, illustrated in Fig. 15, is also
form loss coefficient for the other columns is left at 5, as in the originally
correctly reproduced. However, a small deviation can be seen in the
presented model, to try to change the model as little as possible.
temperature decrease of power plant 1 and power plant 2. The tem­
perature decrease of power plant 2 is much slower, but this could be
5.1.4. Decoupling RH and IP drum
related to the position of the sensors that measured the obtained data.
Fig. 11 shows the comparison between the pressure decay in the IP
However, this information was not shared, so it cannot be assumed with
drum and the pressure decay in the RH using the best configuration
certainty that this is the reason for the different behaviour, but it should
found in the previous sections (0.04 kg/s steam leakage, form loss co­
be considered.
efficient first column 1, and form loss coefficient other columns 5). In the
Fig. 16 shows the sawtooth pattern in the IP drum pressure decay is
model used in this work, the IP drum and the RH are directly connected
well reproduced by the model and the results are in concordance with
via pipes so the obtained results on a shutdown procedure for these two
the data provided by the power plant plants.
sections are very similar, but the real data showed a slightly different
The decoupling procedure showed good results in terms of freeing
behaviour between them. To improve the model prediction, the
the RH from the IP drum, so the simulation gives a better curve that is
decoupling of the RH and IP drum sections as in the real plant should be
more in line with reality. However, as can be seen in Fig. 17, there is still
carried out. To achieve such decoupling, a valve that is closed at the
room for improvement in the RH pressure decay, which should be done
in future work. Fig. 18 shows the temperatures in the economisers and
Table 4 was only shown to demonstrate that the model can reproduce the results
Form loss coefficients used on the other columns with a good degree of confidence. This data was not used in tuning the
Form loss SH and RH Form loss model, but it is good to add to the discussion and feel that the right way
coefficient other leakage coefficient 1st is being taken during the model tunning. There is some degree of
columns column discrepancy between the simulation and the measurements, but that is
RH form loss 3 0.04 kg/s 1 not critical because these measurements were performed outside the flue
coefficient 3 gas path.
RH form loss 5 0.04 kg/s 1
coefficient 5
RH form loss 6 0.04 kg/s 1 6. Conclusion
coefficient 6
RH form loss 6.5 0.04 kg/s 1 In this work, a modern combined cycle power plant with an effi­
coefficient 6.5
ciency of up to 64% and a capacity of 780 MWel was investigated using a

8
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

Fig. 11. The pressure decay of IP drum and RH without decoupling

Table 5
Values simulated for the RH and IP drum decoupling
RH Mass Form loss SH Form loss
steam flow coefficient steam coefficient
leakage rate in other leakage 1st column
RH columns

decoupling 0.02 kg/ 0.00 5 0.04 kg/ 1

0.00 in − s kg/s s
0.02 kg/s
out
decoupling 0.02 kg/ 0.025 5 0.04 kg/ 1
0.025 in s kg/s s
− 0.02
kg/s out
decoupling 0.02 kg/ 0.03 5 0.04 kg/ 1
0.03 in − s kg/s s
0.02 kg/s
Fig. 13. HP pressure decay (leakage 0.04 kg/s; form loss coefficient first col­
out
umn 1, form loss coefficient other columns 5; Plant 1, Plant 2, Plant 3 and
decoupling 0.02 kg/ 0.04 5 0.04 kg/ 1
0.04 in − s kg/s s simulation) for comparing different decoupling IP drum and RH configurations
0.02 kg/s
out dynamic simulation model during the shutdown procedure. The refer­
ence combined cycle power plant “Tambak Lorok” was built in
Indonesia and consists of the GE 9HA.02 high-performance gas turbine
and a three-stage horizontal heat recovery steam generator with a
reheat. Several steps were considered to improve the model prediction,
including varying the steam leakage to the atmosphere to control how
much thermal energy remains in the heat recovery steam generator and
changing the form loss coefficients of the tubes to provide higher or
lower recirculation. Since the real data showed a different behaviour
than the model, a final approach to decouple the reheater and the IP
drum was proposed and gave promising results. The following key
points emerge from this study:

1. A sophisticated 2D dynamic process simulation model was devel­

oped, tuned, and applied to study the shutdown procedure of the
780 MW Tambak Lorok combined cycle power plant. The model
results were compared with real data from three different combined
Fig. 12. RH pressure decay (leakage 0.04 kg/s; form loss coefficient first col­
cycle power plants and showed good agreement. Quantitatively, the
umn 1, form loss coefficient other columns 5; Plant 2 and simulation) for developed model can reproduce well the pressure drop observed in
different decoupling IP drum and RH configurations real power plants, and the magnitude of the values also agrees with a

9
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

Table 6
Conditions for the best configuration
RH steam leakage Mass flow rate in RH Form loss coefficient other columns SH steam leakage Form loss coefficient 1st column

Best configuration 0.04 kg/s 0.05 kg/s 5 0.04 kg/s 1

Fig. 14. HP pressure decay best configuration (Plant 1, Plant 2, Plant 3, Fig. 17. RH pressure decay best configuration (Plant 2 and simulation)
and simulation)

Fig. 18. RH pressure decay best configuration (Plant 2 and simulation)

Fig. 15. HP SH temperature decay best configuration (Plant 1, Plant 2, Plant 3,
and simulation) 2. The developed 2D dynamic process simulation model of the 780 MW
Tambak Lorok combined cycle power plant proved to be very
effective in reproducing the behaviour of natural convection in the
heat recovery steam generator during the shutdown procedure.

Finally, the authors strongly recommend the use of the 2D dynamics

modelling approach instead of 1D dynamics for process simulation of
other power plants (e.g., coal-fired power plants), even if start-up or load
changes are to be considered. This is because the 2D dynamics modelling
approach provides a reliable tool to obtain more accurate and higher-
quality results, at the same time, the computational time does not in­
crease to the same extent. The only disadvantage is the high time
required for model building with this approach compared to 1D dynamic
modelling.

Declaration of Competing Interest

Fig. 16. IP drum pressure decay best configuration (Plant 1, Plant 2, Plant 3,
and simulation) The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
relative error of less than 5%. The temperature drop is also repro­ the work reported in this paper.
duced correctly but with a higher deviation compared to the pressure
drop (about 10% relative error). The exhaust gas temperature and Data availability
mass flow rate of the gas turbine play an important role in further
improving the results obtained with this model, so modifying these The data that has been used is confidential.
values is a consistent way to obtain better simulations.

10
F. Alobaid et al. Applied Thermal Engineering 236 (2024) 121852

References [26] N. Mertens, F. Alobaid, R. Starkloff, B. Epple, H.-G. Kim, Comparative investigation
of drum-type and once-through heat recovery steam generator during start-up,
Appl. Energy 144 (2015) 250–260.
[1] I.E. Agency, CO2 emissions from fuel combustion: overview, Paris 2020 (2020).
[27] Z. Liu, I.A. Karimi, Simulating combined cycle gas turbine power plants in Aspen
[2] S. Dröge, The Paris Agreement 2015: Turning Point for the International Climate
HYSYS, Energ. Conver. Manage. 171 (2018) 1213–1225.
Regime, SWP Research Paper, 2016.
[28] J. McConnell, T. Das, A. Caesar, P. Veeravalli, Modeling and simulation of a
[3] V. Masson-Delmotte, P. Zhai, H.-O. Pörtner, D. Roberts, J. Skea, P.R. Shukla, Global
multistage heat recovery steam generator, Simulation 99 (2023) 169–182.
Warming of 1.5 C: IPCC special report on impacts of global warming of 1.5 C above
[29] G. Sabia, C. Heinze, F. Alobaid, E. Martelli, B. Epple, ASPEN dynamics simulation
pre-industrial levels in context of strengthening response to climate change.
for combined cycle power plant–Validation with hot start-up measurement, Energy
sustainable development, and efforts to eradicate poverty, Cambridge University
187 (2019), 115897.
Press, 2022.
[30] F. Alobaid, R. Starkloff, S. Pfeiffer, K. Karner, B. Epple, H.-G. Kim, A comparative
[4] S. Mor, M. Ghimire, Transparency and nationally determined contributions: A
study of different dynamic process simulation codes for combined cycle power
review of the Paris Agreement, Interdiscip. J. Econ. Business Law. 11 (2022)
plants–Part B: start-up procedure, Fuel 153 (2015) 707–716.
106–119.
[31] A.B. Ata, F. Alobaid, C. Heinze, A. Almoslh, A. Sanfeliu, B. Epple, Comparison and
[5] A. Alshahrani, S. Omer, Y. Su, E. Mohamed, S. Alotaibi, The technical challenges
validation of three process simulation programs during warm start-up procedure of
facing the integration of small-scale and large-scale PV systems into the grid: A
a combined cycle power plant, Energ. Conver. Manage. 207 (2020), 112547.
critical review, Electronics 8 (2019) 1443.
[32] B. El Hefni, Dynamic modeling of a hybrid solar-combined cycle power plant
[6] K. Guerra, P. Haro, R. Gutiérrez, A. Gómez-Barea, Facing the high share of variable
(ISCC) using a solar field based on parabolic trough solar collector-Start-up and
renewable energy in the power system: Flexibility and stability requirements, Appl.
shutdown of the solar field. AIP Conference Proceedings: AIP Publishing LLC;
Energy 310 (2022), 118561.
2018. p. 180005.
[7] F. Alobaid, N. Mertens, R. Starkloff, T. Lanz, C. Heinze, B. Epple, Progress in
[33] Y. Wang, D. Bhattacharyya, R. Turton, Dynamic Modeling and Control of a Natural
dynamic simulation of thermal power plants, Prog. Energy Combust. Sci. 59 (2017)
Gas Combined Cycle Power Plant for Load-following Operation, Elsevier, Computer
79–162.
Aided Chemical Engineering, 2019, pp. 101–106.
[8] J. Li, M.S. Ho, C. Xie, N. Stern, China’s flexibility challenge in achieving carbon
[34] F. Calise, L. Libertini, M. Vicidomini, Dynamic simulation and thermoeconomic
neutrality by 2060, Renew. Sustain. Energy Rev. 158 (2022), 112112.
analysis of a novel solar cooling system for a triple-pressure combined cycle power
[9] C. Vandervort, Advancements in H class gas turbines and combined cycle power
plant, Energ. Conver. Manage. 170 (2018) 82–96.
plants, Turbo Expo: Power for Land, Sea, and Air Am. Soc. Mech. Eng. (2018).
[35] E. Martelli, F. Alobaid, C. Elsido, Design optimization and dynamic simulation of
[10] S. Abens, F. Eulitz, I. Harzdorf, M. Jaenchen, W. Fischer, R. Rudolph, et al.,
steam cycle power plants: a review, Front. Energy Res. (2021:) 319.
Planning for extensive validation of the Siemens H-class gas turbine SGT5-8000H
[36] N. Mertens, F. Alobaid, T. Lanz, B. Epple, H.-G. Kim, Dynamic simulation of a
at the power plant irsching, ASME Power Conference (2009) 435–443.
triple-pressure combined-cycle plant: Hot start-up and shutdown, Fuel 167 (2016)
[11] T. Tanuma, Advances in steam turbines for modern power plants, Woodhead
135–148.
Publishing (2022).
[37] C. Vandervort, D. Leach, D. Walker, J. Sasser, Commercialization and Fleet
[12] Y. Watanabe, T. Takahashi, K. Suzuki, Dynamic simulation of the rapid load-
Experience of the 7/9HA Gas Turbine Combined Cycle. Turbo Expo: Power for
change performance of a large-scale gas turbine combined cycle power plant. The
Land, Sea, and Air: American Society of Mechanical Engineers; 2019. p.
Proceedings of the International Conference on Power Engineering (ICOPE)
V003T08A6.
202115: The Japan Society of Mechanical Engineers; 2021.
[38] M. Hänninen, Phenomenological extensions to APROS six-equation model. Non-
[13] S.C. Gülen, Gas Turbines for Electric Power Generation, Cambridge University
condensable gas, supercritical, 2009.
Press, 2019.
[39] M. Hänninen, J. Ylijoki, The one-dimensional separate two-phase flow model of
[14] H. Liu, Simulation study on the start-up process of heat recovery steam generator in
APROS: VTT Espoo, Finland, 2008.
F-class peak shaving combined cycle power plant, ITM Web Conf.: EDP Sci. (2022)
[40] M. Angerer, S. Kahlert, H. Spliethoff, Transient simulation and fatigue evaluation
03018.
of fast gas turbine startups and shutdowns in a combined cycle plant with an
[15] Y. Yang, T. Nikolaidis, S. Jafari, P. Pilidis, Gas turbine engine transient
innovative thermal buffer storage, Energy 130 (2017) 246–257.
performance and heat transfer effect modelling: A comprehensive review, research
[41] F. Alobaid, R. Starkloff, S. Pfeiffer, K. Karner, B. Epple, H.-G. Kim, A comparative
challenges, and exploring the future, Appl. Therm. Eng. 121523 (2023).
study of different dynamic process simulation codes for combined cycle power
[16] M.S.S. Danish, Z. Nazari, T. Senjyu, AI-coherent data-driven forecasting model for a
plants–Part A: Part loads and off-design operation, Fuel 153 (2015) 692–706.
combined cycle power plant, Energ. Conver. Manage. 286 (2023), 117063.
[42] J. Kurki, Simulation of thermal hydraulic at supercritical pressures with APROS,
[17] Y. Zhao, L.K. Foong, Predicting electrical power output of combined cycle power
2008.
plants using a novel artificial neural network optimized by electrostatic discharge
[43] A. Arkoma, M. Hänninen, K. Rantamäki, J. Kurki, A. Hämäläinen, Statistical
algorithm, Measurement 198 (2022), 111405.
analysis of fuel failures in large break loss-of-coolant accident (LBLOCA) in EPR
[18] M. Dirik, Prediction of NOx emissions from gas turbines of a combined cycle power
type nuclear power plant, Nucl. Eng. Des. 285 (2015) 1–14.
plant using an ANFIS model optimized by GA, Fuel 321 (2022), 124037.
[44] R. Starkloff, R. Postler, W.A.K. Al-Maliki, F. Alobaid, B. Epple, Investigation into
[19] D. Strušnik, J. Avsec, Exergoeconomic machine-learning method of integrating a
gas dynamics in an oxyfuel coal fired boiler during master fuel trip and blackout,
thermochemical Cu–Cl cycle in a multigeneration combined cycle gas turbine for
J. Process Control 41 (2016) 67–75.
hydrogen production, Int. J. Hydrogen Energy 47 (2022) 17121–17149.
[45] C. Schuhbauer, M. Angerer, H. Spliethoff, F. Kluger, H. Tschaffon, Coupled
[20] Y. Farahani, A. Jafarian, O.M. Keshavar, Dynamic simulation of a hybrid once-
simulation of a tangentially hard coal fired 700 C boiler, Fuel 122 (2014) 149–163.
through and natural circulation Heat Recovery Steam Generator (HRSG), Energy
[46] J. Hentschel, H. Zindler, H. Spliethoff, Modelling and transient simulation of a
242 (2022), 122996.
supercritical coal-fired power plant: Dynamic response to extended secondary
[21] A. Temraz, F. Alobaid, J. Link, A. Elweteedy, B. Epple, Development and validation
control power output, Energy 137 (2017) 927–940.
of a dynamic simulation model for an integrated solar combined cycle power plant,
[47] F. Alobaid, W.A.K. Al-Maliki, T. Lanz, M. Haaf, A. Brachthäuser, B. Epple, et al.,
Energies 14 (2021) 3304.
Dynamic simulation of a municipal solid waste incinerator, Energy 149 (2018)
[22] F. Casella, F. Pretolani, Fast start-up of a combined-cycle power plant: a simulation
230–249.
study with Modelica, Modelica Conference: The Modelica Association Linköping,
[48] J. Lappalainen, A. Tourunen, H. Mikkonen, M. Hänninen, J. Kovács, Modelling and
Sweeden (2006) 3–10.
dynamic simulation of a supercritical, oxy combustion circulating fluidized bed
[23] F. Alobaid, R. Postler, J. Ströhle, B. Epple, H.-G. Kim, Modeling and investigation
power plant concept—Firing mode switching case, Int. J. Greenhouse Gas Control
start-up procedures of a combined cycle power plant, Appl. Energy 85 (2008)
28 (2014) 11–24.
1173–1189.
[49] W.A.K. Al-Maliki, F. Alobaid, V. Kez, B. Epple, Modelling and dynamic simulation
[24] F. Alobaid, Numerical Simulation for Next Generation Thermal Power Plants,
of a parabolic trough power plant, J. Process Control 39 (2016) 123–138.
Springer, 2018.
[50] M. Hänninen, Phenomenological extensions to APROS six-equation model. Non-
[25] F. Alobaid, J. Ströhle, B. Epple, H.-G. Kim, Dynamic simulation of a supercritical
condensable gas, supercritical, 2009.
once-through heat recovery steam generator during load changes and start-up
procedures, Appl. Energy 86 (2009) 1274–1282.

11