Modeling and simulation of a complex ThermoSysPro model with

OpenModelica
Dynamic Modeling of a combined cycle power plant

El Hefni Baligh1 Bouskela Daniel1

1
EDF R&D STEP, Electricité de France, {baligh.el-hefni, daniel.bouskela}@edf.fr

Abstract ThermoSysPro is a generic library for the modeling

and simulation of power plants and other kinds of
ThermoSysPro (TSP) is a generic library for the
energy systems. ThermoSysPro library is developed by
modeling and simulation of power plants and other
EDF and released under open source license.
kinds of energy systems. TSP library is developed by
The foundations of the library are based on first
EDF and released under open source license. The library
physical principles: mass, energy, and momentum
features multi-domain modeling such as thermal-
conservation equations, up-to-date pressure losses and
hydraulics, neutronics, combustion, solar radiation,
heat exchange correlations, and validated fluid
instrumentation and control.
properties functions. The correlations account for the
Numerous organizations and individuals worldwide
non-linear behavior of the phenomena of interest. They
now use TSP. Until recently, the TSP library could be
cover all water/steam phases, oil, molten salt and all flue
used only under Dymola for the modeling and
gas compositions. The granularity of the modeling may
simulation of complex power plants. But now, with the
be freely chosen. Some correlations are given by default
latest version of OpenModelica (OM), we can simulate
since they correspond to the most frequent use-cases,
complex models of power plants with complex
but they can be freely modified by the user if needed.
scenarios.
This includes the choice of the pressure drop or heat
To be able to use TSP under OM, some adaptations
transfer correlations. Special attention is given to the
have been applied to our models, essentially the method
handling of two-phase flow, as two-phase flow is a
used to make inverse computation.
common phenomenon in power plants.
The objective of this work is to evaluate the
The library components are written in such a way that
potentiality, capability and efficiency of using
there are no hidden or unphysical equations, that
OpenModelica tools to perform dynamic studies of
components are independent from each other and to
power plants. A combined cycle power plant has been
ensure as much as possible upward and downward
chosen as a representative test case of the complexity of
compatibility across tools and library versions. This is
this type of study.
particularly important in order to control the impact of
The paper describes the dynamic model of a
component, library or tool modifications on the existing
combined cycle power plant, whose objective is to study
models.
a step variation load from 100% to 50% and a full gas
This library is aimed at providing the most frequently
turbine trip, using OM software. Also, the structure of
used model components for the 0D-1D static and
the model, the parameterization data, the results of
dynamic modeling of thermodynamic systems, mainly
simulation runs, the difficulties encountered using OM
for power plants, but also for other types of energy
and the comparison between Dymola and OM are
systems such as industrial processes, energy conversion
presented.
systems, buildings etc. It involves disciplines such as
thermal-hydraulics, combustion, neutronics and solar
Keywords: Modelica; OpenModelica; ThermoSysPro; radiation.
thermal-hydraulics; combined cycle power plant; The ambition of the library is to cover all the phases
dynamic modeling; inverse problems. of the plant lifecycle, from basic design to plant
operation. This includes for instance system sizing,
1 Introduction verification and validation of the instrumentation and
control system, system diagnostics and plant
Modeling and simulation play a key role in the design monitoring. To that end, the library will be linked in the
phase and performance optimization of complex energy future to systems engineering via the modeling of
processes. They also play a significant role in the future systems properties, and to the process measurements via
for power plant maintenance and operation. data reconciliation and data assimilation.

DOI Proceedings of the 12th International Modelica Conference 407

10.3384/ecp17132407 May 15-17, 2017, Prague, Czech Republic
 Modeling and simulation of complex ThermoSysPro model with OpenModelica - Dynamic Modeling of a
combined cycle power plant

The library may be downloaded freely together with The inverse problem basically consists in setting
the OpenModelica software from (fixing) a state variable of the model to a known
https://openmodelica.org/download/download- measurement value to compute by model inversion the
windows. value of a parameter or a boundary condition.
Several test-cases were developed to validate the Modelica allows to express inverse problems, which
library in order to cover the full spectrum of use-cases is a powerful feature for computing operation points,
for power plant modeling: which are defined by their observable outputs, and for
 Dynamic model of a 1300 MWe nuclear power system sizing, to compute parameterised characteristics.
plant covering the primary and secondary loops, To implement the inverse problem under Dymola, it
 Dynamic model of steam generators for sodium is enough simply to fix the value of the state variable
fast reactor (David F., Souyri A. and Marchais G, and declared it to (fixed = true) and released the
2009) parameter to be computed and declared it as (fixed =
 Static and dynamic models of a biomass plant (El false). However, this method is incompatible with
Hefni B. and Péchiné B, 2009), OpenModelica (no standard Modelica language).
 Physics/neutronics model in Modelica for a tool, Here is a simple example to illustrate the deference
to assist the operator, to control the power plant between Dymola and OM for the implementation of the
for infrequent transients and to establish a reverse problem (standard Modelica language).
strategy of optimal operating procedure (El Hefni Furthermore, for the demonstration we use a simple
B., 2011), model to calculate the pressure drop in a tube, so:
 Dynamic model of a concentrated solar power Q Q
Pi  Po  K 
plant (El Hefni B., 2013), 
 Dynamic multi-configuration model of a 145 Pi is the fluid pressure at the singularity inlet (Pa) , Po
MWe concentrated solar power plant with the
ThermoSysPro library (tower receiver, molten is the fluid pressure at the singularity outlet (Pa), Q is
salt storage and steam generator)’, (El Hefni B., the fluid mass flow rate (kg/s), K is the
Soler R., 2014),
friction pressure loss coefficient (m-4 ) and  is the
 Dynamic simulation of a 1MWe CSP tower plant 3
with two-level thermal storage implemented with average density of the fluid (kg/m ).
control system (S.J. Liua et al., 2014),
As the above equation is implemented in a TSP
 Dynamic simulation and experimental validation component model called SingularPressureLoss, this
of Open Air Receiver and Thermal Energy model component is used to illustrate inverse
Storage systems in solar thermal power plant calculation. The model uses the following component
(Qing Li et al., 2015). models (see Figure 1):
The objective of this paper is to show the potentiality,
 SingularPressureLoss model,
capability and efficiency of OpenModelica tools to
perform dynamic studies using complex models such as  SourceP model,
the combined cycle power plant model.  SinkP model.
In order to challenge the dynamic simulation
capabilities of the library, a step load variation from P0= 3e5 Pa P0= 1e5 Pa
100% to 50% and a turbine trip (sudden stopping of the
gas turbine) were simulated.

2 How to use OpenModelica for Figure 1. TestSingularPressureLoss model (test-

case).
inverse problems (model inversion)
As the initial state of the simulation is in general The equation above makes it possible to calculate the
defined by the observable outputs of the system (e.g. the flow rate of the fluid through the tube, provided that the
nominal power output, the pressure inside the boiler, pressure drop in the tube, the coefficient of the pressure
etc.), it is necessary to solve an inverse problem to drop and the fluid density are known (parameters).
compute the initial state. Moreover, it is necessary to The model inversion (calibration) consists in setting
start the simulation from a stationary (or steady) state in the mass flow rate of the fluid through the tube (Q) and
order to avoid the numerical difficulties which arise the friction pressure loss coefficient of the pipe (K) can
when the system is started out of equilibrium be computed. To express this inverse problem with
(oscillations, stiffness …). Dymola, it suffices to write: [Q (fixed = true, start =
500)] and [K (fixed=false, start=100] in the parameters

408 Proceedings of the 12th International Modelica Conference DOI

May 15-17, 2017, Prague, Czech Republic 10.3384/ecp17132407
 Session 6: Poster Session

windows of the SingularPressureLoss model (see Steam Generator (HRSG): Thermal power: 2*360
Figure 2). MW,
Steam Turbine: Nominal power: 277 MW,
Condenser:
Thermal power: 428 MW.
Outlet water temperature: 305 K
Vacuum pressure: 6100 Pa.

3.1 Model description

Currently, two models are used: one to simulate the
power generator step reduction load (see Figure 4), the
other to simulate the full GT trip (see Figure 5). In the
model used to simulate the GT trip, the gas turbine is
replaced by a boundary condition.
The model contains two main parts: the water/steam
cycle and the flue gases subsystem. Only one train is
Figure 2. Data for the SingularPressureLoss model. modelled, so identical behavior is assumed for each
HRSG and for each gas turbine.
On the other hand, to express this inverse problem
with OpenModelica (also valid for Dymola), it is HRSG model:
necessary to write: [Q (fixed = true, start = 500)] and The model consists of 16 heat exchangers (3
K = K_pipe in the parameters windows of the evaporators, 6 economizers, 7 super-heaters), 3
SingularPressureLoss model (see Figure 3). evaporating loops (low, intermediate and high pressure),
Also, the parameter K_pipe (new intermediary 3 drums, 3 steam turbine stages (HP, IP and LP), 3
parameter), must be created (declared) in the main pumps, 9 valves, several pressure drops, several mixers,
model, with [K_pipe (fixed=false, start =1.e2)], see several collectors, 1 condenser, 1 generator, several
Table 1. sensors, sources, sinks and the control system limited to
the drums level control.
An important feature of this model is that the
thermodynamic cycle is completely closed through the
condenser. This is something difficult to achieve,
because of the difficulty of finding the numerical
balance of large closed loops.
The list of components used for the development of the
HRSG model is given in Table 2.

Table 2. Library components used in the HRSG

model.
Type Model name in the library
Condenser DynamicCondenser
Drum DynamicDrum
Figure 3. Data for the SingularPressureLoss model.
Generator Generator
Table 1. The declaration of the K_pipe in the main Heat DynamicExchangerWaterSteamFlue
model. exchanger Gases
model TestSingularPressureLoss
=
parameter Real K_pipe (fixed=false,start=1.e2)
DynamicTwoPhaseFlowPipe
"Pressure loss coefficient";
ExchangerFlueGasesMetal
HeatExchangerWall
equation Pipe LumpedStraightPipe
Pump StaticCentrifugalPump
3 Combined cycle power plant model
Steam StodolaTurbine
The power plant model is a complete model of a real
turbine
combined cycle power plant:
Gas Turbine (GT): Nominal power: 2*226 MW, Valve ControlValve

DOI Proceedings of the 12th International Modelica Conference 409

10.3384/ecp17132407 May 15-17, 2017, Prague, Czech Republic
 Modeling and simulation of complex ThermoSysPro model with OpenModelica - Dynamic Modeling of a
combined cycle power plant

Water VolumeB, VolumeC Based on correlations for the characteristic.

mixer Combustion chamber:
Water VolumeA, VolumeD Based on first principles mass, momentum and
splitter energy balance equations. The pressure loss in the
combustion chamber is taken into account.
Heat Exchanger: The “CombinedCyclePowerPlant” model contains
Based on first principles mass, momentum and 673 component models, generating 10802 variables,
energy balance equations, the following phenomena are 257 differentiated variables, 2752 equations and 1855
represented: nontrivial equations.
 Transverse heat transfer,
3.2 Data implemented in the model
 Mass accumulation,
All geometrical data were provided to the model
 Thermal inertia, (pipes and exchangers lengths and diameters, heat
 Gravity, transfer surfaces of exchangers, volumes …).
 Pressure drop within local flow rate. The plant characteristics are given below.
Gas Turbine (GT)
Drum and Condenser:
Compressor compression rate: 14
Based on first principles mass and energy balance
equations for water and steam, the following Steam Generator (HRSG)
phenomena are represented: HRSG with 3 levels of pressure.
High pressure circuit at nominal power: 127 bar
 Drum level and swell and shrink phenomenon,
Intermediate pressure circuit at nominal power: 27 bar
 Heat exchange between the steam/water and the Low pressure circuit at nominal power: 5.0 bar
wall,
Steam Turbine
 Heat exchange between the outside wall and the
High pressure at nominal power: 124.5 bar, 815 K
external medium.
Intermediate pressure at nominal power: 25.5 bar, 801K
Steam turbine: Low pressure at nominal power: 4.8 bar, 532 K
Based on an ellipse law and an isentropic efficiency.
Condenser
Pump: Steam flow rate: 194 kg/s
Based on the characteristics curves. Water temperature at the inlet: 306 K
Pressure drop in pipes:
Proportional to the dynamic pressure ± the static 3.3 Model calibration
pressure.
The calibration phase consists in setting the
Mixer/splitter: maximum number of thermodynamic variables to
Based on the mass and energy balances for the fluid. known measurement values taken from on-site sensors
GT model: for 100% load. This method ensures that all needed
The model consists of 1 compressor, 1 gas turbine, 1 performance parameters and size characteristics can be
combustion chamber, sources, sinks and 1 air humidity computed. The variables imposed in the model are:
model.  Pressure at the outlet of the pumps,
The list of component models used for the  Pressure at the inlet of the steam turbines,
development of the GT model is given in Table 3.  Specific enthalpy at the inlet of the steam turbines,
 Liquid level in drums and in condenser,
Table 3. Library components used in the GT model.
 Overall heat exchangers coefficients,
Type Model name in the library
 Isentropic efficiency of the compressor,
Air humidity AirHumidity
 Exhaust temperature of the gas turbine,
Compressor GTCompressor  …
Gas turbine CombutionTurbine
Combustion GTCombustionChamber The main computed performance parameters are:
chamber  The characteristics of the pumps,
 The ellipse law coefficients of the steam turbines,
Gas turbine:  The isentropic efficiencies of the turbines,
Based on correlations for the characteristic.  The CV of the valves and the valves positions
Compressor: (openings).

410 Proceedings of the 12th International Modelica Conference DOI

May 15-17, 2017, Prague, Czech Republic 10.3384/ecp17132407
 Session 6: Poster Session

 Heat exchangers fouling coefficients, The computational time is faster than real time.
 Nominal isentropic efficiency of the compressor,
3.7 Comparison between Dymola and
 Nominal isentropic efficiency of the gas turbine,
OpenModelica
 …
The simulation results of OM are very close to the
simulation results of Dymola. The simulation time with
3.4 Simulation scenarios
OM is between 15% and 60% times slower than the
For simulation runs, two scenarios were selected. The simulation time with Dymola, depending on the
first scenario is a power generator step reduction from scenario and the tolerance (see Table 4).
100 to 50% load:
 Initial state (combined cycle): 100 % load
Table 4. Simulation time.
 Final state (combined cycle): 50% load (800 s
Simulation time (s)
slope).
Dymola 2017 OpenModelica1.11.0
The second scenario is a full GT trip (sudden Tolérance=0,001 Tolérance=0,0001 Tolérance=0,001 Tolérance=0,0001
stopping of the gas turbine): Variation de charge 58 73 75 84
 Initial state (GT exhaust): 894 K, 607 kg/s (simulation 2500 s)
Trip TAC 174 310 240 492
 Final state (GT exhaust): 423 K, 50 kg/s (600 s (simulation 10000 s)
slope).
However, OM is still 20 times faster than real time in
the worst case.
3.5 Simulation
Simulation runs were done using Dymola 2017 and 4 Conclusion
OpenModelica 1.11.0. The simulation of the scenarios A dynamic and rather large model of a combined
were mostly successful. However, some difficulties cycle power plant has been developed to validate the
were encountered when simulating large transients, ThermoSysPro library. This model comprises the flue
mainly stemming from the large size of the model: gas side and the full thermo-dynamic water/steam cycle
 Poor debugging facility, closed through the condenser. Two difficult transients
 Large number of values to be manually provided by were simulated with Dymola 2017and OpenModelica
the user for the iteration variables, for the two tools. 1.11.0: a step reduction load of the power generator and
a full gas turbine trip. The results are mostly consistent
In particular, it has been observed that, the two tools with the engineer’s expertise.
cannot calculate the initial states, when all iterations Despite of some simulation difficulties because of the
variables are not set close to their solution values. lack of debugging tools for Modelica models, this work
shows that the library is complete and robust enough for
3.6 Simulation results
the modelling and simulation of complex power plants
The model is able to compute: with the two software.
 The air excess, The simulation results of OM are very close to the
 The distribution of water and steam mass flow rates, simulation results of Dymola. The simulation time with
 The thermal power of heat exchangers, OM is slower than simulation time with Dymola, but
still 20 times faster than real time.
 The electrical power provided by the generator,
This work shows that OpenModelica software is very
 The pressure temperature and specific enthalpy satisfying for thermo-hydraulic modelling and
distribution across the network, simulation.
 The drums levels and the condenser level, Acknowledgements
 The performance parameters of all the equipments, This work was partially supported by the OPENCPS
 The global efficiencies of the water/steam cycle and project.
gas turbine.
References
The results of the simulation runs are given in Figure
David F., Souyri A. and Marchais G., ‘Modeling Steam
6 and Figure 7. They are consistent with the engineer’s
Generators for Sodium Fast Reactors with Modelica’,
expertise. The comparison between simulation Modelica 2009 conference proceedings.
results and GE (General Electric) results
El Hefni B. and Péchiné B., ‘Model driven optimization of
(FOUQUET L, 2004) for 100 % load and 50 % load, biomass CHP plant design’, Mathmod conference 2009,
have shown that the Simulation results are very close Vienna, Austria.
to the GE values (Design results).

DOI Proceedings of the 12th International Modelica Conference 411

10.3384/ecp17132407 May 15-17, 2017, Prague, Czech Republic
 Modeling and simulation of complex ThermoSysPro model with OpenModelica - Dynamic Modeling of a
combined cycle power plant

El Hefni B. ‘Dynamic modeling of concentrated solar

power plants with the ThermoSysPro Library (Parabolic Liua S.J., Faille D., Fouquet M., El Hefni B., Wangc Y.,
Trough collectors, Fresnel reflector and Solar-Hybrid)’, Zhang J.B., Wang Z.F.,.Chen G.F and Soler R., ‘Dynamic
SolarPaces 2013, Elsevier’s Energy Procedia, simulation of a 1 MWe CSP tower plant with two-level
El Hefni B., ‘Modèle physique/neutronique en Modelica thermal storage implemented with control system’,
d’un outil d’aide au pilotage du transitoire sensible de montée SolarPaces 2014, Elsevier’s Energy Procedia.
en puissance à 3%Pn/h après rechargement. Maquettage d’un El Hefni B., Bouskela D. Lebreton G., ‘Dynamic
outil d’aide au pilotage sous Excel/VB’, LMCS 2011. modelling of a combined cycle power plant with
El Hefni B. and Soler R. ‘Dynamic multi-configuration ThermoSysPro’, MODELICA 2011 conference.
model of a 145 MWe concentrated solar power plant with the FOUQUET L, EDF Pôle Industrie, CNET:
ThermoSysPro library (tower receiver, molten salt storage and Y.PM.X.000.PPPP.00.X.0872: PhuMy2.2: Overall plant
steam generator)’, SolarPaces 2014, Elsevier’s Energy operation description, 2004.
Procedia.
Qing Li，Fengwu Bai，Bei Yang，Baligh El Hefni and
Sijie Liu, ‘Dynamic simulation and experimental validation of
Open Air Receiver and Thermal Energy Storage systems in
solar thermal power plant’, SWC 2015 en Koré 2015.

Figure 4. Model of the combined cycle power plant used for the power generator step reduction load.

Figure 5. Model of the combined cycle power plant used for the full GT trip.

412 Proceedings of the 12th International Modelica Conference DOI

May 15-17, 2017, Prague, Czech Republic 10.3384/ecp17132407
 Session 6: Poster Session

Figure 6. Power generator step reduction simulation (-50%): natural gas mass flow rate, air mass flow rate, excess air
temperature at the inlet of the combustion turbine, exhaust temperature (gas turbine), mechanical power of the combustion
turbine, mechanical power produced by each steam turbine, generator power, HRSG temperature at the outlet, steam mass
flow rate produced in each drum, the drums pressure, and the drums level.

DOI Proceedings of the 12th International Modelica Conference 413

10.3384/ecp17132407 May 15-17, 2017, Prague, Czech Republic
 Modeling and simulation of complex ThermoSysPro model with OpenModelica - Dynamic Modeling of a
combined cycle power plant

Figure 7. Gas turbine trip simulation: flue gases temperature at the inlet of the HRSG, flue gases mass flow rate at the inlet
of the HRSG, generator power, the drums pressure, steam mass flow rate produced in each drum, thermal power produced
in each heat exchanger (Evaporators HP, IP, LP and economizers HP, IP, LP), and steam mass flow rate in each steam
turbine.

414 Proceedings of the 12th International Modelica Conference DOI

May 15-17, 2017, Prague, Czech Republic 10.3384/ecp17132407