The ThermoPower Library

Francesco Casella
([email protected])

Dipartimento di Elettronica e Informazione

Politecnico di Milano
Power Plant Dynamic Modelling at DEI - Polimi

1985-1993
• LEGO simulation environment (developed by ENEL)
– Modular modelling, solver for index-1 DAE's, models implemented in
FORTRAN
– Wide availability of tested models, very low flexibility and extensibility,
extremely old-fashion development environment (native FORTRAN code)

1993-2003
• ProcSim simulation environment
(Maffezzoni, Leva, Bartolini)
– Modular modelling, specialized implicit Euler solver for thermo-hydraulic
system equations, models implemented using LabVIEW
– Very good run-time user interface, good avaliability of models, efficient, low
flexibility and extensibility

2003-
• ThermoPower library, using Modelica
(Casella, Leva)
– Object-Oriented paradigm, using the Modelica language
– Maximum flexibility and extensibility, growing model base, highly readable
models
2
Scope of the Library

Support for the generation of power plant models:

– Modularity
– Dynamic models
– Degree of detail suited to system studies (not for component studies)
• Non-linear, first-principles models
• Lumped-parameter models
• 1D Distributed parameter models
• NO 3D CFD models

Usage
– Dynamic plant simulation
– Computer-aided design of control strategies and controller architectures
– Validation and tuning of control systems
– Simulation study of large plant maneuvers (e.g. start-up)
– Plant maneuver optimization

Type of plants:
– Traditional fossil-fired Rankine cycle
– Gas Turbine plants
– Combined cycles
– Combined heat&power plants
– Nuclear plants (gen III & IV)
– Organic Rankine Cycle plants
– Solar plants
3
Issues with Object-Oriented Modelling of Power Plants

ISSUE #1: Abstract model structure

• Some components have clearly defined interfaces (flanges, shafts, etc.)

/
• Other components have a far less obvious structure

(A)
(B)

Gas flow ?
The basic models must allow to describe
complex heat transfer configurations
4
Issues with Object-Oriented Modelling of Power Plants

ISSUE #2: Many completely different models

to describe the same physical object

Liquid-filled pipe, with heat • One-phase vs. Two-phase

transfer through the lateral • Nature of the fluid
boundary • Radial velocity distributions
• Compressibility and inertial effects
• Pressure wave propagation
• Discretization method for the PDE
• Etc. etc. ...
•

It is hard to define standard o basic thermo-

hydraulic models for general-purpose use

5
Structure of the Library – I

Problems with extensive use of inheritance

– Difficult to define a hyerarchy of heterogeneous models
– Modifying the “ancestor” equations can lead to hard-to-understand consequences on the
“siblings”
– The equations of a single model are scattered through many classes, thus hampering:
• readability
• modifiability / extensibility
Judicious use of inheritance
– Flat library structure
– Inheritance is used to:
• change the fluid property model
• define families of very similar components (e.g. pumps, valves)
• add advanced features to already functional components
Modelling assumptions are selected through conditional equations, depending on
boolean/integer parameters ⇒ low number of models with many options, e.g:
– Options for the computation of the friction losses
– Dynamic term in the momentum balance equation
– Geometric characteristics
– Type of initialization
– ...

6
Structure of the Library – II

Type of fluids
– Water / steam for the steam side of the generation processes
(fluids other than water may be used)
– Ideal gases (air, natural gases, flue gases etc.)

Models of the same component (e.g. a valve) with different fluids

– Choice 1:independent models
– Choice 2: generic model + extensions with fluid-specific equations
– Problem: changing the fluid, the modelling assumptions can vary drastically
– The reuse of part of the equations actually make things more confusing

• Two separate categories of models: water/steam and ideal gases

• Within each family, the fluid model is replaceable

7
Old Design of Flange Connectors (ver. 2.x)

• Thermodynamic conditions are specified with a minimum number of non-redundant

variables (Pressure, mass flow rate, specific enthalpy).
• Maximum independence from how the internal equations of each component are
written (encapsulation).
• Support of reverse flow

A B A B

B A

connector WaterFlangeB connector WaterFlangeA

Pressure p; Pressure p;
flow MassFlowRate w; flow MassFlowRate w;
output SpecificEnthalpy hBA; input SpecificEnthalpy hBA;
input SpecificEnthalpy hAB; output SpecificEnthalpy hAB;
end WaterFlangeB end WaterFlangeA

8
 Using Flange Connectors: a Simple Example

model SimpleMixer
parameter Volume V=0 "Inner volume";
FlangeA in;
FlangeB out;
Pressure p;
Density rho; hBA hBA
Mass M;
Energy E;
hAB hAB
SpecificEnthalpy h,hi,ho; in out
equation (FlangeA) (FlangeB)
rho = rho_ph(p, h);
M = V*rho;
E = M*h – p*V;
der(M) = in.w+out.w;
der(E) = in.w*hi+out.w*ho;
hi = if in.w >= 0 then in.hBA else h;
ho = if out.w >= 0 then out.hAB else h;
in.hAB = h;
out.hBA = h;
in.p = p;
out.p = p;
end SimpleMixer;

9
DHT Connectors (Distributed Heat Transfer)

Describe the heat transfer between distributed parameters models

connector DHT;
parameter Integer N=2 "Number of nodes";
Temperature T[N];
flow HeatFlux phi[N];
end DHT;

Extended DHT connector with heat transfer coefficient

(computed internally based on the fluid conditions)

connector DHThtc;
extend DHT;
CoefficientOfHeatTransfer gamma[N];
end DHThtc;

10
Computing the Fluid Properties: Modelica.Media

The fluid properties are computed using the models contained in the standard
Modelica.Media library. The fluid propery models and computational routines are
contained in replaceable packages.
Usage example::
model Mixer
replaceable package Medium = Modelica.Media.Interfaces.PartialMedium;
Medium.BaseProperties fluid;
...
equation
// The fluid properties are determined prescribing two thermodynamic
// variables + the composition (for multiple-component fluids)
// pressione, temperatura e composizione
fluid.p = p;
fluid.T = T;
fluid.Xi = Xi;

// all other base properties (u,d,T) are automatically computed by

// the fluid model
h = fluid.h;
d = fluid.d;

// it is also possible to compute additional fluid properties, using

// the thermodynamic state vector contained in the fluid model
lambda = Medium.thermalConductivity(fluid.state);
...
end Mixer;
11
Computing the Fluid Properties: Modelica.Media

• The models are written for a generic fluid (PartialMedium, or

PartialTwoPhaseMedium);
• When models are instantiated, it is possible to redeclare the medium package to
specify the required fluid model (which extends PartialMedium or
PartialTwoPhaseMedium)
model Plant
Mixer M1(redeclare package Medium = Modelica.Media.Air.MoistAir);
Mixer M2(redeclare package Medium = Modelica.Media.IdealGases.FlueGas);
...
end Plant;

• La standard Modelica.Media library contains several ready-to-use models

– IAPWS IF97 water/steam properties
– Accurate Ideal gas models (based on NASA data)
– Humid air
– Incompressible fluids, with table-based properties
– ...
• It is possible to develop new fluid models in Modelica (using the interfaces
defined in Modelica.Media.Interfaces)
• It is also possible to use FluidProp models, through the ExternalMedia interface.

12
Features of ThermoPower ver. 3

• Uses Modelica Standard Library 3.x

• Standard Modelica.Fluid connectors are now used

– same mathematical structure
– no more complementary ports
– compatible with Modelica.Fluid components

• Homotopy operator used for more robust initialization

– requires homotopy() support
– if not available, dummy homotopy() function must be loaded

• Not yet released officially (later this year)

• Development version available on SVN (anonymous read-only checkout):

https://thermopower.svn.sourceforge.net/svnroot/thermopower/trunk

13
 Complex Heat Transfer Configurations

Complex configurations like the one shown here can be modelled.

The modeller must write specific code for the heat transfer module. The wall models
and the Flow1D models are taken straight from the library

(A)
(B) (Fluid B)

Gas flow (Gas Flow)

(Wall B)

(Heat Transfer
Module)

(Wall A)

(Fluid A)
14
Applications of the ThermoPower library

15
CISE

16
CISE

• First real-life application

• Simulation of a lab model of a
steam generator (100 kW,
electrically powered)

• Available data:
– Dimensional data
– Open-loop transient recordings

• Medium complexity, in terms of

number of components
• Maximum achievable degree of
detail of the single components

17
CISE

• Model diagram
• Complexity (after the symbolic
manipulation phase):
– 50 states
– 300 algebraic variables

18
CISE - Validation

• Pressure: 60 bar
• Step reduction of the power to the evaporators

19
CISE - Validazione

• Pressure: 60 bar
• Step reduction of the power to the evaporators

20
CISE - Validazione

• Pressure: 30 bar
• Step increase in the steam valve opening

21
CISE - Validazione

• Pressure: 60 bar
• Step increase in the steam valve opening

22
IRIS

23
IRIS

• First industrial application: IRIS consortium (PoliMi +

Westinghouse + others)
• Study of the control system for a new generation PWR
nuclear plant
• “Lightweight” models needed to support the control
design activities
• Integration of the ThermoPower and NuKomp libraries
for the modelling of the nuclear core (Fourier
equations for the fuel rods + point neutronic kinetics)
• Set up of a family of consistent plant models
– Accurate reference model, to be compared with
detailed simulation code results (RELAP)
– More/less detailed control system models
– More/less detailed turbine + feedwater train
models
– More/less detailed models of the steam generator
– Modelli for very fast simulations

24
IRIS

• Base model: four empty units (partial models) connected together.

• Actual simulators are obtained by instantiating the base model and replacing the
units with one of their available implementations

25
IRIS

• NSSS diagram

Replaceable models

26
IRIS

Core diagram:

Integration of ThermoPower and

NuKomp components

27
IRIS

• “Library of simulators”
– Interface (empty shell with connectors)
for each unit
– Many implementation for each unit
– Common repository of geometric and
design data
– Replaceable fluid models
• Instantiation of a specific simulator
– A' la carte choice of details
model IRISSimulator_V2
extends IRISSimulatorBase(
redeclare Plants.NSSS_V1 NSSS(
redeclare package PrimaryMedium =
Media.SimpleIncompressibleWater,
Core(N = 7),
redeclare Plants.HelicalCoilFVChen
HelicalCoil(N=15)),
redeclare Controls.CS_V2 CS,
redeclare Controls.SS_V2 SS,
redeclare Plants.TGFWS_V1 TGFWS);
end IRISSimulator_V2;

• All the implementable simulators are

consistent by design! 28
IRIS – Validation vs. RELAP code

• The model has been validated in open loop agains results obtained from a RELAP
model (good for accident simulation and safety analysis, but very slow)

29
IRIS – Test of Closed-Loop transients

• Simulazions carried out with a

simplified model of the digital
control system (5 main control
loops)
• In the future: multi-mode
controller, including start-up and
shut-down
• Using an incompressible fluid
model for the primary loop, a
non-stiff model is obtained
• This can be simulated using a
fixed-time-step forward Euler
algorithm, resulting in 20X real-
time simulation speed.

30
FLUE-GAS PATH
SIMULATOR

31
Flue-Gas Path Simulator

• Study of the pressure control system in the flue gas circuit for a revamped coal-
fired power plant
– SCR unit
– De-NOx unit
– De-SOx unit
– Booster fan
• A CFD study was available for some selected static operating conditions; it was
then necessary to check if the control system can carry out all the foreseen
maneuvers without tripping the plant:
– Load level increase / decrease due to dispatching
– Insertion / Exclusion of the De-SOx unit
– Fan unit trip
• Highlights:
– Ad-hoc con know-how proprietario (operating curves)
– Very detailed modelling of the logic control system

32
Flue-Gas Path Simulator

• Plant model: high-level diagram

33
Flue-Gas Path Simulator

• Model of a fan unit

34
Flue-Gas Path Simulator

• Control system block diagrams

• Each logic blocks is specified through an algorithm

Furnace pressure control Bypass insertion sequencer

35
CESI - CC

36
CESI - CC

• Development of a dynamic model for a typical Combined-Cycle unit, to study fast

start-up strategies
• The model includes the computation of the thermal stresses in the most critical
points of the plant (steam turbine shaft)
• The control strategy must be selected in order to minimize the start-up time (from
shut-down plant up to full load), without exceeding the maximum allowed stress
levels.
• The strategies have been determined by trial-and-error. In the future, the model
could be coupled to an optimization software, to compute the optimal control
signals automatically
• Also, simplified versions of the model could be used to build model-based
controllers (e.g. predictive controllers)
• Highlights:
– High complexity (large number of componentsi)
– Start-up conditions included (low or zero flow rates, very low pressures, etc.)

37
CESI - CC

• High-level structure of the plant simulator

38
CESI - CC

• Heat recovery boiler model

39
CESI - CC

• Steam turbine unit model

40
CESI - CC

• Low-level controllers

41
CESI – CC: Simulation Results

• Reference start-up transient: steam flow rates

42
CESI – CC: Simulation Results

• Reference start-up transient: steam and HP turbine rotor temperatures

43
CESI – CC: Simulation Results

• Reference start-up transient: thermal stresses on HP and MP turbine rotors

44
CESI – CC: Simulation Results

• Minimum-time start-up transient:: thermal stresses on HP and MP turbine rotors

45
Modelling of Combined-Cycle Power Plants

• Circulation type for evaporators

– forced circulation
– natural circulation
– once-through (seldom used)
• Cycle configuration
– two pressure levels no reheat
– three pressure levels w/ reheat
• Steam turbine configuration
– Arrangement of HP/IP/LP stages
– Single- or double-flow
• Plant configuration
– Single shaft (1 GT/HRSG, 1 ST on same generator shaft)
– Double shaft (1 GT/HRSG, 1 ST on separate generator shafts)
– Triple shaft (2 GT/HRSG, 1 ST, three generator shafts)
– …
• Goal of the PowerPlants package in ThermoPower:
– provide abstract interfaces for all these categories
– provide template pre-built models for at least some cases
– avoid repetitive and tedious work when setting up a new model

46
Heat exchanger unit

• Interface has 4 connectors

• Different implementations using
basic components from the library
– 1 phase / 2 phase
– heat transfer phenomena
• fixed/varying h.t.c.
• convection
• radiation
• ad-hoc correlations

47
Heat Recovery Steam Generator body

• Two or three levels of

pressure

• Two interfaces
defined accordingly

48
Heat Recover Steam Generator body

• Example implementation of two-levels of pressure HRSG

49
Drum units

• Two and three levels of • Different implementations

pressure interfaces – natural circulation (no pumps)
– forced circulation (w/ pumps)

50
Complete HRSG unit

• Interfaces defined for two and • Example implementations built with HRSG
three levels of pressure body and drum units

51
Steam turbine unit

• Interfaces for • Different implementations

– two levels of pressure, no reheat – simpler model for normal
– three levels of pressure w/ reheat operation only
– complete model including
bypass paths for start-up
simulation

52
Complete plant model

• Using the unit models defined before, one can easily assemble a complete
plant model with any specific configuration

53
References

• ThermoPower library Home page http://home.dei.polimi.it/casella/thermopower/

• A. Leva, C. Maffezzoni: “Modelling of power plants”, Thermal power plant simulation and
control, D. Flynn (Ed.), IEE, London, pp. 17–60, 2003.
• F. Casella, A. Leva: “Modelica open library for power plant simulation: design and
experimental validation”, Proc. Modelica Conference 2003, Linköping, Sweden, 2003.
F. Casella, A. Leva: “Modelling of thermo-hydraulic power generation processes using
Modelica”, Mathematical and Computer Modelling of Dynamical Systems, Vol. 12, No. 1,
February 2006, 19 – 33, 2006
• Francesco Casella and Alberto Leva, “Modelling of Thermo-Hydraulic Power Generation
Processes Using Modelica”. Mathematical and Computer Modeling of Dynamical Systems,
vol. 12, n. 1, pp. 19-33, Feb. 2006.
• Antonio Cammi, Francesco Casella, Marco Ricotti and Francesco Schiavo, “Object-Oriented
Modeling, Simulation and Control of the IRIS Nuclear Power Plant with Modelica”. In
Proceedings 4th International Modelica Conference, Hamburg, Germany, Mar. 7-8, 2005, pp.
423-432.
• Francesco Casella and Francesco Pretolani, “Fast Start-up of a Combined-Cycle Power
Plant: a Simulation Study with Modelica”. In Proceedings 5th International Modelica
Conference, Vienna, Austria, Sep. 6-8, 2006, pp. 3-10.
• Andrea Bartolini, Francesco Casella, Alberto Leva and Valeria Motterle, “A Simulation Study
of the Flue Gas Path Control System in a Coal-Fired Power Plant”. In Proceedings ANIPLA
International Congress 2006, Rome, Italy, Nov. 13-15, 2006.
• Francesco Casella, “Object-Oriented Modelling of Power Plants: a Structured Approach”. In
Proceedings of the IFAC Symposium on Power Plants and Power Systems Control, Tampere,
Finland, July 5-8, 2009.

54