Upload
abbas-mohajer
View
216
Download
0
Embed Size (px)
Citation preview
7/31/2019 GT2007-27086-pre
1/34
1Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Laboratory of Thermal Turbomachines
National Technical University of AthensNational Aerospace Laboratory, NLR
The Netherlands
Advanced Capabilities For Gas Turbine Engine
Performance Simulations
A. Alexiou, E.H. Baalbergen,O. Kogenhop, K. Mathioudakis, P. Arendsen
http://www.vivaceproject.com
http://www.vivaceproject.com/http://www.vivaceproject.com/7/31/2019 GT2007-27086-pre
2/34
2Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Implementing Component Zooming and Distributed Simulations
in PRopulsion Object-Oriented SImulation Software
PROOSIS
Paper Objective & Definitions
Component Zooming: execution of higher order analysis code and
integration of its results back in the 0-D engine cycle
Distributed Simulations: technologies that enable a simulation
program to execute on a computing system containing multipleprocessors interconnected by a communication network
7/31/2019 GT2007-27086-pre
3/34
3Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
q PROOSIS OVERVIEW
q Compressor Stage-Stacking
q The Engine Model
q COMPONENT ZOOMING
o The de-coupled Approach
o The semi-coupled Approach
o The fully-coupled Approach
q DISTRIBUTED SIMULATIONS
o Implementing Distributed Simulations
o Prototype Development
o Future Developments
q SUMMARY & CONCLUSIONS
Contents
7/31/2019 GT2007-27086-pre
4/34
4Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
PROOSIS OVERVIEW: Code View
Component
Libraries
Library
EL Files
Output Window
Component EL
object-oriented
code containing
mathematical
description of realcomponent
7/31/2019 GT2007-27086-pre
5/34
5Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
PROOSIS OVERVIEW: Schematic View
Library
Palette
Drag-and-drop icons
from palette to
construct engine
model.
Connect
components through
appropriate
communication ports
7/31/2019 GT2007-27086-pre
6/34
6Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
PROOSIS OVERVIEW: Simulation View
Define
EngineMathematical
Model and
Simulation
Cases
Graphical
representation of
results
Describe in EL
calculation
mode or type
7/31/2019 GT2007-27086-pre
7/34
7Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
q PROOSIS OVERVIEW
q Compressor Stage-Stacking
q The Engine Model
q COMPONENT ZOOMING
o The de-coupled Approach
o The semi-coupled Approach
o The fully-coupled Approach
q DISTRIBUTED SIMULATIONS
o Implementing Distributed Simulations
o Prototype Development
o Future Developments
q SUMMARY & CONCLUSIONS
Contents
7/31/2019 GT2007-27086-pre
8/34
8Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Compressor Stage-Stacking: Methodology
Calculation of
individual stage exit properties from
dimensionless stage characteristics
and geometry data
stack stages together to evaluate
overall compressor performance
0
2
4
6
8
10
12
14
PressureRatio
10 15 20 25 30
Corrected Mass Flow
IsentropicEfficiencyContours
7000
7500
8000
8500
9000
9500
10
000
10500 1
1085
11750
0 .86
0.85
0.84
0.82
0.80
0.75
0.70
7/31/2019 GT2007-27086-pre
9/34
9Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Compressor Stage-Stacking: Implementation
FORTRAN: SUBROUTINE stageStack (arguments)
compiled as static library (.lib)PROOSIS: FORTRAN FUNCTION stageStack (arguments)
IN stageStack.lib
OR
C++ wrapper for FORTRAN subroutine:
extern "C" void __stdcall STAGESTACK (arguments);void stageStackClass::stageStack(arguments)
{STAGESTACK (arguments); }
PROOSIS: EXTERN CLASS stageStackClass
METHODS
EXTERN METHOD stageStack (arguments)
END CLASS
INCLUDE stageStack.h IN stageStack.lib
7/31/2019 GT2007-27086-pre
10/34
10Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
q PROOSIS OVERVIEW
q Compressor Stage-Stacking
q The Engine Model
q COMPONENT ZOOMING
o The de-coupled Approach
o The semi-coupled Approach
o The fully-coupled Approach
q DISTRIBUTED SIMULATIONS
o Implementing Distributed Simulations
o Prototype Development
o Future Developments
q SUMMARY & CONCLUSIONS
Contents
7/31/2019 GT2007-27086-pre
11/34
11Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Engine Model
Single-Shaft Industrial Gas Turbine Engine
with 15-stage axial compressor
7/31/2019 GT2007-27086-pre
12/34
12Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Engine Model
Single-Shaft Industrial
Gas Turbine Engine
with 15-stage axial compressor
0
2
4
6
8
10
12
14
PressureRatio
10 15 20 25 30
Corrected Mass Flow
Isentropic EfficiencyContours
7000
7500
8000
8500
9000
9500
1000
0
10500
11085
11750
0.86
0.85
0.84
0.82
0.80
0 .75
0.70
7/31/2019 GT2007-27086-pre
13/34
13Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Component Zooming: Compressor 1-D
GasInGasOut
GasTurbo
AbsCompressor
BETA map
Compressor#SasP
MFT map
Abstract component containing general interface
& equations for standard library components
Abstract component containing general interface
& equations for standard library turbo-components
Abstract component containing core
compressor calculations
AbsCompressor1D
Compressor1D#SasP
Abstract component calling
stage-stacking functionUser-specified
No of bleed ports
User-specified
No of bleed ports
Compressor 1-D Inheritance Tree
7/31/2019 GT2007-27086-pre
14/34
14Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Component Zooming: De-coupled Approach (DCA)
1D
Mass Flow
Pressur
eRatio
Design
Point
Run multi-point
steady-state
simulation
Instantiate
Compressor 1-D
Produce PROOSIS compatible
compressor 1-D map file
7/31/2019 GT2007-27086-pre
15/34
15Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Component Zooming: De-coupled Approach (DCA)
Mass Flow
PressureRa
tio
Design
Point
Run engine simulation using
compressor 1-D map
Select this map
in compressor
Editor
7/31/2019 GT2007-27086-pre
16/34
16Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Component Zooming: Results (DCA)
Mass Flow
PressureRatio
0-D
1-D
Design
Point
Mass Flow
IsentropicEfficiency
0-D1-D
-1.5
-1.1
-0.7
-0.3
0.1
0.5
Load
%
differenceinHeatRateEffect
of
Compressor Zooming
on
Heat Rate vs Load
Characteristic
Comparisonof
0-D & 1-D
Maps
7/31/2019 GT2007-27086-pre
17/34
17Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Component Zooming: Semi-coupled Approach (SCA)
1. Corrected flow zooming scalar2. Isentropic efficiency zooming scalar
3. Fuel flow rate
1. Mass flow error2. Isentropic efficiency error3. Rotational speed
IntrinsicNewton-Raphson
Function
Run Engine Model
All 0-D components
Call External
Compressor Stage-Stacking
Function
Get Compressor 0-D
rotational speed, pressure ratio
Inlet temperature & pressure
Get Compressor 1-D
mass flow and efficiency
Stage geometry
& characteristics
Specified load &
Rotational speed
Independent
Variables
Objective
Variables
PROOSIS
EXPERIMENT
SCHEME
7/31/2019 GT2007-27086-pre
18/34
18Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
The meaning of Modification factorsThe meaning of Modification factors
Transformation of component performance maps
PressureRatio
Corrected Mass Flow
q0
q
0q
qf
q=
Component Zooming: Semi-coupled Approach (SCA)
7/31/2019 GT2007-27086-pre
19/34
19Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Structure of Adaptive modelsStructure of Adaptive models
krefp
kp
kx
x
f,,
,=
Modification factors fkfor components
kp
x,
krefpx ,,
: Actual value for parameter
: Reference value for parameter
Transformation of component performance maps
Component Zooming: Semi-coupled Approach (SCA)
7/31/2019 GT2007-27086-pre
20/34
20Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
0.988
0.992
0.996
1
1.004
1.008
1.012
Load
Scalar
Corrected Mass Flow
Isentropic Efficiency
Component Zooming: Results (SCA)
Variation of Zooming Scalars with Load
7/31/2019 GT2007-27086-pre
21/34
21Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Component Zooming: Fully-coupled Approach (FCA)
TYP MAP
Pt
Cmp1D
Replace Compressor 0-D with 1-D one
7/31/2019 GT2007-27086-pre
22/34
22Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Component Zooming: Results (FCA)
0.583Compressor Power
-0.238Compressor Polytropic Efficiency
0.211Compressor Pressure Ratio
0.438Compressor Delivery Temperature
0.111Compressor Inlet Flow
0.289Fuel Flow Rate
% DIFFERENCEPARAMETER
Design Point Case
1.5% inter-stage bleed from 10th stage
7/31/2019 GT2007-27086-pre
23/34
23Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
q PROOSIS OVERVIEW
q Compressor Stage-Stacking
q The Engine Model
q COMPONENT ZOOMING
o The de-coupled Approach
o The semi-coupled Approach
o The fully-coupled Approach
q DISTRIBUTED SIMULATIONS
o Implementing Distributed Simulations
o Prototype Development
o Future Developments
q SUMMARY & CONCLUSIONS
Contents
7/31/2019 GT2007-27086-pre
24/34
24Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Distributed Simulations: Rationale & Features
q Collaborative modelling among possibly
geographically dispersed engineers
q Easy and efficient deployment of subsystem models
q Protection of ownership and IPR
qReduction of simulation time through load distribution
q Size and complexity of the simulation model may grow
irrespective of capability of computing infrastructure
q Reuse of submodels in different simulations
7/31/2019 GT2007-27086-pre
25/34
25Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Distributed Simulations: Implementing DS
Technologies considered:
CORBA: complex; did not catch up with growing Web developmentsand demands; high run-time costs; difficulties with security; versions &
difficulties in backward compatibility; not supported by Microsoft...
DCOM: serious security problems; did not catch up with growing Web
developments and demands; deprecated in favour of .NET
Java RMI: Java specific; being obscured by Web Service technology
XML and SOAP: slower than e.g. CORBA and RMI but providing good
basis for secure distributed web-based solutions in wide-area contexts
Web Services: uses open standards and protocols (incl. SOAP and
XML); commonly used nowadays to implement secure distributed
solutions in SOA style; standards and tools are emerging
Web Service: state-of-the-art technology enabling software components(clients, servers) to communicate over a network using standard
messages and formats
7/31/2019 GT2007-27086-pre
26/34
26Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Distributed Simulations: Prototype Development (I)
Prototype in VIVACE context: PROOSIS with compressor
stage stacking function, available as User Library,running on a remote computer
Compressor stage stacking function:
developed by, and proprietary code of NTUA
written in Fortran, available as a shared library (DLL) on
Windows
available to PROOSIS users at NTUA as a PROOSIS
User Library (PROOSIS mechanism to include customer
code in engine simulations) code may be used but cannot be installed outside NTUA
7/31/2019 GT2007-27086-pre
27/34
27Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Distributed Simulations: Prototype Development (II)
Compressor Zooming via
Remote Web Service
Invocation between
NLR & NTUA
PROOSIS simulation inThe Netherlands
Stage Stacking Calculation in
Greece
7/31/2019 GT2007-27086-pre
28/34
28Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Distributed Simulations: Prototype Development (III)
Inte
rnet
Application Server
PROOSIS
Web Service
Operation
Stage-Stacking
Function
Server (NTUA)
PROOSIS
Client (NLR)
PROOSIS
Web
Component
Enables remote use
and sharing of User
Libraries over theInternet without
distributing the code
of the libraries
7/31/2019 GT2007-27086-pre
29/34
29Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Distributed Simulations: Prototype Development (IV)
SOAP overinternet
SOAP over internet
3
PROOSIS WebService
Component
(Java)JNI
IntermediateCode(C++)
Stage-StackingFunction
(FORTRAN)
4
5
PROOSISCustom Library
(C++)
Web ServiceClient(Java)
JNI
1
2
Client Side (NLR) Server Side (NTUA)
Layered Structure of Prototype
7/31/2019 GT2007-27086-pre
30/34
30Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Distributed Simulations: Live Public Demo
NLR
NTUA
7/31/2019 GT2007-27086-pre
31/34
31Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Distributed Simulations: Future Developments
Design of more generic (re-usable) interface
multiple function implementationsnot all layers need to be modified
additional overhead and delays in communication
Reduction of overhead caused by conversions & data transfers
use pure C++ development environment
C++ support for Web Services limited/unstable
Java platform allows integration with other collaborative tools
Reduction of overhead in DLL loading and unloading
load DLLs once and dispose after final calculation
Multi-user and security
Use Web Service Security specification
allow multi-user access
7/31/2019 GT2007-27086-pre
32/34
32Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
q PROOSIS OVERVIEW
q Compressor Stage-Stackingq The Engine Model
q COMPONENT ZOOMING
o The de-coupled Approach
o The semi-coupled Approach
o The fully-coupled Approach
q DISTRIBUTED SIMULATIONS
o Implementing Distributed Simulations
o Prototype Development
o Future Developments
q SUMMARY & CONCLUSIONS
Contents
7/31/2019 GT2007-27086-pre
33/34
33Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Summary & Conclusion (I)
PROOSIS is a standalone, multi-platform, object-oriented
simulation environment for gas turbine engine performancesimulations. It can be used to create, run, manage and share
engine models using either the standard or custom libraries
of engine components. The feasibility of performing multi-
fidelity and distributed simulations with PROOSIS was
demonstrated in this paper.
Using the model of an industrial gas turbine engine and a 1-D
compressor stage stacking code as an example, different
implementations for integrating high fidelity component
analysis in overall engine simulations were presented. The
tools flexible and extensible architecture gives the user thefreedom to select the most suitable approach for a particular
simulation case.
7/31/2019 GT2007-27086-pre
34/34
34Advanced Capabilities For Gas Turbine Engine Performance Simulations
Alexiou, Baalbergen, Kogenhop, Mathioudakis, Arendsen
Summary & Conclusion (II)
The stage stacking code is also used to demonstrate
distributed simulations. A prototype of a Web Component has
been created and successfully tested that remotely invokes
the code from an engine simulation, via the internet, using
Web Services technology.
These demonstrations prove that the tools architecture isadaptable enough to integrate different modelling methods
and its potential to fulfil its role as a shared simulation
environment.