new methods for loads & aeroelastics - upm · upstream and downstream codes e.g. usnewpan,...
TRANSCRIPT
1
© Copyright QinetiQ Limited 2009
QinetiQ Proprietary
QINETIQ/09/01276
New Methods for Loads & Aeroelastics
Dr Ian Roberts & Tom WallaceA presentation to: KATNET 2009, Bremen
[email protected] May 2009
QinetiQ Proprietary
2
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
1 Vision & Background
The main goal of the Flexible Wing Programme is:
Closely integrate aeroelastics with its neighbouring disciplines
Motivation:
• Shorten the aircraft design process by more closely integrating the disciplines
• Develop a common framework under which analyses could be performed
• Address the issues presented by highly flexible and HALE aircraft
• Introduce novel methods to rapidly analyse non-linear systems
QinetiQ Proprietary
3
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
1 Vision & Background
The current aircraft design process comprises of a number of silos between which information is transferred.
The challenge is to integrate these silos such that the data flow is continuous.
The aeroelastics & loads domain interfaces with most disciplines, hence the choice to develop a common framework around aeroelastics.
In order to address these issues we need to consider which components of the design space have to be automated and integrated.
The identified core components are therefore:
• Flow physics modelling (atmospheric, aerodynamic, high-lift, engines)
• Structural modelling (structural mass, stiffness and damping)
• Systems modelling (mass, stiffness incl. actuators etc.)
• Control system modelling (loop gains as a function of aircraft position/mode)
• Non-structural/systems (furniture (mass, stiffness), passengers, freight)
QinetiQ Proprietary
4
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
1 Vision & Background
The envisaged capability of the framework will allow the following tasks to be performed:
• Single domain analysis e.g. aeroelastics, flight dynamics etc.
• Flexible-rigid coupling
• Aircraft performance calculations incorporating aeroelastic distortion
• Integration with control system design tools e.g. Matlab-Simulink
• Direct extraction of stress-strain (& their rates) in addition to loads and moments
QinetiQ Proprietary
5
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
1 Vision & Background
The CFMS Core ProgrammeThe CFMS Core Programme
Vision
Transform design processes for aerospace, marine & automotive industries
Goals
• Develop highly complex & advanced mathematical models within computer-based systems & processes to solve problems for the future design of aircraft, ships, motorsport and beyond
• Revolutionary change in the speed of the design process
• Replace the requirement for expensive windtunnel test campaigns
• Reduce the environmental impact of the industries involved by improving the end product through more efficient and flexible design capabilities
QinetiQ Proprietary
6
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
1 Vision & BackgroundFlexible Aircraft Programme within the CFMS core programme
• QinetiQ’s activities draw on the knowledge and technology within the various work packages related to mesh movement, parallel efficiency, data storage, Reduced Order Modelling (ROM) and parametric modelling.
• Within QinetiQ’s analysis framework the contributing technologies are integrated to deliver a step change in fidelity & efficiency in the field of loads and aeroelastics.
• The QinetiQ framework will enable CFMS Core Programme partners to integrate and improve their processes by taking aeroelastics in to account more routinely with a minimal penalty on their execution time.
This work was undertaken as part of the wider CFMS Framework, aimed at delivering a paradigm shift in the capability of fluid mechanics simulation systems. This Framework has been established to manage a sustained programme of research projects, both private venture and government supported. More details can be found on www.cfms.org.uk.
QinetiQ Proprietary
7
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
Control Systems
Non-linear Structure
Vehicle Definition
2 First Steps – Framework
CFD FE/Mass
Modes:Rigid
Flexible Control
Unsteady CFD
Unsteady Aero
Linear UAM
Control UAM
Non-linear UAM
Non-linear Control UAM
Time/Freq Domain Analysis
Analysis
Output
Data
QinetiQ Proprietary
8
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
2 Framework Tools
Analysis
OutputCatia v5
Newpan Nastran
Nastran
USNEWPAN
Spline
MatlabOctave
MatlabOctave
MatlabOctave
Nastran or User
MatlabOctave
MatlabOctave
MatlabOctave
Data
QinetiQ Proprietary
9
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
Control Systems
Non-linear Structure
CAD
2 Future Enhancements
CFD FE/Mass
Modes:Rigid
Flexible Control
ROM
Unsteady Aero
Linear UAM
Control UAM
Non-linear UAM
Non-linear Control UAM
Numerical Continuation
Time/Freq Domain Analysis
Output
Optimiser
Analysis
Data
QinetiQ Proprietary
10
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
3 Data Generation Process – CAD � CFD/FE
CAD geometry should be compatible with CFD and FE based tools and coincident e.g. QinetiQ’s MDO capability MDCAD
QinetiQ Proprietary
11
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
3 Data Generation Process – Mode Extraction
Extract mode shapes and associated stiffness/mass from the structural models including:
• Elastic modes
• Rigid modes
• Control mode
QinetiQ Proprietary
12
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
3 Data Generation Process – Unsteady Aero
Flow Solution’s USNEWPAN is used to generate unsteady aerodynamic data or direct coupling of Euler/RANS for performance is via spline utility
QinetiQ Proprietary
13
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
3 Data Generation Process – RAA
Using the data from the unsteady aerodynamics, the Rational Aerodynamic Approximation (RAA) method is applied to build a Reduced Order Model (ROM) of the system.
This ROM provides force on mode mdue to oscillation in mode n e.g. lift force due to pitch
Gust modes trialled to capture gust penetration effects
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
-0.5 0 0.5 1 1.5 2
Real Forcing
Imag
inar
y Fo
rcin
g RAA USNEWPAN results
-25
-20
-15
-10
-5
0
-11 -10 -9 -8 -7 -6 -5 -4
Real Forcing
Imag
inar
y F
orci
ng
RAA USNEWPAN results
QinetiQ Proprietary
14
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
4 Initial Results & Validation
Initial testing has verified the linear single domain components for:
− Aerostructural coupling (VC-Euler FE)
− Aeroelastics
− Flight mechanics (including control interaction)
Aerostructural coupling (MDO testcase)
QinetiQ Proprietary
15
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
4 Initial Results & Validation
Aerostructural coupling of complex geometries
Same process for panel & VC Euler (including volume mesh movement)
QinetiQ Proprietary
16
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
4 Initial Results & Validation
Aeroelastics (MDO testcase comparison)
• Following slides show results
• Black lines no symbols current method
• Coloured lines with symbols corrected MSC/Nastran
QinetiQ Proprietary
17
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
Frequency Variation Mach 0.6 Unmatched
0
1
2
3
4
5
6
7
8
100 150 200 250 300 350 400 450 500 550
Velocity m/s
Fre
quen
cy
QinetiQ Proprietary
18
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
Damping Variation Mach 0.6 Unmatched
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
100 150 200 250 300 350 400 450 500 550
Velocity m/s
Dam
ping
g
QinetiQ Proprietary
19
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
4 Initial Results & Validation
Flight Mechanics validation (HALE vehicle testcase)
• Control derivatives
• Rigid body derivatives
QinetiQ Proprietary
20
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
4 Initial Results & Validation
Further research has looked in to the application of Numerical Continuation methods for aeroelastic systems
• Floquet Multipliers to assess LCO stability analogous to Eigenvalues
• Uses predictor-corrector technique to tracks stability branches
QinetiQ Proprietary
21
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
5 Continuing DevelopmentsResearch over the next year:
• Direct extraction of stress/strain information – as the process uses full FE models, stress/strain data is available for the mode shapes which can be reconstructed during time simulations giving strain and strain rates to assess fatigue properties.
• High fidelity aerodynamic correction – use alternative aerodynamic data sources to provide mean flow data and apply around this point (in conjunction with Flow Solutions).
• Direct Aerodynamic Influence Coefficients (AIC) Flow Solutions’ Newpan could provide full AIC matrices therefore removes requirement for modes shapes & gust modes.
• Enhanced interpolation of RAA models to cover non-linear aerodynamic regimes in conjunction with high-fidelity aerodynamic correction.
• Application to rotorcraft stability
QinetiQ Proprietary
22
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
6 Future Applications
This programmes is intended to be a stepping stone towards the routine application of higher-fidelity aerodynamics & improving the analysis timeline.Currently the major challenge is related to reducing the loads process as it depends on a lot of high-fidelity aerodynamics (up to a high angle-of-attack). This programme and associated work will provide some important steps along the way:• Routine use of complex 3-D aerodynamic geometry within the process• Direct link to FE model, therefore access to stress/strain data not just oads/
moments• Move away from full AIC’s as processes may not be practical in the future with
high-fidelity aerodynamics, continuing research in to assumed modes• Remove requirement for flexibilisation step by using aeroelastic modes in
generating aerodynamic data for the loads process (high fidelity)• Research in to the process of loads design space mapping (RSMs or surrogate
models) to reduce the number of calculations required at early stages of aircraft development programmes
• More regular use of full non-linear modes
QinetiQ Proprietary
23
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
7 Conclusions
Progress to date• Method developed and validated against
linearised test cases• Demonstrated for aeroelastics, flight
mechanics and elastic/rigid coupling
• Integrated with Matlab-Simulink to allow control systems to be integrated
Benefits over commercial equivalents• Only reliant on one commercial code,
Flow Solutions’ Unsteady Newpan
• Ability to integrate with control systems• Directly extract stress/strain & rates
• Reconfigurable to user requirements
• Rapid generation of data• Non-linear domain functionality
Continuing developments• Application to rotorcraft stability
• Steady-state correction based on alternative aerodynamics
• Develop full AIC as opposed to mode based approach
• Interpolation between RAA models• Non-linear studies
QinetiQ Proprietary
24
© Copyright QinetiQ Limited 2009
QINETIQ/09/01276
7 Conclusions
Learning Lessons
The work has looked at how the aeroelastic process can be more closely integrated in to the design space and how the process can be adapted to cope with new technologies at a practical level related to:
• Application of high-fidelity CFD-CSM when AIC data is not available.
• Derive processes that can meet the high volume of calculations that would be required from a loads based process.
• Develop generic interfaces that are wherever possible independent of upstream and downstream codes e.g. USNEWPAN, VC-Euler etc.
• Address challenges presented by highly non-linear aircraft, e.g. numerical continuation, interpolation techniques.
• Provide a “invisible” process to interface between aerodynamics & structures
• Perform cross domain analysis: aerodynamics-flight mechanics-aeroelastics-controls-strain data within a single analysis