Benchmarking CEASIOM Software to Predict Flight Control and Flying Qualities of the B-747

Preview:

DESCRIPTION

Benchmarking CEASIOM Software to Predict Flight Control and Flying Qualities of the B-747 A. Da Ronch The University of Liverpool, UK C. McFarlane, C. Beaverstock Bristol University, UK J. Oppelstrup, M. Zhang, A. Rizzi Royal Institute of Technology, Sweden. Introduction - PowerPoint PPT Presentation

Citation preview

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

1

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

2

Benchmarking CEASIOM Software to Predict Flight Control and Flying

Qualities of the B-747

A. Da RonchThe University of Liverpool, UK

C. McFarlane, C. BeaverstockBristol University, UK

J. Oppelstrup, M. Zhang, A. RizziRoyal Institute of Technology, Sweden

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

3

Introduction

• Contemporary aircraft conceptual designo Handbook methods, semi-empirical theoryo Need to recalibrate these empirical methods

• Augmented-stability & extended flight envelopeo More accurate flight dynamics modellingo Computational methods based on first principleo First-Time-Right with the FCS design architecture

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

4

CEASIOMComputerized Environment for Aircraft Synthesis and

Integrated Optimization Methodso SimSAC project under the European Commission 6th

Framework Programmeo Integrates discipline-specific tools for conceptual design

to predict flying & handling qualities

http:/www.simsacdesign.eu

http:/www.ceasiom.com

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

5

Objectives

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

6

CEASIOM main GUI

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

7

Test Case: Boeing 747

• Large 4-engined turbofan; 350+ pax

• Multiple control surfaces: Krueger LE flaps, triple-slotted TE

flaps

• Flight dynamics with FCSDT to evaluate different fidelity-

level approx

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

8

Adaptive Fidelity CFD1. DATCOMo Semi-empirical

2. TORNADOo Vortex-Lattice method

3. EDGEo CFD solver

Fidelity CPU Time Geometry

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

9

Adaptive Fidelity CFD1. DATCOMo Semi-empirical

2. TORNADOo Vortex-Lattice method

3. EDGEo CFD solver

• For conventional aircraft, estimate aero derivatives based on geometry details and flight conditions

• Suspect results for new configuration

• Handbook

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

10

Adaptive Fidelity CFD1. DATCOMo Semi-empirical

2. TORNADOo Vortex-Lattice method

3. EDGEo CFD solver

http:/www.redhammer.se/tornado/

• Modified horse-shoe vortex singularity method

• Steady & low reduced-freq harmonic unsteady flows

• Prandtl-Glauert similarity role for compressibility

• Fuselage can be modelled

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

11

Adaptive Fidelity CFD1. DATCOMo Semi-empirical

2. TORNADOo Vortex-Lattice method

3. EDGEo CFD solver

• 3D NS/Euler, compressible flow solver from FOI, Sweden

• Unstructured grids with arbitrary elements; node-centred FV

• Explicit Runge-Kutta integration to steady state

•Semi-implicit, dual-time method for unsteady problem

• Acceleration techniques, turbulence models, parallel implementation

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

12

CFD Code - EDGEDeflection of control surfaces

1. Generation of a new grid for every new configuration of

deflected control surfaces

clean geometry

tens of grids needed

2. Transpiration BCs

only one single grid needed

limits on min/max deflection

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

13

Challenges

1.How to automate grid generation for CFD?

2.How to do 100k CFD?

3.How to do S&C analysis early in design phase?

“...whether CFD can participate in the design process with sufficient speed to drive down the design cycle time”, Dawes et al.

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

14

Challenges

1.How to automate grid generation for CFD?

2.How to do 100k CFD?

3.How to do S&C analysis early in design phase?

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

15

From Geometry to CFD Grid (1)AcBuilder: sketch-pad

- Edit XML file to match new design

- Visual interpretation

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

16

From Geometry to CFD Grid (2)

SUMO* (SUrface MOdeler)• Rapid generation of 3D water-tight geometry

• Automated generation of unstructured surface mesh

• Triangulation based on in-sphere criterion, better than

Delaunay, for skewed surfaces

• Volume mesh using TetGen

* http:/www.larosterna.com/dwfs.html

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

17

From Geometry to CFD Grid (2)

SUMO surface grid

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

18

From Geometry to CFD Grid (2)

SUMO volume grid

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

19

From Geometry to CFD Grid

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

20

TORNADO Geometry

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

21

TORNADO Geometry

Munk’s theory

Sink/source distribution

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

22

TORNADO Geometry

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

23

Challenges

1.How to automate grid generation for CFD?

2.How to do 100k CFD?

3.How to do S&C analysis early in design phase?

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

24

Flight Variables CoefficientsAoA Mach Beta Elev Rud Ail ... p q r CL CD Cm CY Cl Cn

- - - - - -

- - - - - -

- - - - - -

- - - - - -

- - - - - -

- - - - - -

- - - - - -

- - - - - -

Aerodynamic Table Format

Non-conventional controls

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

25

Brute Force Approach

• Simple example; let’s assume:o 10 values for AoA, Mach, Beta, Elev, Rud, Ailo More than 100k entries needed in tableo 10 seconds each calculation using TORNADO

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

26

Brute Force Approach

• Simple example; let’s assume:o 10 values for AoA, Mach, Beta, Elev, Rud, Ailo More than 100k entries needed in tableo 10 seconds each calculation using TORNADO

106 / (24 * 60 * 60) > 10 days

Brute force approach not feasible to fill-in aero

tables!

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

27

Sampling & Data Fusion

Journal of Aircraft, 46 (3), 2009

Aerodynamic Tables

Flight Dynamics

Kriging

Existing

Table

Increments to

Design

Sampling

Database

New Design

Data Fusion for Aerodynamic Increments

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

28

Sampling & Data Fusion

• STATIC effects:o Sampling for M-α-β dependence o Co-Kriging to calculate increments (controls)

• DYNAMIC effects:o No frequency dependenceo Alpha dependence onlyo Replace unsteady time-accurate with HB method? *o Stability derivatives from DATCOM

* AIAA Journal, 47 (4), 2009

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

29

Challenges

1.How to automate grid generation for CFD?

2.How to do 100k CFD?

3.How to do S&C analysis early in design phase?

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

30

FCSDT

FCSDT (Flight Control System Design Toolkit)• Design of the FCS, FCS architecture design

• Reliability analysis, failure mode analysis

• Control allocation, response simulation

• S&C analysis, HQ assessment, control laws design,

control laws definition, flight simulation

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

31

Aerodynamic Predictions1.Low speed aerodynamics

2.Transonic regime

• DATCOM

• TORNADO

• TORNADO with compressibility correction

• EDGE in Euler mode

More comparisons in the paper; exp data from

Rodney, C.H., Nordwall, D.R., 1970

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

32

CL vs α, Mach = 0.80

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

33

CD vs CL, Mach = 0.80

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

34

Cm vs α, Mach = 0.80

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

35

Mach = 0.80

AoA = 1.0 deg

Positive elev deflection

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

36

Results

1.Cruise condition

• Trim & Stability analysis

• Eigen-structure assignment for feedback

controller A + B *K• Flight Handling Qualities

2.Failed lower rudder segment

• Trim & Stability analysis

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

37

Trimmed AoA

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

38

Trimmed elevator

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

39

Pole plot, Mach = 0.8

Short Period

Dutch-Roll

Phugoid

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

40

Results

1.Cruise condition

• Trim & Stability analysis

• Eigen-structure assignment for feedback

controller A + B *K• Flight Handling Qualities

2.Failed lower rudder segment

• Trim & Stability analysis

Eigen value: -2 ± i *2 for Short Period mode

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

41

Kα: gain value of feedback AoA to elevator

Kq: gain value of feedback pitch rate to elevator

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

42

Results

1.Cruise condition

• Trim & Stability analysis

• Eigen-structure assignment for feedback

controller A + B *K• Flight Handling Qualities

2.Failed lower rudder segment

• Trim & Stability analysis

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

43

Short Period mode

Eigenvalue: ƞ + i *ω T1/2 = ln(2) / |ƞ|

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

44

Phugoid mode

ξ: damping ratio ωn: undamped circular freq

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

45

Dutch Roll mode

ξ: damping ratio ωn: undamped circular freq

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

46

Results

1.Cruise condition

• Trim & Stability analysis

• Eigen-structure assignment for feedback

controller A + B *K• Flight Handling Qualities

2.Failed lower rudder segment

• Trim & Stability analysis

Lower rudder segment failed at -10o for range of Mach numbers

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

47

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

48

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

49

Conclusions• Aero tables for flight mechanics

o Automated generation of CFD grido From low-fidelity methods to CFDo Multiple control surfaceso Smart procedure to fuse data

• Test case: Boeing 747, trim analysis & poles ploto Cruise conditiono Failure analysis: lower rudder segment jammed

• Demonstratedo Robust process for S&C analysis in early designo CFD needed for good prediction for a realistic test case

ICAS Paper no. 282Nice, September 2010A.Da-Ronch@liverpool.ac.uk

50

Future Works

• Flight manoeuvre replayo Aero table with dynamic derivatives from HBo Replay with CFDo When does prediction fail? * Unsteady effects?

• Need to review model for flight mechanicso System IDo Indicial (successfully used in gust analyses)o State Space

• Towards modelling of unsteady effects

* AIAA-2009-6273

Recommended