Copyright © 2009 Boeing. All rights reserved. The Impact of High Performance Computing and Computational Fluid Dynamics on Aircraft Development Edward.

Copyright © 2009 Boeing. All rights reserved.

The Impact of High Performance Computing and Computational

Fluid Dynamics on Aircraft Development

Edward N. TinocoTechnical Fellow Enabling Technology & Research Airplane Configuration, Integration & Performance Boeing Commercial Airplanes


http://www.casc.org/index.html


Computational Fluid Dynamics (CFD)

Tools for Aerodynamic Development of Aircraft Configurations

Flight Test Wind Tunnel



Direct Numerical Simulation

Large Eddy Simulation

Detached Eddy Simulation w/RANS

Reynolds Averaged Navier-Stokes

Aerodynamic flows are characterized as compressible, viscous (high Reynolds number turbulent) flows.

Computational Fluid Dynamics

Practical Limit for Complete Airplane

Applications

Incr

easi

ng Com

putatio

nal C

omple

xity

,

Decre

asin

g Em

pirici

sm

Euler

Full PotentialWith Coupled Boundary Layer

Linear Potential (Panel Methods) 1960’s

1970-80’s

1980’s

1990’s

2000’s

2045

2080



1985 1990 1995 2000 2005 2010

777 737NG 787

21% thicker faster wingthan 757, 767 technology.Best economics in class

Highly constrained wingdesign. Faster wing than737-300. Highest sellingcommercial airplane ever

Multipointoptimization

design

Faster and more effi-cient than previousmedium size aircraft

747-8

lowest operatingcosts and besteconomics of anylarge airplane

1 GFLOP 100 GFLOP10 GFLOP

UnstructuredAdaptive Grid

3-D N-S

Full-PotentialTransonic Design

Wing-BodyReynolds Averaged

Navier-Stokes

General 3-DReynolds Averaged

Navier-Stokes

General 3-DFull-Potential

Transonic Analysis

737-300

Modern close couplednacelle installation, 0.02Mach faster than 737-200Enabled by CFD

767 757

1980 state of the art

19851965 1970 1975 19801960

h

xc

Joint CFD/Wind TunnelStudies unlock the secretof nacelle/wing interfer-

ence drag

SupersonicTransport

2-D AirfoilDevelopment

2-D LinearPotential

LinearizedSupersonic

General 3-DLinear Potential

Wing-BodyFull-Potential

Transonic Analysis

1 MFLOP 10 MFLOP 100 MFLOP

Timeline of the Use of Computational Fluid Dynamics in Aircraft Development




What is the Measure of Value in Computational Fluid Dynamics?

• The value of reduced wind tunnel testing due to the use of CFD

In the past 20 years the use of CFD has provided significant cost savings

767(1980)

777-200(1990)

787(2005)

-25 %

-30 %

767(1980)

777-200(1990)

787(2005)

-25 %

-30 %

Win

d T

un

nel

Ho

urs

Win

d T

un

nel

Ho

urs

YEAR

WRIGHT FLYER

DC-3

B-17

B-29 DC-6

B-47

B-52DC-8

F-111

707

737

747

B-1

F-15

SHUTTLE

106

105

104

103

102

10 1900 1920 1940 1960 1980 2000

CFDImpact

YEAR

WRIGHT FLYER

DC-3

B-17

B-29 DC-6

B-47

B-52DC-8

F-111

707

737

747

B-1

F-15

SHUTTLE

106

105

104

103

102

10 1900 1920 1940 1960 1980 2000

CFDImpact



The Challenge

• One complete airplane development requires about 50,000 to 100,000 aerodynamic simulations.

• Flight test is used to validate and certify that the aircraft is safe over the entire range flight conditions mandated by law.

• The challenge is to further push the use of CFD into the edges of the flight envelope.

Velocity - VEAS

The Flight Envelope

Higher quality data earlier in the design phase for Multidisciplinary Design Optimization – big driver on reducing cost “Good enough” aerodynamic data base to reduce number of design cycles Higher quality full scale flight simulation – avoid costly surprises in flight test



What is the Measure of Value in Computational Fluid Dynamics?

• The value of reduced wind tunnel testing due to the use of CFD

In the past 20 years the use of CFD has provided significant cost savings. This is a small fraction of the value CFD delivered.

A much greater value of CFD in the Commercial arena is………..

• The added value of the product due to the use of CFD Achieving design solutions that are otherwise unreachable. Shortening the design development process. Getting it right the first time. NOT getting it right the first time results in:

Very lengthy and costly development to fix it Possible cancelation/termination of the program Putting the Company at risk





Extra Material



Boeing Puget SoundHPC Environment

2001• Cray T916• SGI Origin• ~0.100 Tflops• Full Potential + BL

– e.g. Tranair

2009• Cray X1• PC clusters• ~50 Tflops• Navier-Stokes

– e.g. CFD++, CFL3D, OVERFLOW



CFD Contributions to 787

High-Speed WingDesign Cab Design

Engine/Airframe Integration

Inlet DesignInlet Certification

Exhaust-System Design

CabinNoise

Wing-BodyFairing Design

Vertical Tail and Aft Body Design

Design ForStability &

Control

High-Lift Wing Design

APU InletAnd Ducting

ECS Inlet DesignAPU and Propulsion

Fire Suppression

Nacelle Design

• Thrust-Reverser Design

• Community Noise

Design for FODPrevention

Aeroelastics

Icing

Air-Data System

Location

Vortex Generators

Planform Design

Buffet Boundary

Reynolds-Number Corrections

Flutter

Control-Surface Failure Analysis

Wind-Tunnel Design Validation

Wing-Tip DesignWing

Controls

Avionics Cooling

Interior Air

Quality

Engine-Bay Thermal Analysis



Cost and Flowtime Characteristics of Wind Tunnels and CFD

One complete airplane development requires about 50,000 toOne complete airplane development requires about 50,000 to100,000 aerodynamic simulations100,000 aerodynamic simulations

Today Desired Future State

10 100 1,000 10,000 100,000

Cost,Flowtime

Number of Simulations10 100 1,000 10,000 100,000

Number of Simulations

CFD – Design, MostData Base Building

Wind Tunnel – Validation

Special Conditions

CFD

Wind Tunnel

Data BaseBuilding

The use of new CFD is driven by desperation.Desperation to remain competitive!



Closing Thoughts

• CFD exists to enable new solutions to problems, reduce airplane development cost, and reduce time to market

• CFD can allow you to safely explore areas of the flight regime without putting a pilot at risk

• CFD can allow you to analyze conditions for which physical simulation is either very expensive or not possible, such as hypersonic propulsion systems and full flight Reynolds number testing

• Accuracy, robustness and timeliness are the keys to acceptance and use in an industrial environment

• Impediments: applications that do not scale well (to 1000’s of processors with sufficient memory) – this is science; resources to run 1000s of flight conditions on 100’s of processors – this is the business of engineering


Copyright © 2009 Boeing. All rights reserved. The Impact of High Performance Computing and Computational Fluid Dynamics on Aircraft Development Edward.

Documents