➨ 㻞 ᅇ 㻱㻲㻰㻛㻯㻲㻰 ⼥ྜ䝽䞊䜽䝅䝵䝑䝥㻌

➨ 㻞 ᅇ 㻱㻲㻰㻛㻯㻲㻰 ⼥ྜ䝽䞊䜽䝅䝵䝑䝥㻌

㼀㼔㼑㻌㻞㼚㼐㻌㼃㼛㼞㼗㼟㼔㼛㼜㻌㼛㼚㻌㻵㼚㼠㼑㼓㼞㼍㼠㼕㼛㼚㻌㼛㼒㻌㻱㻲㻰㻌㼍㼚㼐㻌㻯㻲㻰㻌

㛤ദ᪥㻌 㻌 䠖ᖹᡂ 㻞㻝 ᖺ 㻞 ᭶ 㻞㻟㻙㻞㻠 ᪥㻌

㛤ദሙᡤ䠖Ᏹᐂ⯟✵◊✲㛤Ⓨᶵᵓ䠄㻶㻭㼄㻭䠅㻌 ㄪᕸ⯟✵Ᏹᐂ䝉䞁䝍䞊㻌

Ᏹᐂ⯟✵◊✲㛤Ⓨᶵᵓ

Japan Aerospace Exploration Agency

2010 ᖺ 1 ᭶ January 2010

1. A System Concept for EFD/CFD Integration: JAXA Digital/Analog Hybrid Wind Tunnel...5

Ώ㎶ 㔜ဢ㸦JAXA㸧

2. Reducing Uncertainty in EFD and CFD Through Data/Model Fusion ...15 Andrew Meade㸦William Marsh Rice University, USA㸧

3. Validation and Minimizing CFD Uncertainty for Commercial Aircraft Applications

- The Integration of EFD and CFD ...43 Edward N. Tinoco㸦Boeing Commercial Airplanes , USA㸧

4.

ᬑẁ╔ࡢEFD/CFD㐃ᦠ 㹼㉸㡢㏿ᢠຊపῶࢆ㢟ᮦ࡜ࡋ࡚㹼

This is Our Usual EFD/CFD Collaboration -An Example in Supersonic Drag Reduction Study- ....69

బ᐀ ❶ᘯ㸦ྡ኱ᕤ㸧

5.

ὶయ࠿ࡽⓎ⏕ࡍࡿ㡢ࡢᩘ್ゎᯒ࡜ᐇ㦂ィ 

Numerical Computation and Experimental Measurement of Aerodynamic Noise ...75

ຍ⸨ ༓ᖾ㸦ᮾ኱⏕⏘◊㸧

6.

ᐇ✵㛫࡜᫬✵㛫࡟࠾ࡅࡿὶయྍど໬

ఀ⸨ ㈗அ㸦࠾ⲔࡢỈዪᏊ኱㸧

7.

⮬ື㌴✵ຊศ㔝࡟࠾ࡅࡿEFD࣭CFDᢏ⾡ࡢືྥ࡜㠀ᐃᖖ✵ຊࢩ࣑࣮ࣗࣞࢱࡢ㛤Ⓨ

Current Status of EFD/CFD Techniques for Road Vehicle Aerodynamics and Development of the Unsteady Aerodynamic Simulator ...115

ᆤ಴ ㄔ (໭኱ᕤ)

6.

㏲ḟࢹ࣮ࢱྠ໬㸸ࢩ࣑࣮ࣗࣞࢩࣙࣥ࡜ほ ࢹ࣮ࢱࡢࣜ࢔ࣝࢱ࢖࣒⤫⼥ྜࢆࡵࡊࡋ࡚

Sequential Data Assimilation: Online Information Fusion Platform for Simulation and Observation Data ...131

ᵽཱྀ ▱அ (⤫ィᩘ⌮◊)

9.

෌౑⏝ᆺᏱᐂ 㑏ᶵ࡟ྥࡅࡓ◊✲࡟࠾ࡅࡿEFD/CFDࡢྲྀࡾ⤌ࡳ

EFD/CFD Activities in Research for Reusable Launch Vehicles ...155

㇂ Ὀᐶ㸦஑኱ᕤ㸧

10. Discontinuous Galerkinἲࢆ⏝࠸ࡓAGARD-Bᶆ‽ᶍᆺࡢ㟼ⓗ✵ຊᙎᛶゎᯒ

Static Aeroelasticity Analyisis of AGARD-B Wind Tunnel Calibration Model Using Discontinuous Galerkin CFD Solver...183

ಖỤ ࠿࡞Ꮚ (ᮾ໭኱㝔)

13. JAXAࢹࢪࢱࣝ/࢔ࢼࣟࢢ࣭ࣁ࢖ࣈࣜࢵࢻ㢼Ὕ:

࢔ࢼࣟࢢ㢼Ὕࡢ㧗㏿໬

JAXA Digital/Analog Hybrid Wind Tunnel: Speed-Up Technique of Analog Wind Tunnel ...221

⸨⏣ ┤⾜ (JAXA)

5.

ࣃࢿࣝࢹ࢕ࢫ࢝ࢵࢩࣙࣥࠕCan EFD/CFD Integration Minimize Uncertainty?ࠖ

➨

ᅇ

⼥ྜ࣮࣡ࢡࢩࣙࢵࣉ

㛤ദ㊃ព᭩

ᚑ᮶ࠊᩘ್ὶయຊᏛ㸦Computational Fluid Dynamics, CFD㸧࡜ᐇ㦂ὶయຊᏛ㸦Experimental

ࢀࠊࡑࢀࡒࢀேᮦࡶࣜࢯ࣮ࢫࡶ⊂❧࡟ࠊ⊂⮬ࡢ❧ሙ࡛⾜ࢃࢀ࡚ࡁࡲࡋࡓࠋࡋ࠿ࡋ࡞ࡀࡽࠊ

ࡣ≀⌮⌧㇟ࢆࣔࢹࣝ໬ࡋ࡚ᩘ್ⓗ࡟ゎࢆồࡵ࡚࠸ࡿ௨ୖࠊ⤖ᯝࡢጇᙜᛶࢆᐇ㦂ࢹ࣮ࢱࢆ⏝࠸࡚

᳨ドࡍࡿᚲせࡀ࠶ࡾࠊࡑࡢព࿡࡛CFDࡣEFD࡟୍᪉ⓗ࡟౫Ꮡࡋ࡚࠸ࡓ࡜ゝ࠼ࡲࡍࠋEFDࡶࡇ

ࢀࡲ࡛CFD࡟ᑐࡋ࡚ഐほ⪅ⓗ࡞❧ሙ࡟⤊ጞࡋ࡚ࡁࡓࡇ࡜ࡣྰࡵ࡞࠸୍᪉ࠊEFD࡟ࡣEFDᅛ᭷

ࡢ୙☜࠿ࡉࡀ࠶ࡾࠊࡲࡓᚓࡽࢀࡿ᝟ሗ࡟ࡶไ㝈ࡀ⏕ࡌࡲࡍࠋ☜ᐇ࡟ゝ࠼ࡿࡇ࡜ࡣࠊEFD/CFD ༢⊂࡛ᚓࡽࢀࡿࢹ࣮ࢱࡢ⢭ᗘࡸಙ㢗ᛶ࡟ࡣ⮬ࡎ࠿ࡽ㝈⏺ࡀ⏕ࡌࡿ࡜࠸࠺ࡇ࡜࡛ࡋࡻ࠺ࠋ

୍᪉ࠊ኱Ꮫࡢ◊✲ᐊ➼࡟࠾࠸࡚ࡣࠊᐇ㦂࡜ィ⟬ࡢ୧㠃࠿ࡽࡢ࢔ࣉ࣮ࣟࢳࡣ᪥ᖖⓗ࡞ᡭẁ࡛࠶

ࡾࠊᐇ㦂࡜ィ⟬ࡢ༢⣧࡞ẚ㍑࠿ࡽ⪃ᐹࢆࡋ࡚⾜ࡃ࡜࠸࠺ព࿡࡛ࡣEFD࡜CFDࡣᖖ࡟ᐦ᥋࡞㛵ಀ

࡟࠶ࡾࡲࡍࠋ

ࡇࡢࡼ࠺࡞⌧≧࡟㚷ࡳࠊ◊✲ᡤ࡟࠾࠸࡚ࡣࠊ஧ඖㄽⓗ࡞⪃࠼᪉ࢆᨵࡵ཮᪉ࡢಙ㢗ᛶࢆྥୖࡉ ࡏ┿࡟ᐇ⏝࡟౪ࡍࡿࢶ࣮ࣝ࡜࡞ࡍࡓࡵ࡟ࠊࡲࡓࠊ኱Ꮫ➼࡟࠾࠸࡚ࡣࠊ༢⣧ẚ㍑ࢆ㉸࠼ࡓࡼࡾ῝

࠸Ὕᐹ࣭▱ぢࢆᚓࡽࢀࡿࡼ࠺࡟ࡍࡿࡓࡵࠊEFD/CFDࡢ஫࠸ࡢၥ㢟Ⅼࡢ⿵᏶ࡸ᪂ࡓ࡞ᯟ⤌ࡳࡢ ᵓ⠏࡟ࡼࡗ࡚ᚓࡽࢀࡿࢩࢼࢪ࣮ຠᯝࢆぢ࠸ࡔࡍࡇ࡜ࡀ㔜せ࡛ࡣ࡞࠸࡛ࡋࡻ࠺࠿ࠋ

ᮏ࣮࣡ࢡࢩࣙࢵࣉࡣࡇࡢࡼ࠺࡞EFD࡜CFDࡢ⼥ྜࢆࢸ࣮࣐࡜ࡋࠊὶయຊᏛ࡟ᦠࢃࡿ◊✲⪅ࡸ

ᢏ⾡⪅ࡀㅮ₇ࡸࢹ࢕ࢫ࢝ࢵࢩࣙࣥࢆ㏻ࡋ࡚ࡑࡢᚲせᛶ࣭㔜せᛶ࡟ࡘ࠸࡚ㄆ㆑ࢆ῝ࡵࠊ࠿ࡘ▱ぢ

ࢆᗈࡆࡿࡇ࡜ࢆ┠ⓗ࡜ࡋ࡚࠾ࡾࡲࡍࠋ

➨1ᅇ఍ྜࡣࠊ2008ᖺ2᭶26᪥㸦ⅆ㸧࡟Ᏹᐂ⯟✵◊✲㛤Ⓨᶵᵓ㸦JAXA㸧ㄪᕸ⯟✵Ᏹᐂࢭࣥࢱ

࣮࡟࠾࠸࡚㛤ദࡉࢀࡲࡋࡓࠋࡋ࠿ࡋࠊཧຍ⪅ࢆ㝈ᐃࡋࡓ㠀බᘧ࡞࣮࣡ࢡࢩࣙࢵࣉ࡛࠶ࡗࡓࡇ࡜

࠿ࡽࠊ௒ᅇࡣつᶍࢆᣑ኱ࡋ࡚ᗈࡃཧຍࢆ࿧ࡧ࠿ࡅࡲࡋࡓࠋᅜෆᣍᚅㅮ₇࡟ࡘ࠸࡚ࡣࠊ⯟✵Ᏹᐂ

࡟㝈ࡽࡎᵝࠎ࡞ศ㔝࠿ࡽㅮ₇ࢆ࠾㢪࠸ࡋࠊ⯆࿡῝ࡃ࠿ࡘ㠀ᖖ࡟ࣂ࢚࣮ࣜࢩࣙࣥ࡟࡜ࢇࡔෆᐜ࡟

࡞ࡗࡓ࡜⮬㈇ࡋ࡚࠾ࡾࡲࡍࠋࡉࡽ࡟ᮏศ㔝࡟࠾࠸࡚ᾏእ࡛ඛ㥑ⓗ࡞◊✲άືࢆ⾜ࡗ࡚ࡇࡽࢀࡓࠊ

ࡶ௻⏬࠸ࡓࡋࡲࡋࡓࠋ

ࡇࡢ࣮࣡ࢡࢩࣙࢵࣉࡀࠊEFD/CFD⼥ྜ࡜࠸࠺ྂࡃ࡚᪂ࡋ࠸ࢸ࣮࣐࡟㛵ࡋ࡚᝟ሗ஺᥮ࢆࡍࡿ

ࡼ࠸ᶵ఍࡜࡞ࡾࠊ᪂ࡓ࡞Ⓨ᝿࡟ࡼࡿ◊✲㛤Ⓨάືࡀᅜෆእ࡛ࡼࡾ୍ᒙᒎ㛤ࡉࢀࡿࡼ࠺࡟࡞ࢀࡤࠊ

୺ദ⪅࡜ࡋ࡚ఱࡼࡾࡢ႐ࡧ࡛ࡍࠋࡲࡓࠊᮏ࣮࣡ࢡࢩࣙࢵࣉࡣ௒ᚋࡶ⥅⥆ࡉࡏ࡚࠸ࡃணᐃ࡛ࡍࡢ

࡛ࠊෆᐜ࡟ࡘ࠸࡚ࡈពぢࡸࡈᥦ᱌➼ࡈࡊ࠸ࡲࡋࡓࡽࡐࡦ࡜ࡶ࠾▱ࡽࡏ࠸ࡓࡔࡁࡓࡃࠊᐅࡋࡃ࠾

㢪࠸⏦ࡋୖࡆࡲࡍࠋ

ᖹᡂ

ᖺ

᭶

᪥

➨

ᅇ

⼥ྜ࣮࣡ࢡࢩࣙࢵࣉᐇ⾜ጤဨ఍ ጤဨ㛗 Ᏹᐂ⯟✵◊✲㛤Ⓨᶵᵓ ◊✲㛤Ⓨᮏ㒊 ᯇᑿ ⿱୍

ᮾ໭኱Ꮫὶయ⛉Ꮫ◊✲ᡤ ኱ᯘⱱ

ጤဨ㛗 ᯇᑿ ⿱୍ -$;$ ◊✲㛤Ⓨᮏ㒊ᩘ್ゎᯒࢢ࣮ࣝࣉ

኱ᯘ ⱱ ᮾ໭኱Ꮫὶయ⛉Ꮫ◊✲ᡤ㝃ᒓὶయ⼥ྜ◊✲ࢭࣥࢱ࣮

ጤဨ 㟷ᒣ๛ྐ -$;$ ◊✲㛤Ⓨᮏ㒊ᩘ್ゎᯒࢢ࣮ࣝࣉ

బ᐀❶ᘯ ྡྂᒇ኱Ꮫ኱Ꮫ㝔ᕤᏛ◊✲⛉⯟✵ᏱᐂᕤᏛᑓᨷ

㕥ᮌᏹ஧㑻 ᮾி኱Ꮫ኱Ꮫ㝔᪂㡿ᇦ๰ᡂ⛉Ꮫ◊✲⛉ඛ➃࢚ࢿࣝࢠ࣮ᕤᏛᑓᨷ ᯇᒣ᪂࿃ -$;$ ◊✲㛤Ⓨᮏ㒊ᩘ್ゎᯒࢢ࣮ࣝࣉ

ᒣᮏ୍⮧ -$;$ ⯟✵ࣉࣟࢢ࣒ࣛࢢ࣮ࣝࣉᅜ⏘᪑ᐈᶵࢳ࣮࣒

ྜྷ⏣᠇ྖ -$;$ ⯟✵ࣉࣟࢢ࣒ࣛࢢ࣮ࣝࣉ㉸㡢㏿ᶵࢳ࣮࣒

Ώ㎶㔜ဢ -$;$ ◊✲㛤Ⓨᮏ㒊㢼Ὕᢏ⾡㛤Ⓨࢭࣥࢱ࣮

஦ົᒁ ┦᭮⚽᫛ -$;$ ◊✲㛤Ⓨᮏ㒊ᩘ್ゎᯒࢢ࣮ࣝࣉ

ཱྀ▼ ⱱ -$;$ ◊✲㛤Ⓨᮏ㒊㢼Ὕᢏ⾡㛤Ⓨࢭࣥࢱ࣮

㻲㼑㼎㻚㻌㻞㻟㻌㻔㻹㼛㼚㻚㻕㻘㻌㻞㻜㻜㻥㻌

㻝㻟㻦㻜㻜㻙㻝㻟㻦㻜㻡㻌 ᯇᑿ㻌 ⿱୍㻌 㻔㻶㻭㼄㻭㻕㻌

㼅㼡㼕㼏㼔㼕㻌㻹㼍㼠㼟㼡㼛㻌㻔㻶㻭㼄㻭㻕㻌 㻻㼜㼑㼚㼕㼚㼓㻌㻭㼐㼐㼞㼑㼟㼟㻌

㻿㼑㼟㼟㼕㼛㼚㻌㻝㻌㻨㻶㻭㼄㻭㻌㻷㼑㼥㼚㼛㼠㼑㻌㻿㼜㼑㼑㼏㼔㻪㻌 ྖ఍㻦㻌 ᯇᑿ㻌 ⿱୍㻌 㻔㻶㻭㼄㻭㻕

㻯㼔㼍㼕㼞㼜㼑㼞㼟㼛㼚㻦㻌㼅㼡㼕㼏㼔㼕㻌㻹㼍㼠㼟㼡㼛㻌㻔㻶㻭㼄㻭㻕 㻝㻟㻦㻜㻡㻙㻝㻟㻦㻟㻡㻌 Ώ㎶㻌 㔜ဢ㻌 㻔㻶㻭㼄㻭㻕㻌

㻿㼔㼕㼓㼑㼥㼍㻌㼃㼍㼠㼍㼚㼍㼎㼑㻌㻔㻶㻭㼄㻭㻕㻌

㻭㻌㻿㼥㼟㼠㼑㼙㻌㻯㼛㼚㼏㼑㼜㼠㻌㼒㼛㼞㻌㻱㻲㻰㻛㻯㻲㻰㻌㻵㼚㼠㼑㼓㼞㼍㼠㼕㼛㼚㻦㻌㻶㻭㼄㻭㻌㻰㼕㼓㼕㼠㼍㼘㻛㻭㼚㼍㼘㼛㼓㻌㻴㼥㼎㼞㼕㼐㻌㼃㼕㼚㼐㻌 㼀㼡㼚㼚㼑㼘㻌

㻿㼑㼟㼟㼕㼛㼚㻌㻞㻌㻨㻵㼚㼢㼕㼠㼑㼐㻌㻸㼑㼏㼠㼡㼞㼑㻏㻝㻪㻌 ྖ఍㻦㻌 ᯇᑿ㻌 ⿱୍㻌 㻔㻶㻭㼄㻭㻕

㻯㼔㼍㼕㼞㼜㼑㼞㼟㼛㼚㻦㻌㼅㼡㼕㼏㼔㼕㻌㻹㼍㼠㼟㼡㼛㻌㻔㻶㻭㼄㻭㻕 㻝㻟㻦㻠㻡㻙㻝㻠㻦㻠㻡㻌 㻭㼚㼐㼞㼑㼣㻌 㻹㼑㼍㼐㼑㻌 㻔㼃㼕㼘㼘㼕㼍㼙㻌 㻹㼍㼞㼟㼔㻌 㻾㼕㼏㼑㻌

㼁㼚㼕㼢㼑㼞㼟㼕㼠㼥㻘㻌㼁㻿㻭㻌㻕㻌 㻾㼑㼐㼡㼏㼕㼚㼓㻌㼁㼚㼏㼑㼞㼠㼍㼕㼚㼠㼥㻌㼕㼚㻌㻱㻲㻰㻌㼍㼚㼐㻌㻯㻲㻰㻌㼀㼔㼞㼛㼡㼓㼔㻌㻰㼍㼠㼍㻛㻹㼛㼐㼑㼘㻌㻲㼡㼟㼕㼛㼚㻌 㻝㻠㻦㻡㻜㻙㻝㻡㻦㻡㻜㻌 㻱㼐㼣㼍㼞㼐㻌㻺㻚㻌㼀㼕㼚㼛㼏㼛㻌㻔㻌㻮㼛㼑㼕㼚㼓㻌㻯㼛㼙㼙㼑㼞㼏㼕㼍㼘㻌

㻭㼕㼞㼜㼘㼍㼚㼑㼟㻌㻘㻌㼁㻿㻭㻕㻌

㼂㼍㼘㼕㼐㼍㼠㼕㼛㼚㻌㼍㼚㼐㻌㻹㼕㼚㼕㼙㼕㼦㼕㼚㼓㻌㻯㻲㻰㻌㼁㼚㼏㼑㼞㼠㼍㼕㼚㼠㼥㻌㼒㼛㼞㻌㻯㼛㼙㼙㼑㼞㼏㼕㼍㼘㻌㻭㼕㼞㼏㼞㼍㼒㼠㻌㻭㼜㼜㼘㼕㼏㼍㼠㼕㼛㼚㼟 㻌 㻙㻌㼀㼔㼑㻌㻵㼚㼠㼑㼓㼞㼍㼠㼕㼛㼚㻌㼛㼒㻌㻱㻲㻰㻌㼍㼚㼐㻌㻯㻲㻰㻌

㻼㼍㼚㼑㼘㻌㻰㼕㼟㼏㼡㼟㼟㼕㼛㼚㼟㻌 䝁䞊䝕䜱䝛䞊䝍㻦㻌 ྜྷ⏣㻌 ᠇ྖ㻌 㻔㻶㻭㼄㻭㻕

㻯㼛㼛㼞㼐㼕㼚㼍㼠㼛㼞㻦㻌㻷㼑㼚㼖㼕㻌㼅㼛㼟㼔㼕㼐㼍㻌㻔㻶㻭㼄㻭㻕

㻝㻢㻦㻜㻜㻙㻝㻤㻦㻜㻜㻌

㻯㼍㼚㻌㻱㻲㻰㻛㻯㻲㻰㻌㻵㼚㼠㼑㼓㼞㼍㼠㼕㼛㼚㻌㻹㼕㼚㼕㼙㼕㼦㼑㻌㼁㼚㼏㼑㼞㼠㼍㼕㼚㼠㼥㻫㻌 㻼㼍㼚㼑㼘㼕㼟㼠㼟㻦㻌㻭㼚㼐㼞㼑㼣㻌㻹㼑㼍㼐㼑㻌㻔㻾㼕㼏㼑㻌㼁㼚㼕㼢㻚㻕㻌

㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻱㼐㼣㼍㼞㼐㻌㻺㻚㻌㼀㼕㼚㼛㼏㼛㻌㻔㻌㻮㼛㼑㼕㼚㼓㻌㻯㼛㼙㼙㼑㼞㼏㼕㼍㼘㻌㻭㼕㼞㼜㼘㼍㼚㼑㼟㻌㻕㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 బ᐀㻌 ❶ᘯ㻌 㻭㼗㼕㼔㼕㼞㼛㻌㻿㼍㼟㼛㼔㻌㻔㻺㼍㼓㼛㼥㼍㻌㼁㼚㼕㼢㻚㻕㻌

㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 ㉺ᬛ㻌 ❶⏕㻌 㻭㼗㼕㼛㻌㻻㼏㼔㼕㻌㻔㻷㼍㼣㼍㼟㼍㼗㼕㻌㻴㼑㼍㼢㼥㻌㻵㼚㼐㼡㼟㼠㼞㼕㼑㼟㻕㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 㻌 ཱྀ▼㻌 ⱱ㻌 㻿㼔㼕㼓㼑㼞㼡㻌㻷㼡㼏㼔㼕㻙㻵㼟㼔㼕㻌㻔㻶㻭㼄㻭㻕㻌

㻝㻤㻦㻜㻜㻙㻞㻜㻦㻜㻜㻌 㻮㼍㼚㼝㼡㼑㼠㻌㻔㻶㻭㼄㻭㻌㻾㼑㼟㼠㼍㼡㼞㼍㼚㼠㻕㻌

㻌 㻌 㻌

㻲㼑㼎㻚㻌㻞㻠㻌㻔㼀㼡㼑㻚㻕㻘㻌㻞㻜㻜㻥㻌

㻿㼑㼟㼟㼕㼛㼚㻌㻟㻌㻨㻵㼚㼢㼕㼠㼑㼐㻌㻸㼑㼏㼠㼡㼞㼑㻏㻞㻪㻌 ྖ఍㻦㻌 ᯇᑿ㻌 ⿱୍㻌 㻔㻶㻭㼄㻭㻕㻘㻌 ኱ᯘ㻌 ⱱ㻌 㻔ᮾ໭኱ὶయ◊㻕

㻯㼔㼍㼕㼞㼜㼑㼞㼟㼛㼚㻦㻌 㼅㼡㼕㼏㼔㼕㻌 㻹㼍㼠㼟㼡㼛㻌 㻔㻶㻭㼄㻭㻕㻌 㼍㼚㼐㻌 㻿㼔㼕㼓㼑㼞㼡㻌 㻻㼎㼍㼥㼍㼟㼔㼕㻌 㻔㼀㼛㼔㼛㼗㼡㻌 㼁㼚㼕㼢㻚㻕 㻝㻜㻦㻜㻜㻙㻝㻜㻦㻠㻜㻌 బ᐀㻌 ❶ᘯ䠄ྡ኱ᕤ䠅㻌

㻭㼗㼕㼔㼕㼞㼛㻌㻿㼍㼟㼛㼔㻌㻔㻺㼍㼓㼛㼥㼍㻌㼁㼚㼕㼢㻚㻕㻌

ᬑẁ╔䛾㻱㻲㻰㻛㻯㻲㻰㐃ᦠ䡚㉸㡢㏿ᢠຊపῶ䜢㢟ᮦ䛸䛧䛶䡚㻌

㼀㼔㼕㼟㻌㼕㼟㻌㻻㼡㼞㻌㼁㼟㼡㼍㼘㻌㻱㻲㻰㻛㻯㻲㻰㻌㻯㼛㼘㼘㼍㼎㼛㼞㼍㼠㼕㼛㼚㻙㻌㻭㼚㻌㻱㼤㼍㼙㼜㼘㼑㻌㼕㼚㻌㻿㼡㼜㼑㼞㼟㼛㼚㼕㼏㻌㻰㼞㼍㼓㻌 㻾㼑㼐㼡㼏㼠㼕㼛㼚㻌㻿㼠㼡㼐㼥㻌

㻝㻜㻦㻠㻜㻙㻝㻝㻦㻝㻜㻌 ຍ⸨㻌 ༓ᖾ㻌 㻔ᮾ኱⏕⏘◊㻕㻌 㻯㼔㼕㼟㼍㼏㼔㼕㻌㻷㼍㼠㼛㻌㻔㼁㼚㼕㼢㻚㻌㼀㼛㼗㼥㼛㻕㻌

ὶయ䛛䜙Ⓨ⏕䛩䜛㡢䛾ᩘ್ゎᯒ䛸ᐇ㦂ィ 㻌

㻺㼡㼙㼑㼞㼕㼏㼍㼘㻌㻯㼛㼙㼜㼡㼠㼍㼠㼕㼛㼚㻌㼍㼚㼐㻌㻱㼤㼜㼑㼞㼕㼙㼑㼚㼠㼍㼘㻌㻹㼑㼍㼟㼡㼞㼑㼙㼑㼚㼠㻌㼛㼒㻌㻭㼑㼞㼛㼐㼥㼚㼍㼙㼕㼏㻌㻺㼛㼕㼟㼑㻌 㻝㻝㻦㻝㻜㻙㻝㻝㻦㻡㻜㻌 ఀ⸨㻌 ㈗அ㻌 㻔䛚Ⲕ䛾ỈዪᏊ኱㻕㻌

㼀㼍㼗㼍㼥㼡㼗㼕㻌㻵㼠㼛㻌㻔㻻㼏㼔㼍㼚㼛㼙㼕㼦㼡㻌㼁㼚㼕㼢㻚㻕㻌

ᐇ✵㛫䛸᫬✵㛫䛻䛚䛡䜛ὶయྍど໬㻌 㻲㼘㼛㼣㻌㼂㼕㼟㼡㼍㼘㼕㼦㼍㼠㼕㼛㼚㻌㼕㼚㻌㻾㼑㼍㼘㻌㼍㼚㼐㻌㼀㼕㼙㼑㻌㻿㼜㼍㼏㼑㼟㻌 㻝㻝㻦㻡㻜㻙㻝㻞㻦㻞㻜㻌 ᆤ಴㻌 ㄔ㻌 㻔໭኱ᕤ㻕㻌

㻹㼍㼗㼛㼠㼛㻌㼀㼟㼡㼎㼛㼗㼡㼞㼍㻌㻔㻴㼛㼗㼗㼍㼕㼐㼛㻌㼁㼚㼕㼢㻚㻕㻌

⮬ື㌴✵ຊศ㔝䛻䛚䛡䜛㻱㻲㻰䞉㻯㻲㻰ᢏ⾡䛾ືྥ䛸㠀ᐃᖖ✵ຊ䝅䝭䝳䝺䞊䝍䛾㛤Ⓨ

㻯㼡㼞㼞㼑㼚㼠㻌㻿㼠㼍㼠㼡㼟㻌㼛㼒㻌㻱㻲㻰㻛㻯㻲㻰㻌㼀㼑㼏㼔㼚㼕㼝㼡㼑㼟㻌㼒㼛㼞㻌㻾㼛㼍㼐㻌㼂㼑㼔㼕㼏㼘㼑㻌㻭㼑㼞㼛㼐㼥㼚㼍㼙㼕㼏㼟㻌㼍㼚㼐㻌 㻰㼑㼢㼑㼘㼛㼜㼙㼑㼚㼠㻌㼛㼒㻌㼠㼔㼑㻌㼁㼚㼟㼠㼑㼍㼐㼥㻌㻭㼑㼞㼛㼐㼥㼚㼍㼙㼕㼏㻌㻿㼕㼙㼡㼘㼍㼠㼛㼞㻌

㻝㻞㻦㻞㻜㻙㻝㻟㻦㻟㻜㻌 㻸㼡㼚㼏㼔㻌

㻝㻟㻦㻟㻜㻙㻝㻠㻦㻝㻜㻌

ᵽཱྀ㻌 ▱அ㻌 㻔⤫ィᩘ⌮◊㻕㻌

㼀㼛㼙㼛㼥㼡㼗㼕㻌 㻴㼕㼓㼡㼏㼔㼕㻌 㻔㻵㼚㼟㼠㼕㼠㼡㼠㼑㻌 㼛㼒㻌 㻿㼠㼍㼠㼕㼟㼠㼕㼏㼍㼘㻌㻹㼍㼠㼔㼑㼙㼍㼠㼕㼏㼟㻕㻌

㏲ḟ䝕䞊䝍ྠ໬䠖㻌 䝅䝭䝳䝺䞊䝅䝵䞁䛸ほ 䝕䞊䝍䛾䝸䜰䝹䝍䜲䝮⤫⼥ྜ䜢䜑䛦䛧䛶 㻿㼑㼝㼡㼑㼚㼠㼕㼍㼘㻌㻰㼍㼠㼍㻌㻭㼟㼟㼕㼙㼕㼘㼍㼠㼕㼛㼚㻦㻌㻻㼚㼘㼕㼚㼑㻌㻵㼚㼒㼛㼞㼙㼍㼠㼕㼛㼚㻌㻲㼡㼟㼕㼛㼚㻌㻼㼘㼍㼠㼒㼛㼞㼙㻌㼒㼛㼞㻌㻿㼕㼙㼡㼘㼍㼠㼕㼛㼚㻌 㼍㼚㼐㻌㻻㼎㼟㼑㼞㼢㼍㼠㼕㼛㼚㻌㻰㼍㼠㼍㻌

㻝㻠㻦㻝㻜㻙㻝㻠㻦㻡㻜㻌 ㇂㻌 Ὀᐶ䠄஑኱ᕤ䠅㻌

㼅㼍㼟㼡㼔㼕㼞㼛㻌㼀㼍㼚㼕㻌㻔㻷㼥㼡㼟㼔㼡㻌㼁㼚㼕㼢㻚㻕㻌

෌౑⏝ᆺᏱᐂ 㑏ᶵ䛻ྥ䛡䛯◊✲䛻䛚䛡䜛 㻱㻲㻰㻛㻯㻲㻰 䛾ྲྀ䜚⤌䜏㻌 㻱㻲㻰㻛㻯㻲㻰㻌㻭㼏㼠㼕㼢㼕㼠㼕㼑㼟㻌㼕㼚㻌㻾㼑㼟㼑㼍㼞㼏㼔㻌㼒㼛㼞㻌㻾㼑㼡㼟㼍㼎㼘㼑㻌㻸㼍㼡㼚㼏㼔㻌㼂㼑㼔㼕㼏㼘㼑㼟㻌

㻿㼑㼟㼟㼕㼛㼚㻌㻠㻌㻨㼀㼑㼏㼔㼚㼕㼏㼍㼘㻌㻿㼑㼟㼟㼕㼛㼚㻪㻌 ྖ఍㻦㻌 ┦᭮㻌 ⚽᫛㻌 㻔㻶㻭㼄㻭㻕

㻯㼔㼍㼕㼞㼜㼑㼞㼟㼛㼚㻦㻌㻴㼕㼐㼑㼍㼗㼕㻌㻭㼕㼟㼛㻌㻔㻶㻭㼄㻭㻕 㻝㻡㻦㻜㻜㻙㻝㻡㻦㻟㻜㻌 ಖỤ㻌 䛛䛺Ꮚ㻌 㻔ᮾ໭኱㝔㻕㻌

㻷㼍㼚㼍㼗㼛㻌㼅㼍㼟㼡㼑㻌㻔㼀㼛㼔㼛㼗㼡㻌㼁㼚㼕㼢㻚㻕㻌

㻰㼕㼟㼏㼛㼚㼠㼕㼚㼡㼛㼡㼟㻌㻳㼍㼘㼑㼞㼗㼕㼚 ἲ䜢⏝䛔䛯 㻭㻳㻭㻾㻰㻙㻮 ᶆ‽ᶍᆺ䛾㟼ⓗ✵ຊᙎᛶゎᯒ㻌 㻿㼠㼍㼠㼕㼏㻌㻭㼑㼞㼛㼑㼘㼍㼟㼠㼕㼏㼕㼠㼥㻌㻭㼚㼍㼘㼥㼟㼕㼟㻌㼛㼒㻌㻭㻳㻭㻾㻰㻙㻮㻌㼃㼕㼚㼐㻌㼀㼡㼚㼚㼑㼘㻌㻯㼍㼘㼕㼎㼞㼍㼠㼕㼛㼚㻌㻹㼛㼐㼑㼘㻌㼁㼟㼕㼚㼓㻌 㻰㼕㼟㼏㼛㼚㼠㼕㼚㼡㼛㼡㼟㻌㻳㼍㼘㼑㼞㼗㼕㼚㻌㻹㼑㼠㼔㼛㼐㻌

㻝㻡㻦㻟㻜㻙㻝㻢㻦㻜㻜㻌 ຍ⸨㻌 ༤ྖ㻌 㻔ᮾ໭኱㝔㻕㻌 㻴㼕㼞㼛㼟㼔㼕㻌㻷㼍㼠㼛㻌㻔㼀㼛㼔㼛㼗㼡㻌㼁㼚㼕㼢㻚㻕㻌

ᐇ኱Ẽ⎔ቃ䜢⪃៖䛧䛯ᚋ᪉஘Ẽὶண ᢏ⾡㻌

㻼㼞㼑㼐㼕㼏㼠㼕㼛㼚㻌㼛㼒㻌㼃㼍㼗㼑㻌㼀㼡㼞㼎㼡㼘㼑㼚㼏㼑㻌㼡㼚㼐㼑㼞㻌㻭㼏㼠㼡㼍㼘㻌㻭㼠㼙㼛㼟㼜㼔㼑㼞㼕㼏㻌㻯㼛㼚㼐㼕㼠㼕㼛㼚㻌 㻝㻢㻦㻜㻜㻙㻝㻢㻦㻟㻜㻌 ᶫᮏ㻌 ᩔ㻌 㻔㻶㻭㼄㻭㻕㻌

㻭㼠㼟㼡㼟㼔㼕㻌㻴㼍㼟㼔㼕㼙㼛㼠㼛㻌㻔㻶㻭㼄㻭㻕㻌

㻶㻭㼄㻭䝕䝆䝍䝹㻛䜰䝘䝻䜾䞉䝝䜲䝤䝸䝑䝗㢼Ὕ㻦㻌 䝕䝆䝍䝹㢼Ὕ䛾㛤Ⓨ㻌

㻶㻭㼄㻭㻌㻰㼕㼓㼕㼠㼍㼘㻛㻭㼚㼍㼘㼛㼓㻌㻴㼥㼎㼞㼕㼐㻌㼃㼕㼚㼐㻌㼀㼡㼚㼚㼑㼘㻦㻌㻰㼑㼢㼑㼘㼛㼜㼙㼑㼚㼠㻌㼛㼒㻌㻰㼕㼓㼕㼠㼍㼘㻌㼃㼕㼚㼐㻌㼀㼡㼚㼚㼑㼘㻌 㻝㻢㻦㻟㻜㻙㻝㻣㻦㻜㻜㻌 ⸨⏣㻌 ┤⾜㻌 㻔㻶㻭㼄㻭㻕㻌

㻺㼍㼛㼥㼡㼗㼕㻌㻲㼡㼖㼕㼠㼍㻌㻔㻶㻭㼄㻭㻕㻌

㻶㻭㼄㻭䝕䝆䝍䝹㻛䜰䝘䝻䜾䞉䝝䜲䝤䝸䝑䝗㢼Ὕ㻦㻌 䜰䝘䝻䜾㢼Ὕ䛾㧗㏿໬㻌

㻶㻭㼄㻭㻌㻰㼕㼓㼕㼠㼍㼘㻛㻭㼚㼍㼘㼛㼓㻌㻴㼥㼎㼞㼕㼐㻌㼃㼕㼚㼐㻌㼀㼡㼚㼚㼑㼘㻦㻌㻿㼜㼑㼑㼐㻙㼁㼜㻌㼀㼑㼏㼔㼚㼕㼝㼡㼑㻌㼛㼒㻌㻭㼚㼍㼘㼛㼓㻌㼃㼕㼚㼐㻌 㼀㼡㼚㼚㼑㼘㻌

㻝㻣㻦㻜㻜㻙㻝㻣㻦㻜㻡㻌 ኱ᯘ㻌 ⱱ㻌 㻔ᮾ໭኱ὶయ◊㻕㻌

㻿㼔㼕㼓㼑㼞㼡㻌㻻㼎㼍㼥㼍㼟㼔㼕㻌㻔㼀㼛㼔㼛㼗㼡㻌㼁㼚㼕㼢㻚㻕㻌 㻯㼘㼛㼟㼕㼚㼓㻌㻭㼐㼐㼞㼑㼟㼟㻌

A System Concept for EFD/CFD Integration:

A System Concept for EFD/CFD Integration:

Digital/Analog Hybrid Wind Tunnel Digital/Analog Hybrid Wind Tunnel

Shigeya Watanabe Shigeya Watanabe

Wind T nnel Technolog Center Wind Tunnel Technology Center

Aerospace Research and Development Directoratep p Japan Aerospace Exploration Agency (JAXA) Japan Aerospace Exploration Agency (JAXA)

The 2^ndWorkshop on EFD/CFD Integration The 2 Workshop on EFD/CFD Integration

Feb 23 2009 at JAXA Chofu Aerospace Center Tokyo Japan Feb. 23, 2009 at JAXA Chofu Aerospace Center, Tokyo, Japan

Contents Contents

„ B k d M ti ti d Obj ti

„ Background, Motivation, and Objectives

„ ³'Lgital-Analog HyEULG:LQG7XQQHO´g g y

‡ Concept

‡ Concept

S t A hit t

‡ System Architecture

„ Technical element developments for Hybrid WTp y

‡ 'LJLWDO :7

‡ 'LJLWDO WT

A l WT A l ti f d ti f fl i i d t

‡ Analog WT: Acceleration of reduction of flow imaging data

‡ ,QWHgUDWLRQRI()'DQG&)'g

„ Summary

„ Summary

„ EFD: Analog, Real

Experimental Fluid Dynamics g,

Experimental Fluid Dynamics (Wind tunnel test)

(Wind tunnel test)

JAXA 2mx2m JAXA 2mx2m Transonic WT

Pressure measurement by PSP

Digital Virtual

„ CFD: Digital, Virtual

„ CFD:

Computational Fluid Dynamics Computational Fluid Dynamics

Two major methods for aerodynamic characteristics predictionp

„ I t ti f EFD/CFD

„ Integration g of EFD/CFD䋺

䋨or fusion synergy䋩 䋨or fusion, synergy䋩

䊶More than simple collaboration

䊶Aiming a new world where 1 + 1 > 2 䊶Aiming a new world where 1 + 1 > 2

Motivation 1

Challenges of EFD & CFD in Aerodynamic Prediction

Motivation 1

Challenges of EFD & CFD in Aerodynamic Prediction

„EFD side:

„EFD side:

9Diff f l fli ht diti R b Wi d t l

9Differences from real flight condition䋺 Re number, Wind tunnel

ll/ t M d l d f ti t

wall/support, Model deformation, etc.

9Restriction of measurement items䋺 Mainly, force/moment/discrete y, pressure. Additionally, flow visualizations and quantitative image pressure. Additionally, flow visualizations and quantitative image measurements (PIV PSP)

measurements (PIV, PSP).

ЈDiffi lti i d t di fl fi ld d t bl h ti ЈDifficulties in understanding flowfield and troubleshooting

9Relativelyy long period of a whole WT test campaigng p p g including model g manufacturing.^g Need for a unified platform toNeed for a unified platform to

i lt l l b th h ll

„CFD side: simultaneously solve both challenges.

9Reliability is always a concern especially in turbulence transition 9Reliability is always a concern, especially in turbulence, transition,

separation and reacting flow separation, and reacting flow.

V lid ti b i t i d

Ј Validation by experiments required.

9Long calculation time (low data productivity) for high-fidelity CFD.g ( p y) g y 9Higher skill and long time for grid generation required

9Higher skill and long time for grid generation required.

Ex ) Typical comparison

㘑ᵢ⸳ቯ♖ᐲ䈮䉋䉎䉵䊧

Ex. ) Typical comparison

EFD

㘑ᵢ⸳ቯ♖ᐲ䈮䉋䉎䉵䊧

EFD pretest CFD _(PSP)^{EFD CFD} EFD pretest CFD _(PSP)

Mach number 0.904 0.9

Re number 2 1㬍10⁶ 2 0㬍10⁶

Re number 2.1㬍10 2.0㬍10

A l f tt k 2 6㫦 2 0㫦

Angle of attack 2.6㫦

(i l di d fl ti

2.0㫦

(including deflection of balance and sting)

(= setting angle) of balance and sting)

M d l ^{D i d fi D i d}

Model Designed config.

d f ti d

Designed fi ti configuration ⁺deformation due

t l d

configuration to aero. load

Boundary With WT wall and Model alone in

condition ^{sting freestream} -with deformation of balance/sting

-no deformation of balance /sting

Boundary layer Natural or forced Fully laminar/ of balance/sting -with model

of balance /sting -No model

Boundary layer

transition location ^transition

y

turbulent or fixed

with model deformation

No model deformation

transition location

or predicted -with wall/sting N t l t iti

-with sting F ll t b l t

Uncertainty CD

䋺 㫧0.0005

Uncertainty CD

䋺 㫧0.0005

U k if t d Methodology not established

Unknown if not measured Methodology not established

Motivation 3

Prediction of Aero Characteristics at Real Flight

Motivation 3

Prediction of Aero. Characteristics at Real Flight

Goal of EFD & CFD in design: To predict aerodynamic characteristics at real Goal of EFD & CFD in design: To predict aerodynamic characteristics at real

flight condition accurately and effectively flight condition accurately and effectively

*I fl diti Differences: physics model, Refined H b id

*Inflow condition,

WT wall model e e ces p ys cs ode , Hybrid

Boundary cond * model shape etc Matched

WT wall, model

support, etc. Boundary cond. , model shape, etc. WT

support, etc. WT

EFD CFD

(Wind Tunnel) CFD

(Wind Tunnel)

High Re High-Re Wind Tunnel

Differences: Wind Tunnel Differences:

flow condition (Re), model shape,

Real Flight

o co d t o ( e), model shape

ode s ape, physics model etc

Real Flight (i l Fli ht T t)

model shape, Boundary cond

physics model, etc.

(incl. Flight Test)

Boundary cond., etc

etc.

„ Innovation of both EFD and CFD technologiesg through EFD/CFD integration

through EFD/CFD integration.

„ Reduction of cost and risk and improvement of

„ Reduction of cost and risk and improvement of accuracy and reliability of aerodynamic data in y y y aerospace vehicle development

aerospace vehicle development.

Ј Improved contribution to developments of Japanese Ј Improved contribution to developments of Japanese

i l h MRJ d th JAXA j t

airplanes such as MRJ and other JAXA projects.

Silent Supersonic Technology D t t S³TD (JAXA)

Courtesy of Mitsubishi Heavy Industries, Ltd.

Demonstrator, S³TD (JAXA) Japanese Regional Jet MRJ

Japanese Regional Jet, MRJ

NASA Langley䋺 Virtual Diagnostics Interface (ViDI) System NASA Langley䋺 Virtual Diagnostics Interface (ViDI) System

R. Schwartz, G. Fleming 2008

‡ Features and functions:Features and functions:

3D CAD based virtual diagnostics system - 3D-CAD-based virtual diagnostics system

for experiment design for experiment design

9O ti i ti f t l i

9Optimization of measurement planning

i i t l k

using virtual mockup Evaluation of

i i d

9Real-time comparison and visualization viewing area and shade prior to exp.

of CFD and WT data (LiveView3D)

Evaluation of PSP setting shade prior to exp.

- Similar systems are not known in other

Evaluation of PSP setting using 3-D CAD

y

industrial wind tunnels in the world.

Superiority of the present Hybrid WT

‡ 7HFKQLFDOGLVDGYDQWDges:g

- Data comparison relatively difficult due toData comparison relatively difficult due to different data formats for WT and CFD different data formats for WT and CFD.

ЈUnified data format etc

CFD h d l ( t il

ЈUnified data format, etc.

- CFD has a secondary role (not necessarily d t d)

conducted).ЈPretest CFD always conducted.

Real-time comparison of surface pressure

- No data fusion between WT and CFD data Real-time comparison of surface pressure distribution between EFD and CFD

ЈEFD/CFD integration EFD/CFD integration

EFD (Analog WT) CFD (Digital WT) EFD (Analog WT) CFD (Digital WT)

Optimization of test planning,

technique, and model CFD considering both test model and wind tunnel 䋨Risk reduction, Data productivity improvement䋩

Virtual participation in

model and wind tunnel (wall, model support) Virtual participation in

WTT via internet WT wall&sting correction 䋨Accuracy improvement䋩 䋨Risk reduction, test efficiency

impro ement䋩 Design of Experiment

improvement䋩

Wind ‡4XDVLUHDOWLPH Tuning of CFD

Design of Experiment

䋨Test efficiency improvement) Wind

Tunnel

HigKVpeed reduction of

Tuning of CFD parameters EFD/CFD

comparison

Test

䋨Reliability improvement䋩

g p

2D/3D image measurement data

(Turb. model, grid, etc.) co pa so

‡9DOLGDWLRQGDWD

Automatic/adaptive

id ti

䋨Reliability improvement䋩 measurement data

䋨Data productivity improvement䋩

grid generation Data fusion considering

advantages and reliability of 䋨Data productivity improvement䋩

Fast CFD solver advantages and reliability of

EFD and CFD 䋨Reliability improvement䋩 Database of both EFD and CFD

Expanding the technology integrating experiment Expanding the technology integrating experiment

and numerical simulation to other fields

and numerical simulation to other fields

Digital/Analog Hybrid Wind Tunnel:

System Architecture

Digital/Analog Hybrid Wind Tunnel:

System Architecture

- User access via Web.

- Raw EFD/CFD data stored in a self-descriptive data format with metaRaw EFD/CFD data stored in a self descriptive data format with meta data stored in a XML-database

data stored in a XML database.

Userίᵨᵟᵶᵟὸ

Userίoutside JAXA) Userίᵨᵟᵶᵟὸ ᵟᶂᶋᶇᶌᶇᶑᶒᶐᵿᶒᶍᶐ ᵆᵨᵟᵶᵟὸ Userίoutside JAXA) ᵟᶂᶋᶇᶌᶇᶑᶒᶐᵿᶒᶍᶐᴾᵆᵨᵟᵶᵟὸ

Module Function

㪝㪆㪮

ᵵᵣᵠᴾᵫᶍᶂᶓᶊᶃ

ᵱᶗᶑᶒᶃᶋᴾᵫᵿᶌᵿᶅᶃᶋᶃᶌᶒᶗ ᶅ ᵳᶑᶃᶐᴾᶋᵿᶌᵿᶅᶃᶋᶃᶌᶒᶅ ᵡᶍᶌᶒᶐᶍᶊ

ᵢᵠᴾᵫᶍᶂᶓᶊᶃ ᵴᶇᶑᶓᵿᶊᶇᶘᵿᶒᶇᶍᶌᴾ

ᵫᶍᶂᶓᶊᶃ ᵡᵤᵢᴾᵫᶍᶂᶓᶊᶃ ᵣᵤᵢᴾᵫᶍᶂᶓᶊᶃ ᵡᵟᵢᴾᵫᶍᶂᶓᶊᶃ ᵲᶃᶑᶒᴾᵮᶊᵿᶌᶌᶇᶌᶅᵍ

ᵟᶌᵿᶊᶗᶑᶇᶑ ᵫᶍᶂᶓᶊᶃ ᵢᵠᴾ

ᵫᶍᶂᶓᶊᶃ

ᵧᶌᶒᶃᶐᶎᶍᶊᵿᶒᶇᶍᶌ ᵮᶐᶃᶒᶃᶑᶒᴾᵡᵤᵢ ᵢᵿᶒᵿ

ᵟᶌᵿᶊᶗᶑᶇᶑᴾᵫᶍᶂᶓᶊᶃ ᵭᶎᶒᶇᶋᶇᶘᶇᶌᶅᴾ ᵢᵿᶒᵿᴾ

ᵫᵿᶌᵿᶅᶃᶋᶃᶌᶒ ᶎ

ᵢᶃᶒᵿᶇᶊᶃᶂᴾᵡᵤᵢ

ᵢᵿᶒᵿᴾ ᵲᶐᵿᶌᶑᶄᶍᶐᶋᵿᶒᶇᶍᶌ

ᶎ ᶅ

ᵲᶃᶑᶒᴾᵫᶍᶂᶃᶊ ᵰᶃᶂᶓᶁᶒᶇᶍᶌ

ᵤ ᶒ

ᵢᵠᴾ ᵡᶍᶋᶎᵿᶐᶇᶑᶍᶌ

ᵡᶍᶌᶄᶇᶅᶓᶐᵿᶒᶇᶍᶌ ᵭᶎᶒᶇᶋᶇᶘᶇᶌᶅᴾ ᵲ ᶒ ᵡ ᶂᶇᶒᶇ ᵤᶍᶐᶋᵿᶒᴾ

ᵤᶍᶐᶋᵿᶒᴾ ᵲᶐᵿᶌᶑᶄᶍᶐᶋᵿᶒᶇᶍᶌ ᵰᶃᶅᶇᶑᶒᶐᵿᶒᶇᶍᶌ

ᵴᶇᶑᶓᵿᶊᶇᶘᵿᶒᶇᶍᶌ

ᵫᶍᶂᶇᶄᶇᶁᵿᶒᶇᶍᶌ ᵲᶃᶑᶒᴾᵡᶍᶌᶂᶇᶒᶇᶍᶌᴾ ᵲᶐᵿᶌᶑᶄᶍᶐᶋᵿᶒᶇᶍᶌ

ᵱᶃᵿᶐᶁᶆ

ᵰ ᶒ

ᵟᶓᶒᶍᶋᵿᶒᶇᶁᴾ

ᵥᶐᶇᶂ ᵥᶃᶌᶃᶐᵿᶒᶇᶍᶌ ᵭᶎᶒᶇᶋᶇᶘᶇᶌᶅᴾ ᵮᵿᶐᵿᶋᶃᶒᶃᶐᶑ

ᵰᶃᶎᶍᶐᶒ ᵥᶐᶇᶂᴾᵥᶃᶌᶃᶐᵿᶒᶇᶍᶌ ᵮᵿᶐᵿᶋᶃᶒᶃᶐᶑ

ᵤᶇᶊᶃ ᵫᶍᶂᶓᶊᶃ ᵤᶇᶊᶃᴾᵫᶍᶂᶓᶊᶃ

ᵥᶃᶌᶃᶐᵿᶒᶇᶍᶌᴾᶍᶄᴾᵫᶃᶒᵿᴾᵢᵿᶒᵿ ᵠᵿᶁᶉᶓᶎᵍᵰᶃᶑᶒᶍᶐᶃ

Fastness Optimization of model and

Fastness experimental planning by pretest CFD Generalized civil transport model: DLR F6

Solution: Automatic grid generator model: DLR-F6

+ New high-speed solver (FaSTAR)

- Accurate correction of interaction by ll d ti i t ti

wall and sting interaction

M t b bl h t i ti

Accuracy

- Most probable aero. characteristics prediction based on both EFD/CFD

y

prediction based on both EFD/CFD Solution: TAS* code (an existing RANS code) or FaSTAR

*TAS: Tohoku University Aerodynamics Simulation Solution: TAS code (an existing RANS code) or FaSTAR

TAS: Tohoku University Aerodynamics Simulation

„Status: Automatic grid generator usingg g g Cartesian grid technique and FaSTAR

9Grid around an airplane model with 10⁷grid points

g q

9Grid around an airplane model with 10 grid points

can be constructed within ten min using a regular PC Internal Flow Analysis

can be constructed within ten min. using a regular PC.

9The capability of the grid generator has been 9The capability of the grid generator has been

expanded to internal flow calculation expanded to internal flow calculation.

inlet

9In development of FaSTAR, a NS solver for laminar ^inlet

outlet

flow has been completed. ^outlet

High-speed data reduction of imaging techniques High speed data reduction of imaging techniques

„Objective:to improve data productivity of Analog WT by accelerating

„Objective:to improve data productivity of Analog WT by accelerating data reduction for PIV and PSP images

data reduction for PIV and PSP images.

Ex) Present PIV system

Ex)

f ^Issues

Acceleration of

Several hours for Issues

PIV data reduction Several hours for data reduction data reduction

Image acquisition C_Camera Image acquisition

computer

A l i (PC l )

Acceleration system (PC cluster) Commercial PIV data processing software

Accelerator

Acceleration of Accelerator䋺

Cell(High performance

Acceleration of commercial software

Preliminary test using _Cell(High performance

CPU developed by Preliminary test using

Cellshows a CPU developed by

IBM/Toshiba/Sony) is Accelerator

Cellshows a

possibility of ten-times IBM/Toshiba/Sony) is

the most promising (Cell, GPGPU, etc.)

possibility of ten times

faster data reduction. ^{p g}

candidate.

Several minutes for Goal

faster data reduction.

Several minutes for data reduction

- Objectives䋺

䊶Supporting rapid evaluation of the WT data during the test

䊶Clarifying technical problems of both EFD/CFD 䊶Clarifying technical problems of both EFD/CFD

1-D 2-D 3-D

EFD CFD

-1.5

Pressure Transducer

PSP EFD

(PSP)

-1 CFD

PSP

(PSP)

-0.5

Cp

0

0.5

0 0.2 0.4 0.6 0.8 1

x/C x/C

Chordwise press. distribution Surface press. distribution Spatial velocity field (PIV) Chordwise press. distribution Surface press. distribution Spatial velocity field (PIV)

- Qualitative comparison䋺 Feature extraction methods such as edge detection,

t l t t hi f t ti fl f t ( t ti h k t )

template matching for extracting flow features (vortex, separation, shock, etc.)

Shock wave

Original after edge detection Shock wave

Original after edge detection

Edge detection for a Schlieren image

EFD/CFD Integration Techniques

EFD/CFD Integration Techniques

‡ Data fusion techniques

‡ Data fusion techniques

P di ti t b bl h t i ti b d b th EFD/CFD

- Predicting most probable aero. characteristics based on both EFD/CFD.

Ex: SFA (Sequential Function Ex: SFA (Sequential Function Approximation) Neural Network

CFD Approximation) Neural Network

used to drag coeff. of SC1085

Neural Network used to drag coeff. of SC1085

wing section (Meade 2004) Neural Network

estimation

g ( )

estimation EFD

EFD

‡ ()'ЈCFD

- Optimum selection of turbulence modelOptimum selection of turbulence model

- Refinement of computational gridRefinement of computational grid ^l

modelcemulencTurbuT

‡ CFD ЈEFD

‡ CFD ЈEFD

O f

- Accurate correction of wind tunnel

ll/ t i t ti i CFD

- Optimization of test model

d i d WT t t l i

wall/support interaction using CFD including model and WT

design and WT test planning based on pretest CFD

including model and WT. based on pretest CFD.

Ex: Procedure of sting interaction correction Ex: Procedure of sting interaction correction

Ex: Utilization of response surface method in selection

of WT test conditions C_L,C_D,L/D䊶䊶䊶

M d l Model

M Į,ȕ M

Dassult Co., Sting support ,

AIAA Paper 2008-835

Summary Summary

„ -$;$ K W W G WK G O W I ³Di it l/A l H b id Wi d

„ JAXA has started the development of³Digital/Analog Hybrid Wind Tunnel´DVDSURWRW\SHUHDOL]LQJ()'&)'LQWHJUDWLRQ7KHLQLWLDO system will be compOHWHGLQy p

„ M i t h i l d l t it

„ Main technical development items䋺

9()'&)' LQWHJUDWLRQ WHFKQLTXHV FRPSDULVRQYLVXDOL]DWLRQ 9()'&)'integration WHFKQLTXHV FRPSDULVRQYLVXDOL]DWLRQ,

correction of wall/support interaction data fusion etc ) correction of wall/support interaction, data fusion, etc)

9Automatic grid generation and high-speed solver for DigLWDO:7g g g p g 9A l ti f i d t d ti f d i

9Acceleration of image data reduction for aerodynamic PHDVXUHPHQWVXFKDV3,9

„ ()'&)' LQWHJUDWLRQ WHFKQRORJ\ LV D NH\ IRU LQFUHDVLQJ WKH YDOXH

„ ()'&)' integration technology is a NH\ for increasing the value RI WKH + EULG :LQG 7 QQHO :H LOO WU WR PDNH ULJRUR V

of the Hybrid Wind Tunnel We will try to PDNH rigorous

ll b i i h d i d i d i f hi

collaborations with academia and industries for thiVSXUSRVH

Experimentalists :p

S. Kuchi-Ishi, H. Kato, S. Koike, T. Hirotani, and M. Kohzai., , , , Wind Tunnel Technology Center ARD JAXA

Wind Tunnel Technology Center, ARD, JAXA

CFD Researchers : CFD Researchers :

T. Aoyama, K. Murakami, A. Hashimoto, N. Fujita, and Y. Matsuo T. Aoyama, K. Murakami, A. Hashimoto, N. Fujita, and Y. Matsuo

Numerical Analysis Group ARD JAXA Numerical Analysis Group, ARD, JAXA

Y Y k k Y. Yokokawa

Civil Transport Team, APG, JAXAp , , Ryoyu Systems Co Ltd Ryoyu Systems Co., Ltd.

QUATRE-i Science

Reducing Uncertainty in EFD and CFD Through Data/Model Fusion

JAXA Workshop

February 23, 2009

Andrew Meade

Rice University, Houston, Texas, USA

Acknowledgements

2

Postdocs: B. Zeldin, Michael Kokkolaras,

Graduate Students: Jose Navarrete, Wei Wang, and David Thomson NASA Ames Rotorcraft Branch

As a graduate student working at NASA Ames Research Center I was well exposed to both the EFD and CFD communities.

At the time, they didn’t always get along but we were on the same team.

3

Despite 30 years of advances in CFD, we now understand that it is nothing more than

the third approach. CFD synergistically complements pure theory and experiments but it can never replace either of these other approaches.

Background

Figure: Notional NASA air breathing hypersonic aircraft design

The future advancement of fluid dynamics will rest upon a proper balance of all three methods.

Design of future aircraft and spacecraft will require even greater coupling between physical disciplines and better fidelity of their respective models (e.g., hypersonic aircraft)

5

Inlet Compression Lift

Pitching Moment Nozzle Thrust

Lift Pitching Moment Airframe

Fuel Payload

Strong interactions between vehicle components

Aerodynamics, propulsion, control, structure, tank, thermal protection, etc.

Highly integrated engine and airframe Much

x x x x

Hypersonic Aircraft Example :

of vehicle is engine inlet / nozzle

Large propulsive lift and pitching moments – strong contributor to trim, stability & control Large Mach number and dynamic pressure variations in flight

Severe x

x

x aerodynamic heating

Thermal protection must be integrated with structure

High fuel mass fraction required - majority of volume accommodates fuel x

x

Background

6

We can develop a framework by which this synergy is accessible through the use of tools from scattered data approximation machine learning tools and Generalized Regularization (GR)

[A.N. Tikhonov and V.Y. Arsenin, , 1977]

[M. Ulbrich, TUM-M9810, 1998]

Solution of Ill - Posed Problems

7

Specifically, mathematical analysis of experimental data is treated as an ill-posed problem.

Its regularization involves the introduction of additional information regarding the physical system.

We can then utilize a-priori mathematical models of physical systems at appropriate orders of fidelity for regularization. We can then investigate its potential in reducing uncertainty in EFD and CFD.

Background

What are some of the potential applications of a fusion of mathematical models, computational methods, and experimental data?

Filling in the blanks in experiments (including reducing uncertainx ty in EFD)

Accelerating through test matrices (steering and predicting the amount of data needed) Accelerating CFD solutions using data (including reducing uncertainty in CFD) Noi

x x

x se filtering of data

Knowledge discovery through parameter estimation x

9

[Kurt Long, NASA Ames] [Gloria Yamauchi, NASA Ames] [Gary Sivak, Air Force Research Lab]

Approach

^{1, ,}

th

To begin, define:

denotes the dimensionality of the approximation problem , , = dimensional input

basis function

coefficients corresponding

i i d i

k k

d

x x x d

x k

c I

{

{ }

{

{ to the basis functions

( ) exact response of an unknown or underlying physical process ( ) observable output

random noise of measurements at coordinate ( )

i

e i

u x

f x F u f

x f x

P {

{ Ÿ

{

experimental data points ( ) ( ) approximation of ( ) ( )

The function is designed to pass approximately through the experimental data points ( )

i i

a a k k

k

a a

e i

f x

u x u x u x x c

F u f

f x

P

{

{ Ÿ ¦)

10

²

1

Basic Problem:

The construction of a response surface from the minimization of the standard square error, ( )

constitutes an

s

e i a i

i

f x F u x H ¦ª¬ º¼

ill-posed problem since u_a is nonunique and sensitive to the noise in data . f_e

11

-3 -2 -1 0 1 2 3

-2 0 2 -6 -4 -2 0 2 4 6

y x

u(x,y)

-3 -2 -1 0 1 2 3

-2 0 2 -5 0 5

y x

u(x,y)

Approach

Solution:

The general idea in GTR is the minimization of

is a scalar measure of the agreement between and _e

A B

A f F u

H /

. In Bayesian terms is related to

knowledge.

is a stabilizing functional (i.e. regularizing operator) and is related to information.

is a positive scalar known as the regu

a A

a posteriori

B a priori

/ larization that controls the relevance of

the a posteriori and a priori information to the approximation . u_a

Generalized Tikhonov Regularization

Tikhonov Regularization

Bayesian Estimation

The regularizing operator, , is classified as either quantitative or qualitative.

Can range from systems of time-dependent nonlinear

A B

B

H /

partial differential equations, to statistical correlations, to heuristics.

Finding an approximate solution reduces to (a) finding regularization operators and (b) determining the regularization par

B

ameter from supplementary information pertaining to the problem, e.g., pertaining to the noise level in .

is not unique and depends on the type of physical process and the data type.

fe

B

/

13

Approach

In this context, the GTR formulation acts as a Swiss Army knife and gives us access to a number of popular methods in inverse problems.

Popular inverse problem methods include:

Regularization by no x

²

1

ise filtering Regularization by projection

Conventional Tikhonov regularization ( ) and is a

smoothing functional in variational form.

Scattered data approximation

s

e i a i

i

A f x F u x B

x

x Ÿ ª¬ º¼

x

¦

for Response Surface Modeling RSM includes support vector machines, radial basis functions, artificial neural networks

Kalman Filter Bayesian Estimation

x x

14

With 2 log | , 2 log , and 1, then becomes the figure-of-merit function from Bayesian estimation theory.

Therefore, one can think of the GTR framework as a

a e a

A p f f B p f

determ H

/

Bayesian Estimation :

form of Bayesian estimation.

Assume the measurement errors are Gaussian. The probability density functions act as if a differential operator is used to form B and operates on

inistic

u Kalman Filter :

0 0

0 0

, where is the solution to a mathematical model of the time-dependent physical system of interest. The solution

to the resulting Euler-Lagrange equation is , where ,

a

a cor cor k e k k

u u

u u u u t F u t u t

0 ²

. The basis functions for are the Green's function to a dynamic equation and is

computed to minimize during the time marching.

cor

a

u

u u

W W

¦

15

Approach

0

Back to Generalized Tikhonov Regularization

, and 0,

One approach to determining the correct value for is assuming the error is random,

a a e

f x f x f x f x

/ o f o / o o

/

> @

> @

2 2 2 2

1 1 1

statistically independent and uniformly distributed in the interval , . This will give,

( ) ( ) Var

3

where denotes the operat

s s s

e i a i e i a i i

i i i

f x F u x E f x F u x E s s

E

W W

P P W

ª º ª º

˜

ª º ª º

¬ ¼ « ¬ ¼ » « »

¬ ¼ ¬ ¼

¦ ¦ ¦

or of mathematical expectation.

Otherwise you can determine such that Max ( )

Note in both formulations, is solved for iteratively and requires the solution of ( ) within each itera

e i a i

a

f x F u x

f x W

/ d

/ tion.

The direct solution of is still an area of active research./