Efficient Multidisciplinary Analysis and Optimization of Hypersonic Vehicles Using Multi-Fidelity Surrogate Models
Efficient Multidisciplinary Analysis and Optimization of Hypersonic Vehicles Using Multi-Fidelity Surrogate Models
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20250211153050
- ISBN
- 9798346380337
- DDC
- 790
- 서명/저자
- Efficient Multidisciplinary Analysis and Optimization of Hypersonic Vehicles Using Multi-Fidelity Surrogate Models
- 발행사항
- [Sl] : Stanford University, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 123 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-05, Section: B.
- 주기사항
- Advisor: Alonso, Juan.
- 학위논문주기
- Thesis (Ph.D.)--Stanford University, 2024.
- 초록/해제
- 요약The trade between computational cost and model accuracy is a fundamental challenge in engineering design: models of higher fidelity (i.e. physical accuracy) typically require additional cost. Resource constraints fundamentally limit the amount of high-fidelity information available to engineers during design cycles, often requiring decisions to be made with incomplete information. Simply substituting in simplified, low-fidelity tools may guide the system to a non-optimal or infeasible configuration, introducing risk in the design process.The design of hypersonic glide vehicles exemplifies these challenges. The challenges involved with experimental testing place a strong reliance on numerical models for design insight, but com- putational fluid dynamics simulations of aerodynamic and aerothermal conditions, capturing all relevant physics, are computationally expensive. Furthermore, the tight coupling between trajec- tory performance and vehicle configuration necessitates analysis at a wide range of flight conditions, exacerbating computational cost by requiring a large number of simulations. This thesis examines the use of multi-fidelity modeling strategies to reduce the cost of multidisciplinary analysis and op- timization of hypersonic vehicles while retaining a level of accuracy in results consistent with the highest level of fidelity.A multi-fidelity framework for aerodynamic and aerothermal modeling of hypersonic vehicles is introduced and applied to the simulation of a hypersonic glide vehicle. An integrated, low- fidelity modeling framework for parametric geometries, SHARPE, is developed, providing comparable predictions to the SU2 computational fluid dynamics solver for hypersonic conditions, but with significantly reduced computational cost. Predictions from these tools are used to construct surrogate models using multi-fidelity Gaussian process regression. The resulting surrogate models are then used in the trajectory simulation of a notional hypersonic glide vehicle, and the impact of surrogate accuracy on trajectory performance is examined.Given the observation that realized trajectories comprise a small subset of the vehicle state space, strategies for using trajectory information to improve sampling efficiency are explored. A methodology for sampling based on sequentially refined estimates of the true trajectory is developed, and shown to reduce range prediction error relative to a uniform sampling policy. The sensitivity of trajectory quantities to aerodynamic parameters is efficiently computed using the adjoint equations of the vehicle dynamics. A sampling algorithm combining trajectory and sensitivity information is presented and applied to aerodynamic surrogate models, resulting in consistently more accurate range predictions than other policies examined.Finally, multi-fidelity aerodynamic and aerothermal surrogate models trained over a joint ve- hicle design-state-control space are integrated into a simple hypersonic glide vehicle optimization framework. Multi-fidelity predictions of range performance and thermal protection system sizing are shown to be substantially modified from low-fidelity alone predictions by employing a small amount of multi-fidelity data.
- 일반주제명
- Design optimization
- 일반주제명
- Heat
- 일반주제명
- Reynolds number
- 일반주제명
- Geometry
- 일반주제명
- Normal distribution
- 일반주제명
- Altitude
- 일반주제명
- Design
- 일반주제명
- Fluid mechanics
- 기타저자
- Stanford University.
- 기본자료저록
- Dissertations Abstracts International. 86-05B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008250123s2024 us c eng d■001000017164813
■00520250211153050
■006m o d
■007cr#unu||||||||
■020 ▼a9798346380337
■035 ▼a(MiAaPQ)AAI31643309
■035 ▼a(MiAaPQ)Stanfordfw426xc4025
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a790
■1001 ▼aNeedels, Jacob Troy.
■24510▼aEfficient Multidisciplinary Analysis and Optimization of Hypersonic Vehicles Using Multi-Fidelity Surrogate Models
■260 ▼a[Sl]▼bStanford University▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a123 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-05, Section: B.
■500 ▼aAdvisor: Alonso, Juan.
■5021 ▼aThesis (Ph.D.)--Stanford University, 2024.
■520 ▼aThe trade between computational cost and model accuracy is a fundamental challenge in engineering design: models of higher fidelity (i.e. physical accuracy) typically require additional cost. Resource constraints fundamentally limit the amount of high-fidelity information available to engineers during design cycles, often requiring decisions to be made with incomplete information. Simply substituting in simplified, low-fidelity tools may guide the system to a non-optimal or infeasible configuration, introducing risk in the design process.The design of hypersonic glide vehicles exemplifies these challenges. The challenges involved with experimental testing place a strong reliance on numerical models for design insight, but com- putational fluid dynamics simulations of aerodynamic and aerothermal conditions, capturing all relevant physics, are computationally expensive. Furthermore, the tight coupling between trajec- tory performance and vehicle configuration necessitates analysis at a wide range of flight conditions, exacerbating computational cost by requiring a large number of simulations. This thesis examines the use of multi-fidelity modeling strategies to reduce the cost of multidisciplinary analysis and op- timization of hypersonic vehicles while retaining a level of accuracy in results consistent with the highest level of fidelity.A multi-fidelity framework for aerodynamic and aerothermal modeling of hypersonic vehicles is introduced and applied to the simulation of a hypersonic glide vehicle. An integrated, low- fidelity modeling framework for parametric geometries, SHARPE, is developed, providing comparable predictions to the SU2 computational fluid dynamics solver for hypersonic conditions, but with significantly reduced computational cost. Predictions from these tools are used to construct surrogate models using multi-fidelity Gaussian process regression. The resulting surrogate models are then used in the trajectory simulation of a notional hypersonic glide vehicle, and the impact of surrogate accuracy on trajectory performance is examined.Given the observation that realized trajectories comprise a small subset of the vehicle state space, strategies for using trajectory information to improve sampling efficiency are explored. A methodology for sampling based on sequentially refined estimates of the true trajectory is developed, and shown to reduce range prediction error relative to a uniform sampling policy. The sensitivity of trajectory quantities to aerodynamic parameters is efficiently computed using the adjoint equations of the vehicle dynamics. A sampling algorithm combining trajectory and sensitivity information is presented and applied to aerodynamic surrogate models, resulting in consistently more accurate range predictions than other policies examined.Finally, multi-fidelity aerodynamic and aerothermal surrogate models trained over a joint ve- hicle design-state-control space are integrated into a simple hypersonic glide vehicle optimization framework. Multi-fidelity predictions of range performance and thermal protection system sizing are shown to be substantially modified from low-fidelity alone predictions by employing a small amount of multi-fidelity data.
■590 ▼aSchool code: 0212.
■650 4▼aDesign optimization
■650 4▼aHeat
■650 4▼aReynolds number
■650 4▼aGeometry
■650 4▼aNormal distribution
■650 4▼aAltitude
■650 4▼aDesign
■650 4▼aFluid mechanics
■690 ▼a0389
■690 ▼a0204
■71020▼aStanford University.
■7730 ▼tDissertations Abstracts International▼g86-05B.
■790 ▼a0212
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17164813▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


