보고서 정보
주관연구기관 |
한국과학기술원 Korea Advanced Institute of Science and Technology |
보고서유형 | 최종보고서 |
발행국가 | 대한민국 |
언어 |
한국어
|
발행년월 | 2016-07 |
과제시작연도 |
2015 |
주관부처 |
미래창조과학부 Ministry of Science, ICT and Future Planning |
등록번호 |
TRKO201600010607 |
과제고유번호 |
1711029002 |
사업명 |
첨단융합기술개발 |
DB 구축일자 |
2016-10-22
|
키워드 |
익형.날개 설계.수학적 최적화 알고리즘.설계 변수의 민감도 해석.구배율 기반 최적화 알고리즘.비구배율 기반 최적화 알고리즘.수반법.근사모델.최적설계 프레임워크.Airfoil and Wing design.Mathematical optimization algorithm.Sensitivity analysis.Gradient-based optimization.Gradient-free optimization.Adjoint solution method.Surrogate model.Design optimization framework.
|
DOI |
https://doi.org/10.23000/TRKO201600010607 |
초록
▼
유체역학을 기반으로 한 전산유체역학 기법의 발전에 따라, 이를 활용하는 설계 기법에 대한 연구가 활발하다. 비단 연구영역 뿐 만 아니라, 산업계 전반에서 전산 실험 및 최적화 알고리즘까지 포함하는 전산 최적 설계에 대해 관심이 매우 높아지고 있다. 때문에 고등교육 기관에서 전산 최적 설계에 대한 지식을 갖춘 인력을 양성하는 것은 사회적 요구에 부응하는 일이며, 아직 열악한 관련 공학 교육 인프라의 구축에 대한 노력이 필요시 되는 상황이다. 본 연구 사업에서는 고급 인력 양성을 통한 국가 경쟁력 확보 제고의 차원에서 제안된 항공우주공
유체역학을 기반으로 한 전산유체역학 기법의 발전에 따라, 이를 활용하는 설계 기법에 대한 연구가 활발하다. 비단 연구영역 뿐 만 아니라, 산업계 전반에서 전산 실험 및 최적화 알고리즘까지 포함하는 전산 최적 설계에 대해 관심이 매우 높아지고 있다. 때문에 고등교육 기관에서 전산 최적 설계에 대한 지식을 갖춘 인력을 양성하는 것은 사회적 요구에 부응하는 일이며, 아직 열악한 관련 공학 교육 인프라의 구축에 대한 노력이 필요시 되는 상황이다. 본 연구 사업에서는 고급 인력 양성을 통한 국가 경쟁력 확보 제고의 차원에서 제안된 항공우주공학의 유체 교육 분야의 소프트웨어 개발 사업의 일환으로, 전산 최적 설계 기법에 대한 교육용 소프트웨어 및 콘텐츠 개발을 수행할 것이다. 최종적으로 개발 시뮬레이션 프로그램과 교육 콘텐츠를 기계/항공우주/조선/토목 분야 27개 강좌, 1,440명 이상 활용할 수 있도록 한다.
Abstract
▼
Ⅲ. Contents, Scope of the Research, and Results
Major components of fluid dynamics course are the understanding of various flow analysis methods ranging from simple analytical techniques to complex CFD-based algorithms. Also, the optimization method these days is systematically applied to the mod
Ⅲ. Contents, Scope of the Research, and Results
Major components of fluid dynamics course are the understanding of various flow analysis methods ranging from simple analytical techniques to complex CFD-based algorithms. Also, the optimization method these days is systematically applied to the modern aircraft design, developing educational tools for the students to use for their own design problem such as low-drag airfoil and wing are particularly important. Our research group conducted a research on developing the educational design softwares that can be directly used during the classes.
1) In the first year, various types of sensitivity analysis methodology and basic gradient-based optimization algorithm was developed. Also, using EDISON_CFD, which is currently serving, the framework which provides the environment that users can perform the design optimization of an airfoil and the geometry of wings was developed. For the combination into a framework, the development of extra softwares, like geometric kernel were done along at the same time.
Moreover, for the educationally high-applicable subjects, like the basic concepts of design optimization, and sensibility analysis, contents for users’ understanding were developed.
2) In the second year, software and contents which are related to design optimization methodologies based on approximation modeling were developed. Especially, genetic algorithm, which is one of gradient-free algorithms, and multidisciplinary design optimization algorithms were developed. These algorithms enabled users to solve various design problems in various ways.
3) In the third year, the interface of the e-Science based numerical design optimization framework - EDISON_Airfoil_Design which was developed in the first stage - was updated.
The design framework was also updated to handle a new type of the shape function. In order to solve the practical design problems in the engineering field, the EDISON_flow_path was developed based on the framework only used for the airfoil design, aiming at optimization of flow path to minimize the pressure loss. The advanced content including flow path design problem was made and on the service on the portal. In addition to the gradient-based optimization algorithm, the gradient-free optimization algorithm was implemented and the one of content was dedicated to compare the performances of both optimization algorithms.
4) In the fourth year, the further update was done on the EDISON_Airfoil_Design to use total three of the shape functions. The surrogate model can be utilized and enhance the performance of the global design framework and the Kriging model - one of representative surrogate model – was implemented. Indirect and a direct Co-Kriging models which were developed are state-of-art Kriging models to enhance the accuracy of Kriging model. It is hard to find the Co-Kriging serviced publically. The global design framework was established using the various softwares previously developed in our research group, and it is expected to satisfy the high level of educational/research requirement of the graduated students.
5) In the fifth year, a sensitivity analysis solver based on discrete adjoint method was developed. Using the software, the computational cost is independent of the number of design variable, so it can be applied to the detail design of airfoil and/or wing. Because the adjoint sensitivity solver is based on an unstructured grid, a pre-processing software which converts a structured grid from eMEGA to a unstructured grid was developed. As a compatible design framework with the EDISON_Airfoil_Design which uses the gradient-based optimizer, the various features were included to the transonic airfoil design framework based on the gradient-free design optimizer: EDISON_Design. So, users can select an appropriate optimizer depending on the optimization problem. In addition, a flow analysis software to investigate the performance of compressor and turbine was developed to support industry field. The contents for introduction of the adjoint analysis method and detailed design problems of airfoil and wing were made. The contents for the multidisciplinary design optimization and robust design optimization were made. The rest one content was dedicated for the study about the local optimization characteristics of gradient-based design optimizations.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.