서지주요정보
On the model reduction methods with free-interface substructuring for dynamic analysis = 동적 해석을 위한 자유단 경계 기반의 부구조를 갖는 모델 축소 기법
서명 / 저자 On the model reduction methods with free-interface substructuring for dynamic analysis = 동적 해석을 위한 자유단 경계 기반의 부구조를 갖는 모델 축소 기법 / Jeong-Ho Kim.
발행사항 [대전 : 한국과학기술원, 2018].
Online Access 원문보기 원문인쇄

소장정보

등록번호

8032312

소장위치/청구기호

학술문화관(문화관)B1층 보존서고

DME 18030

휴대폰 전송

도서상태

이용가능(대출불가)

사유안내

반납예정일

리뷰정보

초록정보

In response to the large and complex finite element models in practical engineering, the needs for studies on model reduction methods have been highlighted. Since 1960s, the model reduction methods have been actively studied for various problems such as computational efficiency of the reduction process, improvement of the accuracy of the reduction model. In this dissertation, the effective model reduction methods are proposed. The developed methods divide the entire finite element model into several substructures and consider the free-interface between neighboring substructures. In particular, the new component mode synthesis (CMS) method is provided by improving the accuracy of dual Craig-Bampton (DCB) method. The error estimation method for the DCB method is also proposed. For the degree of freedom based reduction method, the new dynamic condensation method with fully decoupled substructuring is proposed. Through the various numerical problems, the solution accuracy and computational efficiency of the present methods are demonstrated.

현대에 이르러 구조물에 대한 동적 응답 해석은 대형화, 복잡한 구조 설계 및 시공 조건 등으로 인해 어려움을 갖게 되었다. 유한요소 모델 구축 시 자유도가 매우 증가하기 때문에 구조해석을 위해서는 상당한 전산 시간이 소요된다. 이러한 이유로 모델 축소기법(model reduction method)에 대한 연구 필요성이 증가하고 있다. 모델 축소기법은 1960년대에 연구가 시작된 이후로 축소 절차의 전산 효율성, 축소모델의 정확도 개선, 주요 모드 또는 주 자유도의 선정, 부구조 간 경계조건의 처리 등 다양한 문제에 대하여 연구가 활발히 진행되고 있다. 본 연구에서는 이와 같은 주요 이슈를 해결하기 위하여 자유단 경계 기반의 부구조법을 적용한 새로운 축소기법을 개발하였다. 먼저, dual Craig-Bampton (DCB) 기법의 정확도를 향상시킨 새로운 부분구조합성법을 제안하였다. 또한 DCB 기법으로 얻은 축소 모델이 갖는 고유치의 신뢰도를 파악할 수 있는 정확한 오차 추정 기법을 제안하였다. 마지막으로 부구조 독립성을 갖는 자유도 기반 축소 기법을 개발하였다. 특히 개발된 기법은 실제의 설계 및 제작 절차에 맞게 조립을 통해 전체 구조물의 유한요소 모델을 얻는 시스템에서 효과적으로 활용될 수 있을 것으로 기대한다. 다양한 수치예제들을 통하여 개발된 기법의 성능을 검증하였다.

서지기타정보

서지기타정보
청구기호 {DME 18030
형태사항 vii, 144 p. : 삽화 ; 30 cm
언어 영어
일반주기 저자명의 한글표기 : 김정호
지도교수의 영문표기 : Phill-Seung Lee
지도교수의 한글표기 : 이필승
학위논문 학위논문(박사) - 한국과학기술원 : 기계공학과,
서지주기 References : p. 140-144
QR CODE

책소개

전체보기

목차

전체보기

이 주제의 인기대출도서

Partitioning ofglobal FE model and interface handling in the CB method (Ns =2).

Assemblage ofsubstructures and interface handling in the DCB method(N,=4).(a)

Reduction procedure ofIRS method. (a) global structural FE model (b) selection ofmaster nodes, (c) reduced model in the IRS method.

Flow chart for the FE model reduction in the improved DCB method

Rectangular plate problem: (a) Matchingmesh on the interface (12x6 mesh), (b) Non-matching mesh between neighboring substructures, (c) Interface boundary treatment.

Relative eigenfrequency errors in the rectangular plate problem with matching mesh

Number ofdominant modes used and number ofDOFsin original and reduced systems for the rectangular plate problem (12x6 mesh).

Relative eigenfrequency errors in the rectangular plate problem with non-matching mesh.

Number ofdominant modes used and number ofDOFsin original and reduced systems for the rectangular plate problem with non-matching mesh.

Plate structure with a hole.

Number ofdominant modes used and number ofDOFs in original and reduced systems for the plate structure with a hole.

Relative eigenfrequency errors in the plate structure with a hole.

MAC for reduced system in the plate structure with a hole: (a) DCB method, (b) Improved DCB

Hyperboloid shell problem.

Number ofdominant modes used and number ofDOFs in original and reduced systems for the hyperboloid shell problem.

Relative eigenfrequency errors in the hyperboloid shell problem.

MAC for reduced system in the hyperboloid shell problem: (a) DCB method, (b) Improved DCB method

Computational costs for the hyperboloid shell problem.

Bended pipe problem.

Number ofdominant modes used and number ofDOFsin the original and reduced systems fortl bended pipe problem.

Relative eigenfrequency errors in the bended pipe problem.

MAC forreduced system in the bended pipeproblem(Nd =15):(a)DCB method, (b)Improved

Computational costs for the bended pipe problem.

NACA 2415 wing with ailerons problem.

Number ofdominant modes used and number ofDOFsin the original and reduced systems fort] NACA2415 wing with ailerons problem.

Relative eigenfrequency errors in the NACA2415 wing with ailerons problem

MAC for reduced system by the improved DCB method in the NACA2415 wing with ailerons problem.

Cable-stayed bridgeproblem (1 substructure).

Connection of cable-stayed bridge substructures (Ns =2).

Relative eigenfrequency errors in the cable-stayed bridgeproblem(Ns =6).

MAC forreduced system in the cable-stayed bridgeproblem(Ns=6):(a)DCB method, (b Improved DCB method

Eigenvalues calculated for the plate with a hole. Negative eigenvalue

Relative eigenfrequency errors in the plate structure with a hole(Nd =4).

Rectangular plate problem: (a) Matchingmesh on the interface (12x6 mesh), (b) Non-matching mesh between neighboring substructures, (c) Interface boundary treatment.

Number ofdominant modes used and number ofDOFsin original and reduced systems for the rectangular plate problem.

Exact and estimated relative eigenvalue errors in the rectangular plate problem with matching mesh

Relative errors for the corrected eigenvalues in the rectangular plate problem with matching mesh

Exact and estimated relative eigenvalue errors in the rectangular plateproblem with non-matching mesh.

Relative errors for the corrected eigenvalues in the rectangular plate problem with non-matching mesh

Hyperboloid shell problem.

Number ofdominant modes used and number ofDOFsin original and reduced systems for the hyperboloid shell problem.

Exact and estimated relative eigenvalue errors in the hyperboloid shell problem.

Relative errors for the corrected eigenvalues in the hyperboloid shell problem.

Pipe intersection problem.

Exact and estimated relative eigenvalue errors in the Pipe intersection problem.

Relative errors for the corrected eigenvalues in the Pipeintersection problem.

Cable-stayed bridgeproblem (1 substructure).

Connection ofcable-stayed bridge substructures (Ns =2).

Number ofdominant modes used and number ofDOFsin original and reduced systems for the cable- stayed bridge problem (Ns =6).

Exact and estimated relative eigenvalue errors in the cable-stayed bridge problem (Ns =6).

Relative errors for the corrected eigenvalues in the cable-stayed bridgeproblem(N =6).

Flow chart for the FE model reduction

Rectangular plate problem with matching mesh: (a) Selected nodes in the original IRS method, (b) Selected nodes in the present method.

Number of master DOFs usedand number ofDOFs in original and reduced systems for th rectangular plateproblem (12x6 mesh).

Exact and approximated eigenvalues in the rectangular plate problem with matching mesh

Relative eigenvalue errors in the rectangular plate problem with matching mesh

Rectangular plate problem with non-matching mesh: (a) Non-matching mesh between neighboring substructures, (b) Selected nodes in the present method.

Relative eigenvalue errors in the rectangular plate problem with non-matching mesh

Number ofmaster DOFs used and number ofDOFsin original and reduced systems for the plate structure with a hole.

Selected nodes in the plate structure with a hole: (a) only interface nodes selected, (b) interface nodes and 8 interior nodes selected in each substructure.

Relative eigenvalue errors in the plate structure with a hole.

MAC forreduced system by the present method in the plate structure with a hole: (a) case 1, (b) case

Hyperboloid shell problem.

Number ofmaster DOFs used and number ofDOFs in original and reduced systems for the Hyperboloid shell problem.

Relative eigenvalue errors in the hyperboloid shell problem.

Eigenvalues calculated for the hyperboloid shell problem.

Bended pipe problem: (a) Global FE model without substructuring, (b) Matchingmesh on the interface, (c) Non-matching mesh between neighboring substructures.

Number ofmaster DOFsused and number ofDOFs in original and reduced systems for the bended pipe problem.

Relative eigenvalue errors in the bended pipe problem with matching mesh.

MAC for reduced system by the present method in the bended pipe problem with matching mesh

Computational costs for the bended pipe problem.

Relative eigenvalue errors in the bended pipe problem with non-matching mesh

MAC for reduced system by the present method in the bended pipe problem with non-matching mesh.

Wind turbine rotor problem.

Number ofmaster DOFs used and number ofDOFsin original and reduced systems for the wind turbine rotor problem.

Relative eigenvalue errors in the wind turbine rotor problem.

MAC for reduced system by the present method in the wind turbine rotor problem.

Computational costs for the wind turbine rotor problem.

NACA 2415 wing with ailerons problem.

Number ofmaster DOFs used and number ofDOFs in original and reduced systems for the N 2415 wing with ailerons problem.

Relative eigenvalue errors in the NACA2415 wing with ailerons problem

MAC for reduced system by the present method in the NACA2415 wing with ailerons problem

Cable-stayed bridgeproblem (1 substructure).

Connection ofcable-stayed bridge substructures (Ns =2).

Number ofmaster DOFs used and number ofDOFsin original and reduced systems for the cable- stayed bridgeproblem (Ns =6).

Relative eigenvalue errors in the cable-stayed bridgeproblem (Ns =6).

MAC for reduced system by the present methodin the cable-stayed bridgeproblem (N,=6):(a) case 1, (b)case 2.