GPGPU를 이용한 공간 다물체 시스템과 대량 입자들 사이의 접촉 해석 연구
- Alternative Title
- Study on Contact Analysis between Spatial Multibody System and Many Particles using GPGPU
- Abstract
- NVIDIA has been developing the GPGPU(General Purpose Graphics Processing Unit) for not only 3D graphics and image processing, but also signal processing, physical simulation and so on. Currently, a lot of studies carried out by using the parallel programing based GPU. A lot of time is required for solving the EOM of large system by using multi-body dynamics. If the parallel programing by GPU is applied to the multi-body dynamics, the analysis time can be improved.
In this study, parallel computing for collision analysis between 3D multi-body and many particles was carried out by using the CUDA programming. In order to detect collisions between multi-body and particle, grid-cell collision detection algorithm is employed. Hertzian normal and tangential sliding friction contact model are used to calculate the contact force.
The multi-body model consists of the particles and spring-damper force elements and spherical joints. To verify the accuracy, MSC ADAMS is used. To investigate the numerical efficiency for the multi-body dynamics, the sequential program and the parallel program were developed. In case of the multi-body dynamics, when the number of body and spring-damper is 280 and 360, respectively, computation time of the parallel program is faster about 2.17 times then the sequential program. In case of the DEM, when the number of particle is 10648, it shows the better efficiency about 100times. In case of the collision analysis, when the number of particle and spring-damper is 1280 and 360, respectively, it shows the better efficiency about 3times.
- Author(s)
- 전철웅
- Issued Date
- 2013
- Awarded Date
- 2013. 2
- Type
- Dissertation
- Publisher
- 부경대학교
- URI
- https://repository.pknu.ac.kr:8443/handle/2021.oak/24652
http://pknu.dcollection.net/jsp/common/DcLoOrgPer.jsp?sItemId=000001966030
- Alternative Author(s)
- Jun, Chul Woong
- Affiliation
- 부경대학교 대학원
- Department
- 대학원 메카트로닉스공학과
- Advisor
- 손정현
- Table Of Contents
- Abstract V
1. 서 론 1
1.1. 연구 배경 및 동향 1
1.2. 연구 목적 3
2. CUDA 병렬 프로그래밍 4
2.1. 그래픽 카드의 발전 4
2.2. CUDA 6
2.3. 병렬 프로그래밍 7
2.3.1. 병렬 프로그래밍 개발 환경 7
2.3.2. 스레드-블록-그리드 구조 8
2.3.3. CUDA 데이터의 병렬처리 10
2.3.4. CUDA 메모리 11
3. 공간 다물체 동역학 해석 프로그래밍 15
3.1. 운동방정식의 구성 15
3.1.1. Euler Parameters 15
3.1.2. 운동 방정식 17
3.1.3. HHT-I3 내재적 수치적분 18
3.2. 프로그램 검증 22
3.2.1. 검증 모델 정의 22
3.2.2. 해의 정확성 검증 24
4. 대량 입자간의 접촉 모델 29
4.1. 입자 모델링 29
4.1.1. 입자와 공간 모델링 29
4.1.2. 입자 동역학 시뮬레이션 30
4.2. 접촉 판별 알고리즘 32
4.3. 접촉력 계산 모델 33
4.3.1. Normal Force 33
4.3.2. Tangential Force 34
4.4 접촉 모델의 검증 35
4.4.1 접촉 검증 모델링 35
4.4.2 접촉력 검증 36
5. CUDA기반 해석 프로그램 39
5.1. CUDA기반 다물체동역학 해석 39
5.1.1. 병렬 다물체 동역학 해석법 39
5.1.2. 효율성 비교 40
5.2. CUDA기반 대량 입자 해석 43
5.2.1. 병렬 대량 입자 해석법 43
5.2.2. 효율성 비교 44
5.3. 통합 해석 45
5.3.1 통합 해석법 45
5.3.2 통합 해석 시뮬레이션 결과 47
5.3.3 통합 해석의 효율성 비교 56
6. 결 론 58
참고문헌 59
- Degree
- Master
-
Appears in Collections:
- 대학원 > 메카트로닉스공학과
- Authorize & License
-
- Files in This Item:
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.