Collaborative work for aerodynamic optimization
JAXA Supercomputer System Annual Report April 2016-March 2017
Report Number: R16E0036
- Responsible Representative: Takeshi Ito(Aeronautical Technology Directorate, Next Generation Aeronautical Innovation Hub Center)
- Contact Information: Shigeru Kuchiishi(kuchi-ishi.shigeru@jaxa.jp)
- Members: Shigeru Kuchiishi, Minoru Yoshimoto, Shinsuke Nishimura, Kei Nakanishi, Yukinori Morita, Takuya Ogura, Kyohei Sawada
- Subject Category: Aviation(Aircraft)
Abstract
A Multi-Objective Evolutionary Algorithm (MOEA) is employed as an aerodynamic optimization method and the optimization tool is aimed to enable the direct evolutionary computing to perform within a practical computational time by FaSTAR. In the present project, basic programs are developed and validated using JSS2. A series of programs are alsosupplied to the collaborative partners to set in their optimization tools and examine how much speed-up is realized by FaSTAR.
Goal
To develop the fastest aerodynamic optimization tool in the word and serve to Japanese aircraft industries or research institutes.
Objective
To develop an aerodynamic optimization tool by using the unstructured CFD code FaSTAR and to examine its validity and efficiency.
References and Links
N/A
Use of the Supercomputer
Each of aerodynamic optimization processes; CFD grid generation, grid deformation, CFD analysis (sample calculation), and evolutionary computation, are performed using JSS2.
Necessity of the Supercomputer
Aerodynamic optimization using an evolutionary algorithm requires a number of high-fidelity and large-scaled computations (3D RANS analysis) and needs to use the supercomputer.
Achievements of the Year
Large scale multi-objective design optimization for commercial airplane using multi-fidelity CFD solvers including FaSTAR as high-fidelity solver was conducted by JSS2. The design concepts were extracted and it contributed to strengthening competitiveness for national commercial airplanes.
Publications
N/A
Computational Information
- Parallelization Methods: Process Parallelization,Thread Parallelization,Serial
- Process Parallelization Methods: MPI
- Thread Parallelization Methods: OpenMP
- Number of Processes: 96
- Number of Threads per Process: 16-96
- Number of Nodes Used: 1-3
- Elapsed Time per Case (Hours): 0.5
- Number of Cases: 10000-
Resources Used
Total Amount of Virtual Cost(Yen): 525,486
Breakdown List by Resources
System Name | Amount of Core Time(core x hours) | Virtual Cost(Yen) |
---|---|---|
SORA-MA | 245,993.07 | 404,597 |
SORA-PP | 12,041.42 | 102,809 |
SORA-LM | 0.00 | 0 |
SORA-TPP | 0.00 | 0 |
File System Name | Storage assigned(GiB) | Virtual Cost(Yen) |
---|---|---|
/home | 16.02 | 20 |
/data | 10,506.74 | 13,765 |
/ltmp | 3,276.91 | 4,293 |
Archiving System Name | Storage used(TiB) | Virtual Cost(Yen) |
---|---|---|
J-SPACE | 0.00 | 0 |
Note: Virtual Cost=amount of cost, using the unit price list of JAXA Facility Utilization program(2016)
JAXA Supercomputer System Annual Report April 2016-March 2017