JAXA Supercomputer System Annual Report April 2016-March 2017

Report Number: R16E0103

• Responsible Representative: Yuichi Matsuo(Aeronautical Technology Directorate, Numerical Simulation Research Unit)
• Contact Information: Keiichi Murakami(murakei@chofu.jaxa.jp)
• Members: Yuki Matsui
• Subject Category: Basic Research(CFD)

Abstract

In order to obtain a low computation cost and stable high resolution solution for the aircraft Rudimentary Landing Gear (RLG), the CFD solver FaSTAR developed by JAXA is used as the central difference base, and in the place where the calculation is unstable, the numerical viscosity of the upwind difference To extend. As a result, the resolution in the high frequency region was improved.

Goal

Aim for the construction of numerical simulation method which can perform aerodynamic analysis with high resolution by using the unstructured CFD code 'FaSTAR' (developed by JAXA).

Objective

Suppress decay of physical quantity fluctuation due to numerical viscosity in high frequency region, obtain high resolution and stable solution

References and Links

N/A

Use of the Supercomputer

Computation analysis of unsteady aerodynamics

Necessity of the Supercomputer

Since unsteady aerodynamic analysis requires a huge amount of time for calculation, a high-speed computing environment is required.

Achievements of the Year

In order to obtain a low computation cost and stable high resolution solution for the aircraft Rudimentary Landing Gear (RLG).The non-viscous flux calculation is set to the second order central difference base, only the place where the calculation becomes unstable by the central difference is detected, the hybrid scheme which stabilizes the calculation is added by adding the numerical viscosity of the second order upwind difference, We introduced the development CFD solver FaSTAR developed by JAXA and conducted unsteady aerodynamic analysis. As a result, as shown in Fig.1, when visualizing the vortex of Hybrid scheme and Roe scheme constructed this time on the isosurface of Q value, the Hybrid scheme is more finely resolved than the Roe scheme. The pressure spectrum of the object surface is confirmed in Fig. 2, indicating that the Hybrid schemesuppresses the decay of SPL in the high frequency range.

From this, it is found that the non-viscous flux calculation is a form in which the numerical viscosity of the Roe scheme is added to the second order central difference base, and by increasing the numerical viscosity of the Roe scheme only in the point where the calculation is unstable, By using the controlled hybrid scheme, we obtained the knowledge that attenuation of fluctuation of the physical quantity in the high frequency region due to numerical viscosity issuppressed, and high resolution and stable solution can be obtained.

Fig.1:Instantaneous visualization of Q

Fig.2:FFT analysis

Publications

N/A

Computational Information

• Parallelization Methods: Process Parallelization
• Process Parallelization Methods: MPI
• Thread Parallelization Methods: n/a
• Number of Processes: 1024
• Number of Threads per Process: 1
• Number of Nodes Used: 1
• Elapsed Time per Case (Hours): 300
• Number of Cases: 6

Resources Used

Total Amount of Virtual Cost(Yen): 1,655,811

Breakdown List by Resources

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