JAXA Supercomputer System Annual Report February 2025-January 2026

Report Number: R25EFHC0319

Subject Category: Large-Scale Challenge

PDF (to be added)

• Responsible Representative: Naoyuki Fujita, Security and Information Systems Department, SuperComputer Division
• Contact Information: Takeshi Sawada, The University of Tokyo(sawada@flab.isas.jaxa.jp)
• Members: Akira Oyama, Takeshi Sawada

Abstract

Inflatable flexible aeroshells have attracted attention as high-efficiency aerodynamic deceleration devices that can serve as alternatives to parachutes. In STEP1 of the stepwise Mars landing exploration program, multiple aeroshells are planned to be deployed to demonstrate entry, descent, and landing (EDL) technologies, as well as to enable multi-point observations on the Martian surface.

To date, technical knowledge on aeroshells has been accumulated through both experimental and numerical studies. In particular, the subsonic flight regime has been identified as a critical condition where attitude instability occurs. In the suborbital free-flight experiment conducted in 2012 using a sounding rocket (SMAAC), attitude instability was observed in the subsonic regime. Furthermore, similar dynamic instabilities were reported in the RATS experiment in 2021 and the RATS-L experiment in 2023. Such dynamic instabilities may also occur during atmospheric entry into Mars. However, it is difficult to reproduce the Martian environment in ground-based experiments. Therefore, numerical simulations are essential for reproducing the attitude dynamics under Martian atmospheric conditions.

In addition, previous studies have reported that structural deformation due to the flexibility of the aeroshell reduces its static stability. In this study, we further focus on how such structural deformation affects the attitude motion and its dynamic stability.

Accordingly, the objective of this study is to numerically reproduce the attitude dynamics of a flexible inflatable aeroshell in the subsonic regime under Martian atmospheric conditions, and to clarify the effects of structural deformation on its dynamic stability.

Reference URL

N/A

Reasons and benefits of using JAXA Supercomputer System

To reproduce the attitude dynamics of an inflatable flexible aeroshell with structural deformation, it is necessary to develop a solver that simultaneously solves the governing equations of fluid, structure, and motion in a coupled manner. However, to the best of the authors' knowledge, no previous studies have reported such a fully coupled analysis, making this a novel and challenging problem. In the present study, a partitioned approach is adopted, in which individual solvers are used for each physical field. The coupling conditions are approximately satisfied by performing iterative calculations within the same time step. The fluid and motion solvers are based on the fluid–motion coupling functionality of FaSTAR-move, while the structural solver is the open-source finite element solver CalculiX. These solvers are coupled via the open-source coupling library preCICE, enabling a three-way coupled analysis. In this approach, approximately four to five sub-iterations are required within each time step on average. As a result, the computational cost becomes approximately four to five times higher than that of a standalone fluid simulation, making large-scale parallel computation essential.

Achievements of the Year

1. Computational Setup

The flow conditions are set to a Mach number of 0.3 and a Reynolds number of 1.0 x 105. The remaining conditions are determined based on a point-mass simulation of the Mars descent trajectory. The geometry of the aeroshell and the computational grid used in this study are shown in Fig. 1.

The diameter of the aeroshell is 3 m. The center of gravity is fixed as the rotation center, and a one-degree-of-freedom motion allowing only pitch rotation is considered. The moment of inertia about the rotation center is 7.06 kgm2. The angle between the capsule center axis and the freestream is defined as the angle of attack, alpha.

The fluid computation employs approximately 80 million grid cells. The computational grid consists of two overlapping components, a body-fitted grid surrounding the spherical body (shown in red in the figure) and a background grid. A rectangular refinement region is introduced in the wake to capture the recirculation region behind the body.

In this study, a Large Eddy Simulation (LES) is performed to resolve the vortical structures in the wake. The Smagorinsky model is used as the subgrid-scale model. The HLLEW scheme is applied to the convective terms, with second-order spatial accuracy achieved using MUSCL reconstruction. Time integration is carried out using a second-order LU-SGS method. The time step size is set to 2.23 x 10-4 s.

On the other hand, the structural computation uses approximately 17,000 elements. As boundary conditions, the region connected to the capsule is fixed, while fluid forces are applied on the outer surface of the membrane. On the inner surface, the average pressure between the fore and aft sides of the vehicle is uniformly distributed. The torus structure is treated as a rigid body without deformation.

The material properties of the membrane are assumed to be a Young's modulus of 700 MPa and a Poisson's ratio of 0.3, with a linear stress-strain relationship. The membrane is assumed to be homogeneous and isotropic. The time step used in the structural computation is identical to that used in the fluid computation.

2. Results and Discussion

s a preliminary study, the initial pitch angle was varied as 15, 30, 45, and 60 degrees, and the time evolution of the pitch motion was examined. The results show that the oscillation amplitude grows and becomes unstable at 60 degrees, whereas it decays for angles of 45 degrees and below. This indicates that the boundary between dynamically stable and unstable behavior under the present conditions lies between 45 degrees and 60 degrees. In the following analysis, an initial pitch angle of 45 degrees is adopted as a marginally stable condition close to the instability threshold.

A model that accounts for structural deformation is referred to as the flexible model, while a model that neglects deformation is defined as the rigid model. The effect of structural deformation on attitude dynamics is evaluated by comparing the time histories of pitch motion for these two models.

Figure 2 shows the time histories of the pitch angle for both the flexible and rigid models. Figure 3 presents the Mach number distribution around the body and the surface pressure distribution. It is observed that the decay of oscillation amplitude is smaller in the flexible model, indicating that structural deformation contributes to increased dynamic instability.

Focusing on the deformation behavior of the flexible model (Fig. 4), two dominant deformation modes are identified: a swing mode, in which the torus oscillates about the pitch rotation axis, and an axial-stretch mode, in which the structure oscillates along the body axis. These modes are excited by aerodynamic forces. However, their natural frequencies are 9.45 Hz and 11.52 Hz, respectively, which are several tens of times higher than the pitch oscillation frequency of 0.26 Hz. Therefore, their direct influence on the attitude motion is considered to be limited.

On the other hand, from the motion histories of three points on the torus, the angle theta between the normal direction of the torus surface and the body center axis is calculated, and its time history is compared with that of the pitch angle alpha, as shown in Fig. 2. The oscillation of theta consists of a high-frequency component associated with the swing mode and a low-frequency component that is out of phase with alpha. As a result, theta always has the opposite sign to alpha, and the effective angle of attack of the torus, defined as theta + alpha, becomes consistently smaller than the angle of attack alpha defined with respect to the body axis.

This effect reduces the restoring aerodynamic force acting on the body at the same angle of attack compared to the rigid model, leading to weaker damping of the pitch oscillation.

Future work will focus on investigating the influence of structural deformation on the flow field in greater detail.

Fig.1: Geometry and computational meshes

Fig.2: Time history of pitch angle and torus angle

Fig.3(video): Flow field around the aeroshell and surface pressure distribution

Fig.4(video): Time histories of deformation and stress distribution (deformation scaled by a factor of 50)

Publications

- Oral Presentations

1) T. Sawada, A. Oyama, Y. Takahashi, R. Miyashita, and T. Yoshio, "Attitude Dynamics Analysis of an Inflatable Aeroshell Using Fluid–Structure Interaction," 69th Space Sciences and Technology Conference, Sapporo, (2025).

2) T. Sawada, A. Oyama, Y. Takahashi, R. Miyashita, and T. Yoshio, "One-Degree-Of-Freedom Attitude Stability Analysis of a Mars Entry Inflatable Aeroshell Using Fluid-Structure Interaction," AIAA SciTech 2026, Florida, (2026).

Usage of JSS

Computational Information

• Process Parallelization Methods: MPI
• Thread Parallelization Methods: OpenMP
• Number of Processes: 2400 - 4800
• Elapsed Time per Case: 100 Hour(s)

JSS3 Resources Used

Fraction of Usage in Total Resources*1(%): 0.43

Details

Please refer to System Configuration of JSS3 for the system configuration and major specifications of JSS3.

*1: Fraction of Usage in Total Resources: Weighted average of three resource types (Computing, File System, and Archiver).

*2: Fraction of Usage：Percentage of usage relative to each resource used in one year.

ISV Software Licenses Used

*2: Fraction of Usage：Percentage of usage relative to each resource used in one year.

JAXA Supercomputer System Annual Report February 2025-January 2026