Current Research

Computational Fluid Dynamics & Parallel Computing

OUTLINE

Our field of research is large-scale parallel computing in the area of numerical algorithms. This research is closely related to fluid physics. Our studies focus on large-scale parallel computing for the various phenomena in fluid dynamics. Numerical studies of vortical flows near solid bodies require high accuracy and high resolution, since there are vortices with different spatial scales in these regions, and complicated nonlinear interactions are observed near the separation points and the wake in these flows. We investigate various vortical flow phenomena through large-scale numerical simulations using supercomputers as well as by employing numerical schemes for high-performance computation with better accuracy and greater resolution. Our research topics are as follows.

TOPICS

1. Numerical analysis of vortex shedding mechanisms at different Reynolds numbers. Such phenomena are observed in incompressible flows past bluff bodies and/or the flow past 2-D wings with a high angle of attack.
2. Incompressible vortical flow fields in three-dimensional lid-driven cavities at different Reynolds numbers and different aspect ratios. These flow fields are simple models of a short-dwell coater.
3. Numerical estimation of aerodynamical sound and numerical investigation of the sound source phenomena in the flow past wings and in induced flows due to the interaction of two vortex rings. These are basic studies of fan noise and/or noise of high-speed vehicles.
4. Numerical investigation of 3D motions of gas bubbles in liquids and/or interfaces using a three-dimensional level set approach.
5. Numerical analysis of blood flows in/near aneurysms in arteries of the brain. The development and implementation of effective numerical schemes with greater accuracy and higher resolution using high-performance computation is another research subject.
6. Study of finite difference schemes with high accuracy and spectral-like resolution, i.e. the vector-potential method, efficient compressible schemes, and combined compact schemes. In addition, the development and effective implementation of parallel algorithms is also investigated using SIMD and MIMD computers.

It is becoming apparent that grid computing will be an enabler for major advances and new ways of conducting science. We plan to study grid computing environments and applications connecting the supercomputer of Nagoya University with other supercomputers in Japan. We will also study data grids in which a virtual large-scale data set of files is accessed without regard to location or platform.