Noriaki Nishikawa

Parallel Computing of Magnetic Field for Rotating Machines on the Earth Simulator

We developed a parallel computing method of the magnetic field for rotating machines by using the 3-D finite element method (FEM) with edge elements. In this paper, we describe the outline of the developed method and an optimization of the method for the Earth simulator, which is a vector-type parallel supercomputer in Japan. Moreover, the performance of the proposed method running on the Earth simulator is quantitatively clarified.

3-D Finite Element Analysis of Eddy Current in Laminated Cores of the Interior Permanent-Magnet Motor

In this paper, a large-scale numerical analysis for eddy currents in laminated cores of an interior permanent-magnet motor is achieved. The eddy currents in the laminated cores caused by the axial flux are simulated by using the Earth Simulator, which is a vector-type parallel supercomputer.

Parallel computing of magnetic field for rotating machines excited from voltage sources on the Earth Simulator

A parallel computing method for rotating machines excited from voltage source with the three-dimensional finite element method is developed, and the usefulness is verified through the computation of an IPM motor with an off-centered rotor on the Earth Simulator.

Numerical Simulation of Long-Term Fate of CO 2 Stored in Deep Reservoir Rocks on Massively Parallel Vector Supercomputer

As one of the promising approaches for reducing greenhouse-gas content in the atmosphere, CCS (carbon dioxide capture and storage) has been recognized worldwide. CO2 is captured from large emission sources and injected and stored in deep reservoir rocks, including saline aquifers, depleted oil and gas field. Under typical pressure and temperature conditions at deep reservoirs (depths > 800m), CO2 will be stored in supercritical state, subsequently dissolving in groundwater, and eventually forming carbonate minerals through geochemical reactions in a long-term (e.g., thousands of years). To ensure the safety and permanence of the storage, numerical simulation is considered as the most powerful approach for predicting the long-term fate of CO2 in reservoirs. A parallelized general-purpose hydrodynamics code TOUGH2-MP has been used on scalar architectures where it exhibits excellent performance and scalability. However, on the Earth Simulator (ES2), which is a massively parallel vector computer, extensive tune-ups were required for increasing the vector operation ratio. In this paper, the performance of the modified TOUGH2-MP code on ES2 is presented with some illustrative numerical simulations of long-term fate of CO2 stored in reservoirs.

Prediction of flow field and aerodynamic noise of jet fan using large eddy simulation

To conduct a product development of turbomachinery continuously and developmentally, an approach using large eddy simulation (LES) should be one of the mainstream, because it is already essential to utilize computational fluid dynamics (CFD) using Reynolds averaged Navier–Stokes Simulation in industrial field. In recent years, the demand for a high-efficiency and low-noise jet fan, which is used for ventilation of a tunnel of a carriage way, is getting higher and higher. However, to conduct noise prediction of the jet fan with high accuracy, it needs the high-performance computing technique using a supercomputer. In this study, CFD of the jet fan is conducted using LES. Noise simulation is also conducted using the LES result. Moreover, the fine mesh model is generated by refining a coarse mesh, and LES is conducted using the Earth Simulator. The result of surface pressure using the fine mesh is clearer than the one using the coarse mesh.Graphical abstract

A study of real-time and 100 billion agents simulation using the Boids model

In high-performance computing of multi-agent systems, there often exists a load imbalance that slows down the calculation. In this paper, we discuss the parallelization of the Boids model for simulating a swarm intelligence. We apply the multi-level hierarchy of parallelism to the Boids model to mitigate the load-imbalance problem. To eliminate numerical errors due to parallelization, we apply pseudo-quadruple arithmetic. The parallel performance is evaluated on three major architectures, including many-core processors on an x86-based server with GPUs, and the Earth Simulator. The parallelization can decrease the negative effects of a load imbalance to almost zero in a simulation of 50 million agents. In addition, the parallelization guarantees the reproducibility of the results in a sequential execution. The strong scaling shows the potential to complete a simulation in real-time on the Earth Simulator. In addition, the weak scaling shows the ability to calculate 100 billion agents within a reasonable amount of time.