大山 聖

In order to improve an efficiency of failure analysis process by using failure mode and effect analysis (FMEA) and failure tree analysis (FTA), detailed FMEA/FTA support tool is developed. By using this tool, failure mechanisms can be analyzed with FMEA approach as well as FTA approach. In order to improve coverage of the extracted failure modes, new failure analysis approach is proposed, which is based on the interface N2 chart. Interface N2 chart is N2 chart which is extended to express the interface of mechanical system. Interface N2 chart can also be used to visualize the failure analysis result to describe system structure based on the knowledge obtained in failure analysis such as causal relationship of fault propagation. New schematic visualization method is discussed to realize useful visualization tool which promotes comprehensive understanding of failure mechanism.

Flow fields of the supersonic jets impinging on an inclined flat plate at high plate-angles are experimentally investigated using surface pressure measurement with pressure sensitive paint and Schlieren flow visualization. While Type I flow type is dominant at high plate angles, the present research found a new flow type "TYPE I with bubble" at plate angle between 60 and 80 degrees. The flow classification according to L/L_s' and plate angle indicated that the constant x/L'_s curve doesn't represent the boundary of Type I and Type II anymore at high plate angles between 60 and 90 probably because Type II flows at low plate angles and high plate angles is different phenomena. This study also indicates that the curve dividing Type I and Type I with bubble regions is same as the curve dividing Type II and Type II with bubble regions.

An efficient and useful robust optimization approach, design for multi-objective six sigma (DFMOSS), has been developed. The DFMOSS couples the ideas of design for six sigma (DFSS) and multi-objective genetic algorithm (MOGA) to solve drawbacks of DFSS. DFMOSS obtains trade-off solutions between optimality and robustness in one optimization. In addition, it does not need careful parameter tuning. Robust optimizations of a test function and welded beam design problem demonstrated that DFMOSS is more effective and more useful than DFSS.

Low Reynolds number airfoil design optimization is demonstrated. A two-dimensional Navier-Stokes solver is used for aerodynamic performance estimation of the airfoil design candidates because the viscosity effect is not negligible at low Reynolds number flight condition. An evolutionary computation is used for robust airfoil design optimization. The result demonstrated that the optimized airfoil is a very thin airfoil with significant camber. The result also showed that the low Reynolds number airfoil has time variation in its lift and drag.