This paper presents numerical solution to a shape optimization for stationary fluid structure interactive fields. In the fluid structure interactive analysis, a weak coupled analysis is used to alternately analyze the governing equations of the flow field domain and the structural field considering geometrically nonlinear. A mean compliance minimization problem is formulated in order to achieve stiffness maximization on the structural field. Shape derivative, which means the sensitivity in the shape optimization problem, is derived theoretically by using the Lagrange multiplier method and adjoint variable method, and the formulae of the shape derivative with respect to domain variation of the distribution function. Reshaping is carried out by the H¹ gradient method proposed as an approach to solving shape optimization problems. Numerical analysis program for the problem is developed by using FreeFEM, and validity of proposed method is confirmed by numerical results of 2D problems.

A one-dimensional theoretical analysis is made of the pressure field around a train running in a long tube with a constant width slit, which is a simplified model of partially enclosed structures (or vented tubes) of railway, such as snow shelters or whole covered stations. It is shown from field measurement results in Shinkansen that the pressure field produced in such partially enclosed structures exhibits a pattern of ‘a combination of single sawtooth and its reverse’. The present analysis of a tube with a slit reproduces this unique pressure pattern and shows that one-dimensional pressure field ahead of the train nose and tail can be expressed by elementary functions. It also reveals that the magnitude of the pressure change around the tail can exceed that around the nose, which is also a unique feature that is different from the cases of open air spaces and fully enclosed spaces (i.e., tunnels).

In this study, we developed a method for obtaining high contributing part (reference point) to the response point at the operational condition by operational TPA (OTPA) using several measurement systems. OTPA calculates contribution of each reference point to the response point using only operational signals. All reference and response signals are necessary to be measured simultaneously by single measurement system because the method calculates the contribution using their correlation along time. However, this occasionally requires preparing large measurement system depending on the number of measurement points and the size of products. This may decrease the applicability of the method. We then considered a post processing procedure to obtain accurate contribution of each reference point to the response point by using several measurement systems instead of preparing large single measurement system. In the proposed method, all signals are measured using different several systems at around same timing. The exact sampling timing gaps among systems are estimated by using the estimation error between the calculated and actual measured response signal. After then, all reference signals compensated by the estimated time gap in each system are regarded to be measured simultaneously and contribution of all reference signals are calculated by OTPA. As the verification of the proposed method, the procedure was applied to a simple vehicle model. As the result, the contribution of the proposed method was similar with the contribution by single measurement system and clarified to have an ability to obtain correct contribution by several measurement systems.

In this work, we propose a semi-implicit, density-based solver for compressible, evaporating particle-laden flow, and investigate its efficiency. It is established on a Cartesian-grid-based, scalable, numerical framework named CUBE. In this solver, the governing equation system is divided into three subsystems (compressible Navier-Stokes, species transport, and Lagrangian), and these subsystems are weakly coupled in two ways. In the Lagrangian domain, the fuel spray is treated as a set of discrete particles, and the particle-source-in-cell (PSI-Cell) method is employed for the coupling between the Eulerian and Lagrangian domains. Furthermore, the species transport and Lagrangian subsystems are subcycled with smaller time step, and the Navier-Stokes equation is temporally integrated with a larger step size. The proposed solver's verification and evaluation is conducted on the supercomputer Fugaku by comparing the results with those of the original, fully explicit solver where all equations have the same time step. The results show that this solver reduces the computational cost while ensuring similar accuracy. The solution of the proposed solver is consistent with that of the original solver. Finally, we brief our perspective on the future application of the proposed solver to our target problem: the large-scale simulation of evaporating particle-laden flow in a combustor of an aviation engine.

Muscle activities are usually evaluated as the subjects of motion analysis in human performance surveys. Force-myography (FMG), which has a similar objective to electromyography, has been proposed and applied in motion analysis. Contact pressure sensors were adopted to construct the FMG device, designed for measuring muscular activities in motion. The essential factors for the detection of FMG are changes in the thickness and hardness of muscles during motion with muscle activity. This study proposed a simple structure, which is applicable as an FMG sensor, to selectively measure the hardness changes attributable to muscle activity. In addition, compact FMG sensors were manufactured by focusing on the hardness measurements by means of pressure sensing. Sensor characteristics in the dumbbell curl posture and walking motion were investigated, and some applications were provided in this study. The changes in hardness expressed the variations in muscle activity of these motions. The usability and availability of the sensor unit were confirmed in the experiments focused on muscle activity.

One of the challenges facing widespread electric vehicle (EV) adoption is the short driving range. To address this challenge, the development of various EV transmissions is underway, but the shock during EV shifting is more noticeable than for gasoline engines. In order to achieve seamless gear shifting, the motor and clutch must be controlled in response to the torque transmitted through the clutch. This torque is theoretically proportional to the friction coefficient and the thrust force but is difficult to estimate in reality as it changes according to various factors. Therefore, this study applied deep reinforcement learning to automatically learn a gear shifting control method for the two-speed dual clutch transmission of an EV that adapts to the control target (seamless transition of output torque) through trial and error. A model for learning this control method was developed using the Actor/Critic network to explore the possibility of automatically designing motor torque and clutch control rules to achieve seamless gear shifting through repeated learning of the resulting output torque. The gear shifting results with and without the developed control method were then compared. It was found that control rules could be automatically designed to achieve seamless gear shifting by applying deep reinforcement learning to the transmission in this study. Furthermore, it was possible to control the motor torque and clutch to enable seamless gear shifting without directly monitoring the clutch transmission torque.

In this paper, we propose a new visualization method using self-organizing map (SOM) for computed results of fluid flow. Most of existing visualization method gives color depending on a certain physical value, such as pressure, vorticity, etc. However the choice of the physical value is arbitrary and sometimes loses important features. The present method firstly classifies all flow properties, i.e. pressure, velocity components and their spatial gradients at each grid point, by giving these properties as high order vectors to SOM. Then color is given to each grid point based on its location on the map so that the flow field be naturally painted including all flow properties. SOM is originally a two-dimensional (2D) map as the main purpose of SOM is to visualize high order vectors, but in the present study the map is not directly viewed and only used to determine color of each grid point. Hence we try to generate three-dimensional (3D) SOM to give three color components based on the map. Both 2D and 3D map results are shown to demonstrate the capability of the present method.