A non-local eddy diffusivity model for turbulent scalar flux was recently proposed for homogeneous isotropic turbulence. In this study, the model was improved by incorporating the effects of turbulence anisotropy and inhomogeneity and applied to a turbulent channel flow. The one- and two-dimensional profiles of non-local eddy diffusivity were accurately evaluated using DNS data. The exact DNS profiles revealed a contribution to the scalar flux from the mean scalar gradient in a wide upstream region. Temporal profiles of non-local eddy diffusivity move downstream and diffuse anisotropically. The improved model succesfully reproduced this behavior of mean flow convection and anisotropic turbulent diffusion.

Turbulent Rayleigh flow is a test problem to investigate the effects of heating on turbulence. Last year, we implemented the direct numerical simulation of the turbulent Rayleigh flow and showed that counter-gradient diffusion of turbulent heat flux occurs in it, and the mechanism of the counter-gradient diffusion was discussed by analyzing the transport equation of the statistical quantities. In this study, we analyzed the transport equation of the statistical quantities focusing only on the linear terms, which plays crucial role in the mechanism of counter-gradient diffusion. As a result, analytical solution was obtained and was qualitatively consistent with the result of the direct numerical simulation. This result clearly explained the qualitative mechanism of the counter-gradient diffusion in turbulent Rayleigh flow.

A direct numerical simulation of magnetohydrodynamic (MHD) turbulent channel flow was carried out to investigate turbulent dynamo effect. Time history of the volume-averaged magnetic field showed that a positive magnetic field was generated and sustained. To investigate the mechanism of the magnetic field generation, we obtained the mean fields and the statistical quantities. It was shown that the turbulent electromotive force contributes to the magnetic field generation against the Ohmic term. The profiles of the production terms of the turbulent electromotive force clearly showed that the cross-helicity dynamo actually contributes to the magnetic field generation whereas the pumping dynamo plays a role of anti-dynamo decreasing the magnetic field.

In non-mirror-symmetric system, turbulent helicity enters the Reynolds-stress expression as the coupling coefficient of the mean absolute vorticity (anti-symmetric part of the mean velocity shear) (: velocity fluctuation, : vorticity fluctuation, and : ensemble average). This gives marked contrast to the turbulent energy, which plays an essential role in the eddy-viscosity representation of the Reynolds stress as the coupling coefficient of the mean velocity strain (symmetric part of the mean velocity shear). By considering the turbulent vortex-motive force in the mean vorticity equation, it is shown that a large-scale flow can be induced in the direction of the mean absolute vorticity (mean vorticity and system rotation) mediated by the inhomogeneous turbulent helicity. This inhomogeneous helicity effect is applied to the large-scale flow generation and sustainment in the stellar convective zone. The contribution of the inhomogeneous helicity effect to the angular-momentum transport in the stellar convection is discussed with the aid of some direct numerical simulations. Emphasis is also made on the turbulent helicity as a link between the large-scale flow structures, like differential azimuthal rotation and the meridional circulation, and the statistical properties of turbulence.

This study proposes an anisotropic concentration diffusivity model in the Reynolds-averaged Navier-Stokes equations and the Eulerian dispersion model. The proposed model combines models to consider the turbulent anisotropic and near-source limited diffusivity based on generalized gradient-diffusion hypothesis and travel time. The proposed model and conventional isotropic models were applied to predict the pollutant dispersion in an atmospheric boundary layer from an elevated source. The predicted concentration profile and plume half-width were validated with an existing wind tunnel experiment. Both proposed model and the isotropic model using the diffusivity limiter accurately predicted the mean concentration profiles at the central vertical plane. However, the equivalent turbulent Schmidt number in the proposed model were different in each direction. The proposed model predicted counter-gradient turbulent diffusion in the streamwise direction.

This study proposes a method using Physics-Informed Neural Network (PINN) to reconstruct high-resolution ventilation flow fields from sparse measurement data, addressing spatial resolution limitations in traditional experiments. Data from scaled-model aircraft cabin experiments validated the approach, demonstrating effective utilization of mean velocity data to enhance resolution. The results revealed accurate velocity field reconstruction when using the complete dataset, including regions challenging for measurement using particle image velocimetry. Even with reduced measurement data, maintaining approximately 300 or more data points ensured reconstruction accuracy with deviations below 10% of inflow velocity. Using fewer than 100 data points showed localized accuracy decline, yet key flow features remained.

Drones are widely used to conduct various missions in urban districts containing complex layouts of high-rise buildings. To promote safe, energy-efficient flights in urban areas, it is necessary to evaluate the influence of turbulent flow induced by these buildings on the flight stability of drones. In this study, we used a numerical platform to simulate the flight conditions of drones passing through different routes around an isolated high-rise building. The averaged flow fields predicted by large-eddy simulation were input into the platform to induce wind effects. The results showed that the flight conditions were significantly changed when the drone passed through separation flow structures caused by the building.

This study aims to optimize water currents in semi-closed fish cages to enhance water recirculation systems for aquaculture. By combining Particle Image Velocimetry (PIV) experiments and numerical simulations, flow characteristics and dissolved oxygen (DO) are analyzed. Results reveal that parameters like inlet angles and hole numbers significantly impact characteristics of flow fields. For simulating DO, a customized solver is developed. Optimized setups can improve fish health, oxygen levels, and waste removal efficiency, contributing to sustainable aquaculture. The findings offer theoretical foundations for designing advanced aquaculture systems, emphasizing stable environments.

The infrastructure-cooperative automated driving systems has garnered significant interest as a potential avenue for the expeditious social deployment of Level4 automated driving. While demonstration tests have been conducted for cooperative automated driving in limited situations, the development of comprehensive guidelines for the evaluation of the entire cooperative system is a crucial step towards its implementation in a wide range of regions. Once the impact of the integration of diverse functions and the variability in the performance of each function within a cooperative system is fully definite, it will be possible to devise a cooperative system that should be implemented and to set the ODD for cooperative automated driving. This study presents the methodology along with the case example.

The widespread adoption of self-driving cars brings new possibilities for traffic control at intersections. In this paper, we use traffic simulation (SUMO) to investigate the challenges of current traffic control methods in the context of self-driving cars and propose a new traffic control method aimed at improving traffic flow. Specifically, the method utilizes information from each vehicle attempting to enter the intersection to enhance time efficiency. By comparing the proposed method with existing traffic control methods using the simulator, we demonstrate its superior time efficiency, paving the way for more effective intersection management in an era of self-driving cars.

Multi-agent reinforcement learning-based arterial traffic signal control offers flexible, scalable control for complex networks through adaptive responsiveness and distributed coordination. Incorporating neighboring intersection traffic information into state features may improve control performance. Experiments are conducted to test linear RL model’s behavior under two training environments and different combinations of local and neighboring state features. The results show that both joint training environment as well as incorporating information sharing lead to better agent coordination and control performance, while state features with vehicle counts generally outperform those with queuing length at main direction but underperform at other directions.

In rainy conditions, there is a tendency for accidents to increase. Additionally, reduced road friction due to wet conditions can lead to accidents. On the Tokyo Metropolitan Expressway, numerous cameras are installed to monitor traffic. While studies have explored detecting snow and ice through camera images and estimating road conditions using in-vehicle cameras, few studies have specifically focused on detecting wet pavement conditions using existing traffic surveillance cameras. This research aims to develop a system that uses images captured by traffic cameras and applies machine learning-based image classification techniques to identify wet road conditions automatically.

A new railroad crossing control system using mobile networks has been developed, eliminating the need for traditional ground equipment or wired circuits. This system transmits information via mobile networks, allowing cars and pedestrians to receive it as well. Previous studies developed a similar system that detects stopping positions, checks for space after crossings, estimates crossing time, and aids vehicle departure decisions. This research aims to apply this system to electric wheelchairs, improving safety at pedestrian and railroad crossings and evaluating performance through field experiments.

Recently, whereas there have been advancements in autonomous driving across automotive and railway sectors, the demand for driverless trams has also grown due to a shortage of tram drivers, resulting from declining fertility rates. To achieve this, it is crucial to equip trams with vision capabilities to prevent collisions with on-track obstacles, ranging from pedestrians to surrounding cars. Accordingly, this paper aims to propose an algorithm to determine whether the tram can maintain its safe clearance from obstacles using a frontal camera. The proposed vehicle clearance algorithm is developed using traditional image-processing techniques, and its performance is evaluated via simulations.

This paper proposes “ExBIM,” a building information model focusing on externalities, which models buildings exclusively using information obtainable from the external. ExBIM targets building exteriors and outdoor structures, also other information publicly available. It can be created using existing software environment. This modeling and management approach contributes to solving various social issues, including autonomous mobility, environmental impact assessment, description of outdoor structures and elements, and building maintenance. ExBIM complements existing 3D city models and geographic information systems, offering high utility as information corresponding to building externalities. Furthermore, its low information acquisition barriers and cost-effectiveness make it suitable for large-scale information development.