Tracks of typhoons are predicted using a generative adversarial network (GAN) with observational data in form of satellite images and meteorological data from a reanalysis database. Time series of images of typhoons which occurred in the Korean Peninsula in the past are used to train the neural network. The trained GAN is employed to produce a 6-hour-advance track of a typhoon for which the GAN was not trained. The predicted image favorably identifies the future location of the typhoon center as well as the deformed cloud structures. The errors between predicted and real typhoon centers are measured quantitatively in kilometers. 65.5% of all typhoon center predictions have an error of less than 80 km, 31.5% lie within a range of 80− 120 km and the remaining 3.0% are above 120 km. The overall error is 67.2 km, compared to 95.6 km when only observational data are used as input. The cloud structure prediction is evaluated qualitatively. It is shown that the GAN is able to predict trends in cloud motion. It is found that adding physically meaningful meteorological data to satellite images improves the sharpness of predicted images.

A new flow solver scalable on multiple Graphics Processing Units (GPUs) for direct numerical simulation of wall-bounded incompressible flow is presented. This solver utilizes a previously reported work (J. Comp. Physics, vol. 352 (2018), pp.246-264) which proposes a semi-implicit fractional-step method on a single GPU. Extension of this work to accommodate multiple GPUs becomes inefficient when global transpose is used in the Alternating Direction Implicit (ADI) and Fourier-transform-based direct methods. A new strategy for designing an efficient multi-GPU solver is described to completely remove global transpose and achieve high scalability. Parallel Diagonal Dominant (PDD) and Parallel Partition (PPT) methods are implemented for GPUs to obtain good scaling and preserve accuracy. An overall efficiency of 0.89 is shown. Turbulent flat-plate boundary layer is simulated on 607M grid points using 4 Tesla P100 GPUs.

A numerical methodology for simulations of multi-mode heat transfer during a selenization process for CIGS (Copper Indium Gallium Selenide) solar cell films has been developed. Turbulent fluid dynamics and convective heat transfer are simulated using a finite-volume large-eddy simulation (LES) technique while thermal conduction and radiation are predicted using finite-element methods. The computational methodology is validated for three heat transfer modes. Using the technique, a numerical study of heat transfer during the selenization process for deposition of a CIGS layer is performed to analyze the thermo-fluid phenomena occurring during the process. The present method is found to well predict temperature distribution on substrates as a function of both space and time. It is also analyzed that how turbulent fluid motions alter temperature distributions on the substrate during the selenization process.

Achievement of solutions in Navier-Stokes equation is one of challenging quests, especially for its closure problem. For achievement of particular solutions, there are variety of numerical simulations including Direct Numerical Simulation (DNS) or Large Eddy Simulation (LES). These methods analyze flow physics through efficient reduced-order modeling such as proper orthogonal decomposition or Koopman method, showing prominent fidelity in fluid dynamics. Generative adversarial network (GAN) is a reprint of neurons in brain as combinations of linear operations, using competition between generator and discriminator. Current paper propose deep learning network for prediction of small-scale movements with large-scale inspections only, using GAN. Therefore DNS result of three-dimensional mixing-layer was filtered blurring out the small-scaled structures, then is predicted of its detailed structures, utilizing Generative Adversarial Network (GAN). This enables multi-resolution analysis being asked to predict fine-resolution solution with only inspection of blurry one. Within the grid scale, current paper present deep learning approach of modeling small scale features in turbulent flow. The presented method is expected to have its novelty in utilization of unprocessed simulation data, achievement of 3D structures in prediction by processing 3D convolutions, and predicting precise solution with less computational costs.

Tracks of typhoons are predicted using satellite images as input for a Generative Adversarial Network (GAN). The satellite images have time gaps of 6 hours and are marked with a red square at the location of the typhoon center. The GAN uses images from the past to generate an image one time step ahead. The generated image shows the future location of the typhoon center, as well as the future cloud structures. The errors between predicted and real typhoon centers are measured quantitatively in kilometers. 42.4% of all typhoon center predictions have absolute errors of less than 80 km, 32.1% lie within a range of 80 - 120 km and the remaining 25.5% have accuracies above 120 km. The relative error sets the above mentioned absolute error in relation to the distance that has been traveled by a typhoon over the past 6 hours. High relative errors are found in three types of situations, when a typhoon moves on the open sea far away from land, when a typhoon changes its course suddenly and when a typhoon is about to hit the mainland. The cloud structure prediction is evaluated qualitatively. It is shown that the GAN is able to predict trends in cloud motion. In order to improve both, the typhoon center and cloud motion prediction, the present study suggests to add information about the sea surface temperature, surface pressure and velocity fields to the input data.

Direct numerical simulations (DNS) are performed to investigate the effect of the interface position of water-air flow on turbulence statistics and flow structures in wall-bounded water-air turbulent flow through a straight channel. Water depths of 90 and 180 viscous wall units, referred to as shallow-water and deep-water cases, respectively, are examined. Water-to-air density and viscosity ratios of 831.7 and 55.56 are considered to model a realistic flow condition at temperature of 25 oC and pressure of 1 atm. The Reynolds number and Froude number are set to 180 and 1.22 x 10-4, respectively, for both shallow-water and deep-water case, based on the friction velocity at the bottom wall, the half depth of the channel, and water density and viscosity. The Navier-Stokes equations are solved using a timesplitting projection method on an octree grid structure, while the deformation of the interface between water and air is computed using a volume-of-fluid method. With the presence of the water-air interface, velocity profiles in deep-water and shallow-water cases are found to slightly deviate from the log-law profile for a single phase turbulent flow in a channel. The deviation is magnified when the interface is placed closer to the log-law region. Turbulent velocity fluctuations in the water stream are found to be associated with quasi-streamwise vortices and hairpin vortices. The quasi-streamwise vortices which are attached close to the wall are found in both deep-water and shallow-water cases. However, the hairpin vortices of which leading portions are lifted away from the wall are found to be diminished in the shallow-water case while they are clearly observed in the deep-water case.

In the scale-up of chemical production in a microfluidic system, it is challenging to prevent flow maldistribution from a single inlet into stacked multiple microchannel exits. In the present study, a compact flow distributor equipped with a fluidic damper is developed by computational fluid dynamics (CFD) along with experimental validation. A microfluidic flow distributor, which is equipped with an optimized fluidic damper and consists of 25 exit channels, is fabricated as an integrated body using a digital light processing (DLP) type 3D printer. The 3D printed flow distributor with a CFD-optimized fluidic damper is found to achieve a low maldistribution factor (MF) of 2.2% for the average flow rate over 25 exit channels while inducing only a minor increment (<6%) in the pressure drop. A generalized manual is proposed for the design of optimal flow distributors with different scale-up dimensions. Using the manual, an optimal flow distributor with 625 stacked microchannels with a MF of only 1.2% is successfully designed. It is expected that the design manual and the rapid printing platform will allow the efficient development of multi-channel stacked micro-devices such as those in drug delivery and energy conversion systems where equidistribution of fluid flow is highly demanded.

Utility of the computational power of Graphics Processing Units (GPUs) is elaborated for solutions of incompressible Navier–Stokes equations which are integrated using a semi-implicit fractional-step method. The Alternating Direction Implicit (ADI) and the Fourier-transform-based direct solution methods used in the semi-implicit fractional-step method take advantage of multiple tridiagonal matrices whose inversion is known as the major bottleneck for acceleration on a typical multi-core machine. A novel implementation of the semi-implicit fractional-step method designed for GPU acceleration of the incompressible Navier–Stokes equations is presented. Aspects of the programing model of Compute Unified Device Architecture (CUDA), which are critical to the bandwidth-bound nature of the present method are discussed in detail. A data layout for efficient use of CUDA libraries is proposed for acceleration of tridiagonal matrix inversion and fast Fourier transform. OpenMP is employed for concurrent collection of turbulence statistics on a CPU while the Navier–Stokes equations are computed on a GPU. Performance of the present method using CUDA is assessed by comparing the speed of solving three tridiagonal matrices using ADI with the speed of solving one heptadiagonal matrix using a conjugate gradient method. An overall speedup of 20 times is achieved using a Tesla K40 GPU in comparison with a single-core Xeon E5-2660 v3 CPU in simulations of turbulent boundary-layer flow over a flat plate conducted on over 134 million grids. Enhanced performance of 48 times speedup is reached for the same problem using a Tesla P100 GPU.

Unsteady laminar vortex shedding over a circular cylinder is predicted using a deep learning technique, a generative adversarial network (GAN), with a particular emphasis on elucidating the potential of learning the solution of the Navier-Stokes equations. Numerical simulations at two different Reynolds numbers with different time-step sizes are conducted to produce training datasets of flow field variables. Unsteady flow fields in the future at a Reynolds number which is not in the training datasets are predicted using a GAN. Predicted flow fields are found to qualitatively and quantitatively agree well with flow fields calculated by numerical simulations. The present study suggests that a deep learning technique can be utilized for prediction of laminar wake flow in lieu of solving the Navier-Stokes equations.

Aerodynamic drag reduction is directly related to fuel consumption and CO2 emission and is thus a main research interest in heavy vehicles. Approximately half of the total aerodynamic drag is attributed to the flow at the forebody of a vehicle and the gap between the tractor and trailer. Therefore, flow control devices that can reduce aerodynamic drag on the forebodies of heavy vehicles offer a considerably practical significance. Cab-roof fairing (CRF) is one of the most widely used drag reduction devices installed at the roofs of trucks or tractor-trailers. However, the drag-reducing effect and the three-dimensional flow characteristics around forebodies as a function of the external shape of CRFs have yet to be fully investigated. In this study, the drag reduction effects of typical and modified CRF models are quantitatively examined through wind tunnel tests and numerical simulation (coarse large eddy simulation (LES)). The wind tunnel experiment at a Reynolds number of >5.5X10^5 is conducted for a scaled-down model of a 15-tonne truck. The modified CRF significantly changes the flow structure, leading to approximately 19% drag reduction. PIV flow field measurement was conducted to figure out the differences in flow characteristics around the forebody of the vehicle with and without CRFs. Flow characteristics, including vortical structures, turbulent kinetic energy, mean pressure field, and mean velocity field around the forebody of the vehicle model (1/8 scale) with and without CRFs are numerically investigated using coarse LES for further understanding of the mechanism associated with drag reduction. The present results are expected to provide useful information for the design of new CRF models and the improvement of the aerodynamic performance of heavy vehicles, including trucks and tractor-trailers.