scholarly journals Simulations of co-axial jet flows on graphics processing units: the flow and noise analysis

2019 ◽  
Vol 377 (2159) ◽  
pp. 20190083 ◽  
Author(s):  
A. P. Markesteijn ◽  
S. A. Karabasov
Keyword(s):  
High Resolution ◽  
Graphics Processing Units ◽  
Stokes Equations ◽  
Noise Analysis ◽  
Jet Noise ◽  
Jet Flows ◽  
Noise Sources ◽  
Fine Grid ◽  
Graphics Processing ◽  
Noise Models

Large-eddy simulations (LES) are performed for a range of perfectly expanded co-axial jet cases corresponding to conditions of the computation of coaxial jet noise (CoJeN) experiment by QinetiQ. In all simulations, the high-resolution Compact Accurately Boundary-Adjusting high-REsolution Technique (CABARET) is used for solving the Navier–Stokes equations on unstructured meshes. The Ffowcs Williams–Hawkings method based on the penetrable integral surfaces is applied for far-field noise predictions. To correctly model the turbulent flow downstream of the complex nozzle that includes a central body, a wall modelled LES approach is implemented together with a turbulent inflow condition based on synthetic turbulence. All models are run on graphics processing units to enable a considerable reduction of the flow solution time in comparison with the conventional LES. The flow and noise solutions are validated against the experimental data available with 1–2 dB accuracy being reported for noise spectra predictions on the fine grid. To analyse the structure of effective noise sources of the jets, the covariance of turbulent fluctuating Reynolds stresses is computed and their characteristic scales are analysed in the context of the generalized acoustic analogy jet noise models. Motivated by self-similarity of single-stream axi-symmetric jet flows, a suitable non-dimensionalization of the effective jet noise sources of the CoJeN jets is tested, and its implications for low-order jet noise models are discussed. This article is part of the theme issue ‘Frontiers of aeroacoustics research: theory, computation and experiment’.

scholarly journals Data driven global weather predictions at high resolutions

2021 ◽  
Author(s):  
John Taylor ◽  
Pablo Larraonndo ◽  
Bronis de Supinski
Keyword(s):  
High Resolution ◽  
Graphics Processing Units ◽  
Weather Forecasting ◽  
Weather Prediction ◽  
Atmospheric Science ◽  
Data Driven ◽  
Research And Innovation ◽  
Future State ◽  
Graphics Processing ◽  
Quantities Of Interest

Abstract Society has benefited enormously from the continuous advancement in numerical weather prediction that has occurred over many decades driven by a combination of outstanding scientific, computational and technological breakthroughs. Here we demonstrate that data driven methods are now positioned to contribute to the next wave of major advances in atmospheric science. We show that data driven models can predict important meteorological quantities of interest to society such as global high resolution precipitation fields (0.25 degrees) and can deliver accurate forecasts of the future state of the atmosphere without prior knowledge of the laws of physics and chemistry. We also show how these data driven methods can be scaled to run on super-computers with up to 1024 modern graphics processing units (GPU) and beyond resulting in rapid training of data driven models, thus supporting a cycle of rapid research and innovation. Taken together, these two results illustrate the significant potential of data driven methods to advance atmospheric science and operational weather forecasting.

WAVELET-BASED POSTPROCESSING OF JET LES DATA FOR ACOUSTIC FAR-FIELD EXTRAPOLATIONS

2013 ◽  
Vol 21 (04) ◽  
pp. 1350017
Author(s):  
RAMIN KAVIANI ◽  
VAHID ESFAHANIAN ◽  
MOHAMMAD EBRAHIMI
Keyword(s):  
Large Scale ◽  
Stokes Equations ◽  
Near Field ◽  
Flow Simulation ◽  
Computational Time ◽  
Jet Flows ◽  
Small Scale ◽  
Frequency Noise ◽  
Far Field ◽  
Noise Sources

The affordable grid resolutions in conventional large-eddy simulations (LESs) of high Reynolds jet flows are unable to capture the sound generated by fluid motions near and beyond the grid cut-off scale. As a result, the frequency spectrum of the extrapolated sound field is artificially truncated at high frequencies. In this paper, a new method is proposed to account for the high frequency noise sources beyond the resolution of a compressible flow simulation. The large-scale turbulent structures as dominant radiators of sound are captured in LES, satisfying filtered Navier–Stokes equations, while for small-scale turbulence, a Kolmogorov's turbulence spectrum is imposed. The latter is performed via a wavelet-based extrapolation to add randomly generated small-scale noise sources to the LES near-field data. Further, the vorticity and instability waves are filtered out via a passive wavelet-based masking and the whole spectrum of filtered data are captured on a Ffowcs-Williams/Hawkings (FW-H) surface surrounding the near-field region and are projected to acoustic far-field. The algorithm can be implemented as a separate postprocessing stage and it is observed that the computational time is considerably reduced utilizing a hybrid of many-core and multi-core framework, i.e. MPI-CUDA programming. The comparison of the results obtained from this procedure and those from experiments for high subsonic and transonic jets, shows that the far-field noise spectrum agree well up to 2 times of the grid cut-off frequency.

scholarly journals GPU IMPLEMENTATION OF A VISCOUS FLOW SOLVER ON UNSTRUCTURED GRIDS

2016 ◽  
Vol 42 ◽  
pp. 1660167
Author(s):  
TIANHAO XU ◽  
LONG CHEN
Keyword(s):  
Graphics Processing Units ◽  
Stokes Equations ◽  
Unstructured Grids ◽  
Flow Simulation ◽  
Navier Stokes ◽  
Competitive Advantages ◽  
Navier Stokes Equations ◽  
Flow Solver ◽  
Volume Method ◽  
Graphics Processing

Graphics processing units have gained popularities in scientific computing over past several years due to their outstanding parallel computing capability. Computational fluid dynamics applications involve large amounts of calculations, therefore a latest GPU card is preferable of which the peak computing performance and memory bandwidth are much better than a contemporary high-end CPU. We herein focus on the detailed implementation of our GPU targeting Reynolds-averaged Navier-Stokes equations solver based on finite-volume method. The solver employs a vertex-centered scheme on unstructured grids for the sake of being capable of handling complex topologies. Multiple optimizations are carried out to improve the memory accessing performance and kernel utilization. Both steady and unsteady flow simulation cases are carried out using explicit Runge-Kutta scheme. The solver with GPU acceleration in this paper is demonstrated to have competitive advantages over the CPU targeting one.

scholarly journals Data-driven global weather predictions at high resolutions

2021 ◽  
pp. 109434202110398
Author(s):  
John A Taylor ◽  
Pablo Larraondo ◽  
Bronis R de Supinski
Keyword(s):  
High Resolution ◽  
Graphics Processing Units ◽  
Weather Forecasting ◽  
Weather Prediction ◽  
Atmospheric Science ◽  
Data Driven ◽  
Research And Innovation ◽  
Future State ◽  
Graphics Processing ◽  
Quantities Of Interest

Society has benefited enormously from the continuous advancement in numerical weather prediction that has occurred over many decades driven by a combination of outstanding scientific, computational and technological breakthroughs. Here, we demonstrate that data-driven methods are now positioned to contribute to the next wave of major advances in atmospheric science. We show that data-driven models can predict important meteorological quantities of interest to society such as global high resolution precipitation fields (0.25°) and can deliver accurate forecasts of the future state of the atmosphere without prior knowledge of the laws of physics and chemistry. We also show how these data-driven methods can be scaled to run on supercomputers with up to 1024 modern graphics processing units and beyond resulting in rapid training of data-driven models, thus supporting a cycle of rapid research and innovation. Taken together, these two results illustrate the significant potential of data-driven methods to advance atmospheric science and operational weather forecasting.

2012 Freeman Scholar Lecture: Computational Fluid Dynamics on Graphics Processing Units

2013 ◽  
Vol 135 (6) ◽  
Author(s):  
S. P. Vanka
Keyword(s):  
Fluid Dynamics ◽  
Computational Fluid Dynamics ◽  
Graphics Processing Units ◽  
Stokes Equations ◽  
Multicore Processors ◽  
Fluid Flows ◽  
Data Access ◽  
Central Processing ◽  
Data Parallel ◽  
Graphics Processing

This paper discusses the various issues of using graphics processing units (GPU) for computing fluid flows. GPUs, used primarily for processing graphics functions in a computer, are massively parallel multicore processors, which can also perform scientific computations in a data parallel mode. In the past ten years, GPUs have become quite powerful and have challenged the central processing units (CPUs) in their price and performance characteristics. However, in order to fully benefit from the GPUs' performance, the numerical algorithms must be made data parallel and converge rapidly. In addition, the hardware features of the GPUs require that the memory access be managed carefully in order to not suffer from the high latency. Fully explicit algorithms for Euler and Navier–Stokes equations and the lattice Boltzmann method for mesoscopic flows have been widely incorporated on the GPUs, with significant speed-up over a scalar algorithm. However, more complex algorithms with implicit formulations and unstructured grids require innovative thinking in data access and management. This article reviews the literature on linear solvers and computational fluid dynamics (CFD) algorithms on GPUs, including the author's own research on simulations of fluid flows using GPUs.

Sign in / Sign up

Export Citation Format

Share Document