Parallel Navier-Stokes multi-block code to solve industrial aerodynamic design problems on high performance computers

Author(s):  
V. Van Kemenade ◽  
M. J. Daydé ◽  
J. B. Vos
Author(s):  
Jack Dongarra ◽  
Laura Grigori ◽  
Nicholas J. Higham

A number of features of today’s high-performance computers make it challenging to exploit these machines fully for computational science. These include increasing core counts but stagnant clock frequencies; the high cost of data movement; use of accelerators (GPUs, FPGAs, coprocessors), making architectures increasingly heterogeneous; and multi- ple precisions of floating-point arithmetic, including half-precision. Moreover, as well as maximizing speed and accuracy, minimizing energy consumption is an important criterion. New generations of algorithms are needed to tackle these challenges. We discuss some approaches that we can take to develop numerical algorithms for high-performance computational science, with a view to exploiting the next generation of supercomputers. This article is part of a discussion meeting issue ‘Numerical algorithms for high-performance computational science’.


Water ◽  
2021 ◽  
Vol 13 (7) ◽  
pp. 902
Author(s):  
Zhi Li ◽  
Ben R. Hodges

A new high-performance numerical model (Frehg) is developed to simulate water flow in shallow coastal wetlands. Frehg solves the 2D depth-integrated, hydrostatic, Navier–Stokes equations (i.e., shallow-water equations) in the surface domain and the 3D variably-saturated Richards equation in the subsurface domain. The two domains are asynchronously coupled to model surface-subsurface exchange. The Frehg model is applied to evaluate model sensitivity to a variety of simplifications that are commonly adopted for shallow wetland models, especially the use of the diffusive wave approximation in place of the traditional Saint-Venant equations for surface flow. The results suggest that a dynamic model for momentum is preferred over diffusive wave model for shallow coastal wetlands and marshes because the latter fails to capture flow unsteadiness. Under the combined effects of evaporation and wetting/drying, using diffusive wave model leads to discrepancies in modeled surface-subsurface exchange flux in the intertidal zone where strong exchange processes occur. It indicates shallow wetland models should be built with (i) dynamic surface flow equations that capture the timing of inundation, (ii) complex topographic features that render accurate spatial extent of inundation, and (iii) variably-saturated subsurface flow solver that is capable of modeling moisture change in the subsurface due to evaporation and infiltration.


Aerospace ◽  
2021 ◽  
Vol 8 (8) ◽  
pp. 216
Author(s):  
Emanuel A. R. Camacho ◽  
Fernando M. S. P. Neves ◽  
André R. R. Silva ◽  
Jorge M. M. Barata

Natural flight has consistently been the wellspring of many creative minds, yet recreating the propulsive systems of natural flyers is quite hard and challenging. Regarding propulsive systems design, biomimetics offers a wide variety of solutions that can be applied at low Reynolds numbers, achieving high performance and maneuverability systems. The main goal of the current work is to computationally investigate the thrust-power intricacies while operating at different Reynolds numbers, reduced frequencies, nondimensional amplitudes, and mean angles of attack of the oscillatory motion of a NACA0012 airfoil. Simulations are performed utilizing a RANS (Reynolds Averaged Navier-Stokes) approach for a Reynolds number between 8.5×103 and 3.4×104, reduced frequencies within 1 and 5, and Strouhal numbers from 0.1 to 0.4. The influence of the mean angle-of-attack is also studied in the range of 0∘ to 10∘. The outcomes show ideal operational conditions for the diverse Reynolds numbers, and results regarding thrust-power correlations and the influence of the mean angle-of-attack on the aerodynamic coefficients and the propulsive efficiency are widely explored.


1997 ◽  
Vol 119 (1) ◽  
pp. 122-128 ◽  
Author(s):  
S. L. Puterbaugh ◽  
W. W. Copenhaver

An experimental investigation concerning tip flow field unsteadiness was performed for a high-performance, state-of-the-art transonic compressor rotor. Casing-mounted high frequency response pressure transducers were used to indicate both the ensemble averaged and time varying flow structure present in the tip region of the rotor at four different operating points at design speed. The ensemble averaged information revealed the shock structure as it evolved from a dual shock system at open throttle to an attached shock at peak efficiency to a detached orientation at near stall. Steady three-dimensional Navier Stokes analysis reveals the dominant flow structures in the tip region in support of the ensemble averaged measurements. A tip leakage vortex is evident at all operating points as regions of low static pressure and appears in the same location as the vortex found in the numerical solution. An unsteadiness parameter was calculated to quantify the unsteadiness in the tip cascade plane. In general, regions of peak unsteadiness appear near shocks and in the area interpreted as the shock-tip leakage vortex interaction. Local peaks of unsteadiness appear in mid-passage downstream of the shock-vortex interaction. Flow field features not evident in the ensemble averaged data are examined via a Navier-Stokes solution obtained at the near stall operating point.


1992 ◽  
Vol 10 (6) ◽  
pp. 632-632
Author(s):  
Stuart M. Dambrot

2011 ◽  
Vol 64 (2) ◽  
Author(s):  
Giancarlo Alfonsi

The direct numerical simulation of turbulence (DNS) has become a method of outmost importance for the investigation of turbulence physics, and its relevance is constantly growing due to the increasing popularity of high-performance-computing techniques. In the present work, the DNS approach is discussed mainly with regard to turbulent shear flows of incompressible fluids with constant properties. A body of literature is reviewed, dealing with the numerical integration of the Navier-Stokes equations, results obtained from the simulations, and appropriate use of the numerical databases for a better understanding of turbulence physics. Overall, it appears that high-performance computing is the only way to advance in turbulence research through the front of the direct numerical simulation.


Author(s):  
Francesco Balduzzi ◽  
Alessandro Bianchini ◽  
Giovanni Ferrara ◽  
David Marten ◽  
George Pechlivanoglou ◽  
...  

Due to the rapid progress in high-performance computing and the availability of increasingly large computational resources, Navier-Stokes computational fluid dynamics (CFD) now offers a cost-effective, versatile and accurate means to improve the understanding of the unsteady aerodynamics of Darrieus wind turbines and deliver more efficient designs. In particular, the possibility of determining a fully resolved flow field past the blades by means of CFD offers the opportunity to both further understand the physics underlying the turbine fluid dynamics and to use this knowledge to validate lower-order models, which can have a wider diffusion in the wind energy sector, particularly for industrial use, in the light of their lower computational burden. In this context, highly spatially and temporally refined time-dependent three-dimensional Navier-Stokes simulations were carried out using more than 16,000 processor cores per simulation on an IBM BG/Q cluster in order to investigate thoroughly the three-dimensional unsteady aerodynamics of a single blade in Darrieus-like motion. Particular attention was payed to tip losses, dynamic stall, and blade/wake interaction. CFD results are compared with those obtained with an open-source code based on the Lifting Line Free Vortex Wake Model (LLFVW). At present, this approach is the most refined method among the “lower-fidelity” models and, as the wake is explicitly resolved in contrast to BEM-based methods, LLFVW analyses provide three-dimensional flow solutions. Extended comparisons between the two approaches are presented and a critical analysis is carried out to identify the benefits and drawbacks of the two approaches.


Author(s):  
Francesco Balduzzi ◽  
David Marten ◽  
Alessandro Bianchini ◽  
Jernej Drofelnik ◽  
Lorenzo Ferrari ◽  
...  

Due to the rapid progress in high-performance computing and the availability of increasingly large computational resources, Navier–Stokes (NS) computational fluid dynamics (CFD) now offers a cost-effective, versatile, and accurate means to improve the understanding of the unsteady aerodynamics of Darrieus wind turbines and deliver more efficient designs. In particular, the possibility of determining a fully resolved flow field past the blades by means of CFD offers the opportunity to both further understand the physics underlying the turbine fluid dynamics and to use this knowledge to validate lower-order models, which can have a wider diffusion in the wind energy sector, particularly for industrial use, in the light of their lower computational burden. In this context, highly spatially and temporally refined time-dependent three-dimensional (3D) NS simulations were carried out using more than 16,000 processor cores per simulation on an IBM BG/Q cluster in order to investigate thoroughly the 3D unsteady aerodynamics of a single blade in Darrieus-like motion. Particular attention was paid to tip losses, dynamic stall, and blade/wake interaction. CFD results are compared with those obtained with an open-source code based on the lifting line free vortex wake model (LLFVW). At present, this approach is the most refined method among the “lower-fidelity” models, and as the wake is explicitly resolved in contrast to blade element momentum (BEM)-based methods, LLFVW analyses provide 3D flow solutions. Extended comparisons between the two approaches are presented and a critical analysis is carried out to identify the benefits and drawbacks of the two approaches.


PAMM ◽  
2015 ◽  
Vol 15 (1) ◽  
pp. 495-496 ◽  
Author(s):  
Lennart Schneiders ◽  
Jerry H. Grimmen ◽  
Matthias Meinke ◽  
Wolfgang Schröder

Sign in / Sign up

Export Citation Format

Share Document