scholarly journals Parallel performance for a real time Lattice Boltzmann code

2018 ◽  
Vol 173 ◽  
pp. 237-258 ◽  
Author(s):  
Mark A. Woodgate ◽  
George N. Barakos ◽  
Rene Steijl ◽  
Gavin J. Pringle
2015 ◽  
Vol 8 (4) ◽  
pp. 405-414 ◽  
Author(s):  
M. Amirul Islam Khan ◽  
Nicolas Delbosc ◽  
Catherine J. Noakes ◽  
Jonathan Summers

Author(s):  
Simone Ferrari ◽  
Simone Ambrogio ◽  
Andrew J Narracott ◽  
Adrian Walker ◽  
Paul D Morris ◽  
...  

Abstract Medical device design for personalised medicine requires sophisticated tools for optimisation of biomechanical and biofluidic devices. This paper investigates a new real-time tool for simulating structural and fluid scenarios - ANSYS Discovery Live - and we evaluate its capability in the fluid domain through benchmark flows that all involve steady state flow at the inlet and zero pressure at the outlet. Three scenarios are reported: i. Laminar flow in a straight pipe, ii. vortex shedding from the Karman Vortex, and iii. nozzle flows as characterised by an FDA benchmark geometry. The solver uses a Lattice Boltzmann method requiring a high performance GPU (nVidiaGTX1080, 8GB RAM). Results in each case were compared with the literature and demonstrated credible solutions, all delivered in near real-time: i. The straight pipe delivered parabolic flow after an appropriate entrance length (plug flow inlet conditions), ii. the Karman Vortex demonstrated appropriate vortex shedding as a function of Reynolds number, characterised by Strouhal number in both the free field and within a pipe, and ii the FDA benchmark geometry generated results consistent with the literature in terms of variation of velocity along the centreline and in the radial direction, although deviation from experimental validation was evident in the sudden expansion section of the geometry. This behaviour is similar to previous reported results from Navier-Stokes solvers. A cardiovascular stenosis example is also considered, to provide a more direct biomedical context. The current software framework imposes constraints on inlet/outlet boundary conditions, and only supports limited control of solver discretization without providing full field vector flow data outputs. Nonetheless, numerous benefits result from the interactive interface and almost-real-time solution, providing a tool that may help to accelerate the arrival of improved patient-specific medical devices.


1997 ◽  
Vol 08 (04) ◽  
pp. 869-877 ◽  
Author(s):  
Giorgio Amati ◽  
Sauro Succi ◽  
Renzo Piva

High resolution lattice-Boltzmann simulations of turbulent channel flow on the Quadrics parallel machine are presented. The parallel performance is discussed together with some preliminary results concerning the vorticity structures which appear near the wall layer and their influence on the scaling laws.


2019 ◽  
Vol 9 (10) ◽  
pp. 2000
Author(s):  
Liangjun Wang ◽  
Xiaoxiao Zhang ◽  
Wenhao Zhu ◽  
Kangle Xu ◽  
Weiguo Wu ◽  
...  

The lattice Boltzmann method (LBM) is an important numerical algorithm for computational fluid dynamics. This study designs a two-layer parallel model for the Sunway TaihuLight supercomputer SW26010 many-core processor, which implements LBM algorithms and performs optimization. Numerical experiments with different problem sizes proved that the proposed model has better parallel performance and scalability than before. In this study, we performed numerical simulations of the flows around the two-dimensional (2D) NACA0012 airfoil, and the results of a series of flows around the different angles of attack were obtained. The results of the pressure coefficient and lift coefficient were in good agreement with those in the literature.


Author(s):  
Aditya C. Velivelli ◽  
Kenneth M. Bryden

The reduction of computation times is an important aspect of interactive computational fluid dynamics simulations. The lattice Boltzmann method has proved to be an important technique for the numerical solution of partial differential equations because it has nearly ideal scalability on parallel computers for many applications. Utilizing the two-dimensional diffusion equation, Tt=μ(Txx+Tyy), this paper examines the parallel performance for the lattice Boltzmann method and the alternating direction implicit (ADI) method. In this study for 50 time steps the non-cache optimized parallel lattice Boltzmann method was on average two times faster than the parallel ADI method. The cache optimized parallel lattice Boltzmann was on average seven times faster than the parallel ADI method.


Sign in / Sign up

Export Citation Format

Share Document