A high-throughput hybrid task and data parallel Poisson solver for large-scale simulations of incompressible turbulent flows on distributed GPUs

Zolfaghari, Hadi; Obrist, Dominik

doi:10.1016/j.jcp.2021.110329

Cited by 13 publications

(14 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2010 b ; Zolfaghari et al. 2019; Zolfaghari & Obrist 2021). The solver has been exhaustively validated and used to study various transitional flows (Henniger, Kleiser & Meiburg 2010 a ; Obrist, Henniger & Kleiser 2012; John, Obrist & Kleiser 2014, 2016).…”

Section: Direct Numerical Simulationmentioning

confidence: 99%

“…We model the systolic flow within a 3-D model, which involves fully open leaflets following the kinematic assumptions given in and Bellofiore et al (2011). We make use of our in-house multi-GPU data-parallel incompressible Navier-Stokes solver (Zolfaghari & Obrist 2021) for generating the DNS data at very high resolution (a total of 337 644 801 grid points were used). This simulation was completed in 3 days on 20 GPUs of the Cray XC40/50 (Piz Daint).…”

Section: Scope and Structure Of The Present Workmentioning

confidence: 99%

“…where U 0 = 0.75 m s −1 (one half of the inflow velocity at peak flow rate,), L 0 = 3 × r r = 36 mm (three times the aortic root radius) and ν = 2.7 × 10 −6 m 2 s −1 are the velocity scale, length scale and the kinematic viscosity (ν = ρ/μ), respectively. The equations (2.1a,b) are solved using a sixth-order finite-difference scheme in space on a staggered Cartesian grid, and a third-order explicit low-storage Runge-Kutta scheme in time (Henniger et al 2010b;Zolfaghari & Obrist 2021). The solver has been exhaustively validated and used to study various transitional flows (Henniger, Kleiser & Meiburg 2010a;Obrist, Henniger & Kleiser 2012;John, Obrist & Kleiser 2014.…”

Section: Direct Numerical Simulationmentioning

confidence: 99%

“…A highly resolved numerical simulation using degrees of freedom in a cuboid domain of size is performed. Such a high spatial resolution is set to deal with the non-conforming geometry of the leaflets, and also to resolve the spatio-temporal instability waves and their interactions (refer to the effect of grid resolution in Zolfaghari & Obrist 2021). Grid stretching is applied in all three directions, so that a grid resolution similar to the one reported by Zolfaghari & Obrist (2019) is achieved near the leaflets (31 grid points are placed along each leaflet's thickness length ).…”

Section: Direct Numerical Simulationmentioning

confidence: 99%

“…(2011). We make use of our in-house multi-GPU data-parallel incompressible Navier–Stokes solver (Zolfaghari & Obrist 2021) for generating the DNS data at very high resolution (a total of 337 644 801 grid points were used). This simulation was completed in 3 days on 20 GPUs of the Cray XC40/50 ().…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Sensitivity and downstream influence of the impinging leading-edge vortex instability in a bileaflet mechanical heart valve

et al. 2022

Self Cite

View full text Add to dashboard Cite

Bileaflet mechanical heart valves (BMHV) create unphysiological turbulent flow. Such turbulent flow involves multiple instability mechanisms interacting with one another in a confined geometry. For instance, an impinging leading-edge vortex (ILEV) instability creates disturbances at the leading edge of the valve leaflets, while potentially promoting turbulence downstream of the BMHV (Zolfaghari and Obrist, Phys. Rev. Fluids, vol. 4, 2019). In this article, we use adjoint-based methods to study the structural sensitivity of the ILEV instability in the BMHV, and to quantify the role of this instability in the maximum disturbance energy growth in the wake of the BMHV. We first present a direct numerical simulation to show the effect of the ILEV instability on the turbulent flow in the wake of the valve. Second, we perform a modal linear stability analysis on a two-dimensional subdomain attached to the leading edge of one leaflet. We investigate the sensitivity of the global modes associated with this flow using their adjoints, and then show a passive control scenario using a local feedback source. This results in a partial improvement in the flow oscillations downstream of the leaflets. We finally present a non-modal approach to identify the optimal initial conditions for achieving maximum energy growth at arbitrary locations. We show that, for sufficiently large times, the optimal initial condition for highest energy growth in the wake points at the leading edge, which includes the ILEV instability. Our study illustrates that an improved leading-edge shape can effectively reduce turbulence in the wake of the BMHV.

show abstract