Parallel Adaptive Boundary Layer Meshing for CFD Analysis

Ovcharenko, A. G.; Chitale, Kedar C.; Sahni, Onkar; Jansen, Kenneth E.; Shephard, Mark S.

doi:10.1007/978-3-642-33573-0_26

Cited by 11 publications

(14 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since each torus node is connected to its neighbor via a dedicated link, torus networks can yield high throughput for nearest-neighbor communication patterns. Therefore, they are well suited for applications such as anisotropic mesh adaptation in CFD solvers [17]. Torus networks also offer good path diversity because several distinct paths exist between any given source and destination node.…”

Section: Torus Networkmentioning

confidence: 99%

A case study in using massively parallel simulation for extreme-scale torus network codesign

Mubarak

Carothers

Ross

et al. 2014

Proceedings of the 2nd ACM SIGSIM Conference on Principles of Advanced Discrete Simulation

View full text Add to dashboard Cite

A high-bandwidth, low-latency interconnect will be a critical component of future exascale systems. The torus network topology, which uses multidimensional network links to improve path diversity and exploit locality between nodes, is a potential candidate for exascale interconnects.The communication behavior of large-scale scientific applications running on future exascale networks is particularly important and analytical/algorithmic models alone cannot deduce it. Therefore, before building systems, it is important to explore the design space and performance of candidate exascale interconnects by using simulation. We improve upon previous work in this area and present a methodology for modeling and simulating a high-fidelity, validated, and scalable torus network topology at a packet-chunk level detail using the Rensselaer Optimistic Simulation System (ROSS). We execute various configurations of a 1.3 million node torus network model in order to examine the effect of torus dimensionality on network performance with relevant HPC traffic patterns. To the best of our knowledge, these are the largest torus network simulations that are carried out at such a detailed fidelity. In terms of simulation performance, a 1.3 million node, 9-D torus network model is shown to process a simulated exascale-class workload of nearest-neighbor traffic with 100 million message injections per second per node using 65,536 Blue Gene/Q cores in a simulation run-time of only 25 seconds. We also demonstrate that massive-scale simulations are a critical tool in exascale system design since small-scale torus simulations are not always indicative of the network behavior at an exascale size. The take-away message from this case study is that massively parallel simulation is a key enabler for effective extreme-scale network codesign.

show abstract

Section: Torus Networkmentioning

confidence: 99%

A case study in using massively parallel simulation for extreme-scale torus network codesign

Mubarak

Carothers

Ross

et al. 2014

Proceedings of the 2nd ACM SIGSIM Conference on Principles of Advanced Discrete Simulation

View full text Add to dashboard Cite

show abstract

“…A few works have been conducted to adapt the boundary layer mesh. For instance, in [124], specific hybrid operators have been derived to refine the boundary layer mesh. This aspect is of crucial interest for transonic flows, where shocks may interact with the boundary layer.…”

Section: Mesh Adaptation For Turbulent Navier-stokes Equationsmentioning

confidence: 99%

A decade of progress on anisotropic mesh adaptation for computational fluid dynamics

Alauzet

Loseille

2016

Computer-Aided Design

156

View full text Add to dashboard Cite

Please cite this article as: Alauzet F, Loseille A. A decade of progress on anisotropic mesh adaptation for computational fluid dynamics. Computer-Aided Design (2015), http://dx. AbstractIn the context of scientific computing, the mesh is used as a discrete support for the considered numerical methods. As a consequence, the mesh greatly impacts the e ciency, the stability and the accuracy of numerical methods. The goal of anisotropic mesh adaptation is to generate a mesh which fits the application and the numerical scheme in order to achieve the best possible solution. It is thus an active field of research which is progressing continuously. This review article proposes a synthesis of the research activity of the INRIA Gamma3 team in the field of anisotropic mesh adaptation applied to inviscid flows in computational fluid dynamics since 2000. It shows the evolution of the theoretical and numerical results during this period. Finally, challenges for the next decade are discussed.

show abstract

“…Mesh size: 235 095 vertices, 1 310 082 tetrahedra a boundary faces 87 Anisotropic Mesh Adapt of high-order curved meshes [3,80,84,90] and parallel (adaptive) mesh generation [9,23,65,73,78].…”

Section: D Double Mach Reflectionmentioning

confidence: 99%

Unstructured Mesh Generation and Adaptation

Loseille¹

2017

Handbook of Numerical Analysis

View full text Add to dashboard Cite

We first describe the well established unstructured mesh generation methods as involved in the computational pipeline, from geometry definition to surface and volume mesh generation. These components are always a preliminary and required step to any numerical computations. From an historical point of view, the generation of fully unstructured mesh generation in 3D has been a real challenge so as to the design of robust and accurate second order schemes on such unstructured meshes. If the issue of generating volume meshes for geometries of any complexity is now mostly solved, the emergence of robust numerical schemes on unstructured meshes has paved the way to adaptivity. Indeed, unstructured meshes in contrast with structured or block structured grids have the necessary flexibility to control the discretization both in size and orientation.In the second part, we review the main components to perform adaptative computations: (i) anisotropic mesh prescription via a metric field tensor (ii) anisotropic error estimates, and (iii) anisotropic mesh generation. For each component, we focus on a particularly simple method to implement. In particular, we describe a simple but robust strategy for generating anisotropic meshes. Each adaptation entity, ie surface, volume or boundary layers, relies on a specific metric tensor field. The metric-based surface estimate is then used to control the deviation to the surface and to adapt the surface mesh. The volume estimate aims at controlling the interpolation error of a specific field of the flow.Several 3D examples issued from steady and unsteady simulations from systems of hyperbolic laws are presented. In particular, we show that despite the simplicity of the introduced adaptive meshing scheme a high level of anisotropy can be reached. This includes the direct prediction of the sonic boom of an aircraft by computing the flow from the cruise altitude to the ground, the interaction between shock waves and boundary layer, or the prediction of complex unsteady phenomena in 3D.

show abstract

Parallel Adaptive Boundary Layer Meshing for CFD Analysis

Cited by 11 publications

References 34 publications

A case study in using massively parallel simulation for extreme-scale torus network codesign

A case study in using massively parallel simulation for extreme-scale torus network codesign

A decade of progress on anisotropic mesh adaptation for computational fluid dynamics

Unstructured Mesh Generation and Adaptation

Contact Info

Product

Resources

About