Accelerating the FlowSimulator: Profiling and scalability analysis of an industrial-grade CFD-CSM toolchain

Huismann, Immo; Reimer, Lars; Strobl, Severin; Eichstädt, Jan; Tschüter, Ronny; Rempke, Arne; Einarsson, Gunnar

doi:10.23967/coupled.2021.008

Cited by 5 publications

(4 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This enables the meshing of the grain geometry and the enclosed fluid volume with a conformal boundary as the initial interface between. For the handling of mesh data and coupling to external programs, the DLR FlowSimulator DataManager environment is used [37]. Originally designed to unify flow simulation capabilities with Python APIs, this framework is used here for the extensive import/export capabilities as well as simple data access as NumPy structures [38] for efficient calculations.…”

Section: Meshingmentioning

confidence: 99%

Hybrid Rocket Engine Burnback Simulations Using Implicit Geometry Descriptions

Zeriadtke,

Martin,

Wartemann

2024

Aerospace

View full text Add to dashboard Cite

The performance of hybrid rocket engines is significantly influenced by the fuel geometry. Burnback simulations, to determine the fuel surface and fluid volume, are therefore an important tool for preliminary design. This work presents a method for the simulation of spatially constant burn-ups on arbitrary geometries. An implicit surface definition by means of a signed distance function is used to represent the fluid volume and the fuel block on tetrahedral meshes. Two methods each are used to determine the fluid volume and the burning surface. The first method is based on a direct integration of the signed distance function with the Heaviside function or the Dirac delta distribution, respectively. The second method linearly interpolates the position of an isosurface and thus reconstructs the fuel surface. Both methods are compared and validated with analytical results of four example geometries. Both calculations of the fluid volume and the calculation of the surface content with the interpolation method are characterized as first-order methods. With practicable mesh resolutions of one million computational cells, errors below two percent can be achieved. With the interpolation method, numerical meshes can also be exported for any time points of the burn. Finally, the application of the program to the fuel geometry of the Viserion hybrid rocket engine is demonstrated.

show abstract

Section: Meshingmentioning

confidence: 99%

Hybrid Rocket Engine Burnback Simulations Using Implicit Geometry Descriptions

Zeriadtke,

Martin,

Wartemann

2024

Aerospace

View full text Add to dashboard Cite

show abstract

“…Unlike its conventional use of repairing the bad quality cells (e.g. negative volumes) arising from the Radial Basis Functions (RBF) method [12], the elastic analogy method serves as the primary deformation method in this study, providing a more robust mesh deformation method and offering potential scalability on HPC systems. Following the mesh deformation, the flow is recalculated based on the modified mesh, using the previous solution as a initial condition.…”

Section: Partitioningmentioning

confidence: 99%

“…In previous studies, the partitioning of the mesh was identified as a bottleneck of the simulation chain [12]. Not only did the runtime of the partitioning increase with the number of processes, the memory requirements also became prohibitive.…”

Section: High-lift Crm: Partitioningmentioning

confidence: 99%

“…6.4, and with the flow solver CODA) and because the dies with shared L3 caches in CARO's architecture consist of four cores, four threads are used per process. This raises the amount of available memory per process compared to previous studies by a factor of four [12]. Figure 6 depicts the runtimes of the partitioning process decomposed into three phases: the RCB prepartitioning, the extraction of the graph, and lastly the graph partitioning with ParMETIS using the default ParMETIS V3 PartGeomKway algorithm.…”

Section: High-lift Crm: Partitioningmentioning

confidence: 99%

See 1 more Smart Citation

Accelerating the FlowSimulator: Improvements in FSI Simulations for the HPC Exploitation at Industrial Level

Cristofaro,

Fenske,

Huismann

et al. 2023

10th Edition of the International Conference on Computational Methods for Coupled Problems in Science and Engineering

View full text Add to dashboard Cite

High-performance computing systems enable the use of computationally intensive models in aircraft design simulations, but challenges arise with communication times and memory consumption as the number of cores scales up. This paper presents improvements to the mesh partitioner and mesh deformation methods for highly parallelized simulations within the FlowSimulator framework. The results are obtained for a simple wing test case, a full-aircraft configuration and an aircraft research model. Improvements to the mesh partitioner are presented that allow for even larger and more parallelized simulations than was previously possible in the framework. In addition, scaling results from a mesh deformation method based on the elastic analogy are presented, which shows superior scalability compared to the conventional radial basis function method. This improves the overall suitability of the toolchain for large parallelization and highlights its potential for industrial applications. The results demonstrate a step forward towards an optimal exploitation of high performance computing systems for solving coupled simulations in the aerospace industry.

show abstract

A Look at Performance and Scalability of the GPU Accelerated Sparse Linear System Solver Spliss

Mohnke,

Wagner

2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

A significant part in computational fluid dynamics (CFD) simulations is the solving of large sparse systems of linear equations resulting from implicit time integration of the Reynolds-averaged Navier-Stokes (RANS) equations. The sparse linear system solver Spliss aims to provide a linear solver library that, on the one hand, is tailored to these requirements of CFD applications but, on the other hand, independent of the particular CFD solver. Spliss allows leveraging a range of available HPC technologies such as hybrid CPU parallelization and the possibility to offload the computationally intensive linear solver to GPU accelerators, while at the same time hiding this complexity from the CFD solver.This work highlights the steps taken to establish multi-GPU capabilities for the Spliss solver allowing for efficient and scalable usage of large GPU systems. In addition, this work evaluates performance and scalability on CPU and GPU systems using a representative CODA test case as an example. CODA is the CFD software being developed as part of a collaboration between the French Aerospace Lab ONERA, the German Aerospace Center (DLR), Airbus, and their European research partners. CODA is jointly owned by ONERA, DLR and Airbus. The evaluation examines and compares performance and scalability in a strong scaling approach on Nvidia A100 GPUs and the AMD Rome architecture.

show abstract

Accelerating the FlowSimulator: Profiling and scalability analysis of an industrial-grade CFD-CSM toolchain

Cited by 5 publications

References 0 publications

Hybrid Rocket Engine Burnback Simulations Using Implicit Geometry Descriptions

Hybrid Rocket Engine Burnback Simulations Using Implicit Geometry Descriptions

Accelerating the FlowSimulator: Improvements in FSI Simulations for the HPC Exploitation at Industrial Level

A Look at Performance and Scalability of the GPU Accelerated Sparse Linear System Solver Spliss

Contact Info

Product

Resources

About