Dynamic Load Balancing Techniques for Particulate Flow Simulations

Rettinger, Christoph; Rüde, Ulrich

doi:10.3390/computation7010009

Cited by 9 publications

(6 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fortunately, the framework and the main ideas could also be transferred to most platforms as similar architectures are becoming popular, although much work is needed from the coding point of view. In addition, some dynamic load balance methods might be incorporated into the two‐layer DD method to further improve the load balance for extreme‐scale parallel computing.…”

Section: Resultsmentioning

confidence: 99%

Virtual process engineering on a three‐dimensional circulating fluidized bed with multiscale parallel computation

Liu

et al. 2019

J Adv Manuf & Process

View full text Add to dashboard Cite

The combination of (quasi-)real-time simulation on industrial processes and virtual reality technologies may lead to a new paradigm of research and development in chemical engineering, that is, virtual process engineering (VPE). However, as the main engines of VPE, accurate and efficient simulation methods are still in urgent demand. In this paper, with substantial improvements to a discrete particle method (DPM), namely, energy-minimization multiscale (EMMS)-DPM, such an engine was established for a typical multiphase system, the circulating fluid beds (CFBs).To improve the accuracy, the coupling of particle and fluid flow solvers was updated to be more sensitive to local flow structures, and different drag laws were used for different flow regimes. To speed up the computation, heterogeneous central processing unit (CPU)-graphics processing unit (GPU) computing is developed for solving the fluids and particles. A two-layer domain decomposition method is developed to improve the load balance for real applications with complex geometries. Thus, a computer virtual experiment on a three-dimensional (3D) full-loop CFB has been achieved by simulating 1.27 × 10 11 real particles with 1.27 × 10 8 coarse-grained particles at the speed of 1.5 × 10 7 particle updates per GPU per second on 135 NVIDA K80 GPUs. To our knowledge, this is the largest-scale and highest-performance DPM simulation of a 3D full-loop CFB in terms of the computational particles used. It is a strong indication that VPE can be realized for industrial systems in the near future.K E Y W O R D S discrete particle method, energy-minimization multiscale model, three-dimensional circulating fluidized bed, virtual process engineering

show abstract

Section: Resultsmentioning

confidence: 99%

Virtual process engineering on a three‐dimensional circulating fluidized bed with multiscale parallel computation

Liu

et al. 2019

J Adv Manuf & Process

View full text Add to dashboard Cite

show abstract

“…Thus, to avoid unnecessary idle times of some MPI processes, we developed novel load balancing techniques to further increase the performance of these coupled simulations in Ref. 93. The previously presented coupling functionalities, combined with our efficient fluid and particle simulation methods, allow us to investigate large particulate systems numerically on a massively parallel scale.…”

Section: Geometrically Fully Resolved Simulationsmentioning

confidence: 99%

waLBerla: A block-structured high-performance framework for multiphysics simulations

Bauer

Eibl

Godenschwager

et al. 2021

Computers & Mathematics with Applications

Self Cite

View full text Add to dashboard Cite

Programming current supercomputers efficiently is a challenging task. Multiple levels of parallelism on the core, on the compute node, and between nodes need to be exploited to make full use of the system. Heterogeneous hardware architectures with accelerators further complicate the development process. waLBerla addresses these challenges by providing the user with highly efficient building blocks for developing simulations on block-structured grids. The block-structured domain partitioning is flexible enough to handle complex geometries, while the structured grid within each block allows for highly efficient implementations of stencil-based algorithms. We present several example applications realized with waLBerla, ranging from lattice Boltzmann methods to rigid particle simulations. Most importantly, these methods can be coupled together, enabling multiphysics simulations. The framework uses meta-programming techniques to generate highly efficient code for CPUs and GPUs from a symbolic method formulation. To ensure software quality and performance portability, a continuous integration toolchain automatically runs an extensive test suite encompassing multiple compilers, hardware architectures, and software configurations.

show abstract

“…This limits the number of particles in a simulation based on the available computational resources. In the last two decades, parallel computing has helped researchers simulate granular systems containing millions of particles [4,[8][9][10][11][12][13]. To this end, several parallel DEM softwares using multiple central processing units (CPU) to carry-out a single simulation have been developed [5,9,12,[14][15][16].…”

Section: Introductionmentioning

confidence: 99%

“…Cintra et al [21,22] used a recursive coordinate bisection in a simulation of hopper discharge and landslide using the DEMOOP software. The performances of different load-balancing algorithms were compared in other works [10,11,23]. Markauskas and Kačeniauskas [23] simulated the discharge of 5.1 M spherical particles on 128-2048 cores and reported a speed-up of 1785 on 2048 cores.…”

Section: Introductionmentioning

confidence: 99%

Load-Balancing Strategies in Discrete Element Method Simulations

Golshan

Blais

2021

Processes

View full text Add to dashboard Cite

In this research, we investigate the influence of a load-balancing strategy and parametrization on the speed-up of discrete element method simulations using Lethe-DEM. Lethe-DEM is an open-source DEM code which uses a cell-based load-balancing strategy. We compare the computational performance of different cell-weighing strategies based on the number of particles per cell (linear and quadratic). We observe two minimums for particle to cell weights (at 3, 40 for quadratic, and 15, 50 for linear) in both linear and quadratic strategies. The first and second minimums are attributed to the suitable distribution of cell-based and particle-based functions, respectively. We use four benchmark simulations (packing, rotating drum, silo, and V blender) to investigate the computational performances of different load-balancing schemes (namely, single-step, frequent and dynamic). These benchmarks are chosen to demonstrate different scenarios that may occur in a DEM simulation. In a large-scale rotating drum simulation, which shows the systems in which particles occupy a constant region after reaching steady-state, single-step load-balancing shows the best performance. In a silo and V blender, where particles move in one direction or have a reciprocating motion, frequent and dynamic schemes are preferred. We propose an automatic load-balancing scheme (dynamic) that finds the best load-balancing steps according to the imbalance of computational load between the processes. Furthermore, we show the high computational performance of Lethe-DEM in the simulation of the packing of 108 particles on 4800 processes. We show that simulations with optimum load-balancing need ≈40% less time compared to the simulations with no load-balancing.

show abstract

Dynamic Load Balancing Techniques for Particulate Flow Simulations

Cited by 9 publications

References 54 publications

Virtual process engineering on a three‐dimensional circulating fluidized bed with multiscale parallel computation

Virtual process engineering on a three‐dimensional circulating fluidized bed with multiscale parallel computation

waLBerla: A block-structured high-performance framework for multiphysics simulations

Load-Balancing Strategies in Discrete Element Method Simulations

Contact Info

Product

Resources

About