A performance analysis of ensemble averaging for high fidelity turbulence simulations at the strong scaling limit

Makarashvili, Vakhtang; Merzari, Elia; Obabko, Aleksandr; Siegel, A.; Fischer, Paul

doi:10.1016/j.cpc.2017.05.023

Cited by 12 publications

(13 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An interesting observation is discussed in [27]. The authors noted that the averaging over multiple realizations of the same turbulent flow (ensemble averaging) may provide higher accuracy compared to the conventional long-range simulation and averaging in time.…”

Section: Discussionmentioning

confidence: 99%

“…In case the flow considered is not statistically steady, the ensemble averaging must be performed. The time averaging of statistically steady flows can be combined with the ensemble averaging [27]. Instead of averaging in time over the single flow with time interval T A , m flows can be averaged over the interval T A /m ( Figure 2).…”

Section: Time-averaged and Ensemble-averaged Statisticsmentioning

confidence: 99%

See 1 more Smart Citation

An approach for accelerating incompressible turbulent flow simulations based on simultaneous modelling of multiple ensembles

Krasnopolsky

2018

Computer Physics Communications

View full text Add to dashboard Cite

The present paper deals with the problem of improving the efficiency of large scale turbulent flow simulations. The high-fidelity methods for modelling turbulent flows become available for a wider range of applications thanks to the constant growth of the supercomputers performance, however, they are still unattainable for lots of real-life problems. The key shortcoming of these methods is related to the need of simulating a long time integration interval to collect reliable statistics, while the time integration process is inherently sequential.The novel approach with modelling of multiple flow states is discussed in the paper. The suggested numerical procedure allows to parallelize the integration in time by the cost of additional computations. Multiple realizations of the same turbulent flow are performed simultaneously. This allows to use more efficient implementations of numerical methods for solving systems of linear algebraic equations with multiple right-hand sides, operating with blocks of vectors. The simple theoretical estimate for the expected simulation speedup, accounting the penalty of additional computations and the linear solver performance improvement, is presented. The two problems of modelling turbulent flows in a plain channel and in a channel with a matrix of wall-mounted cubes are used to demonstrate the correctness of the proposed estimates and efficiency of the suggested approach as a whole. The simulation speedup by a factor of 2 is shown.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Time-averaged and Ensemble-averaged Statisticsmentioning

confidence: 99%

An approach for accelerating incompressible turbulent flow simulations based on simultaneous modelling of multiple ensembles

Krasnopolsky

2018

Computer Physics Communications

View full text Add to dashboard Cite

show abstract

“…While the results presented in [1,4] demonstrate the proof of concept for the suggested approach to parallelize the simulations in time, several questions affecting overall simulation speedup are not discussed in detail in these publications. Among them is the problem of generation of multiple uncorrelated turbulent flow states to start averaging in time.…”

Section: Introductionmentioning

confidence: 99%

“…An approach to model incompressible turbulent flows, combining conventional time averaging with the ensemble averaging has been proposed in [1,4]. This approach allows to parallelize the simulations in time and replace the single simulation with long time averaging by multiple simulations with much shorter time averaging intervals.…”

Section: Introductionmentioning

confidence: 99%

“…This approach allows to parallelize the simulations in time and replace the single simulation with long time averaging by multiple simulations with much shorter time averaging intervals. The paper [4] suggests to perform each of these simulations independently and to obtain the simulation speedup by increasing the amount of utilized compute resources. This approach increases the corresponding computational costs for the simulation, but allows to speedup the calculations beyond the strong scalability limit.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Generation of Multiple Turbulent Flow States for the Simulations with Ensemble Averaging

2018

JSFI

View full text Add to dashboard Cite

The paper deals with the problem of improving the performance of high-fidelity incompressible turbulent flow simulations on high performance computing systems. The ensemble averaging approach, combining averaging in time together with averaging over multiple ensembles, allows to speedup the corresponding simulations by increasing the computing intensity of the numerical method (flops per byte ratio). The current paper focuses on further improvement of the proposed computational methodology, and particularly, on the optimization of procedure to generate multiple independent turbulent flow states.

show abstract

Modeling of low Mach number unsteady turbulent pipe flows

Di Nucci,

Michele,

Di Risio

2024

Meccanica

View full text Add to dashboard Cite

Under adiabatic conditions, and neglecting temperature variations due to entropy production, we present a set of Reynolds Averaged Navier–Stokes (RANS) equations for fluids of low compressibility, i.e., fluids in the liquid state. In the low Mach number limit, we specialize the RANS equations to the one-dimensional unsteady pipe flow, and we deduce the dimensionless number that plays a predominant role in the flow behavior. We reduce the system of equations to a linear damped wave equation, and use its analytical solution to investigate the propagation of large amplitude pressure waves in liquid-filled pipes (water hammer phenomenon). We test the model reliability by comparing the analytical solution of the proposed model against experimental data available in the literature.

show abstract

A performance analysis of ensemble averaging for high fidelity turbulence simulations at the strong scaling limit

Cited by 12 publications

References 11 publications

An approach for accelerating incompressible turbulent flow simulations based on simultaneous modelling of multiple ensembles

An approach for accelerating incompressible turbulent flow simulations based on simultaneous modelling of multiple ensembles

Generation of Multiple Turbulent Flow States for the Simulations with Ensemble Averaging

Modeling of low Mach number unsteady turbulent pipe flows

Contact Info

Product

Resources

About