Toward one-dimensional turbulence subgrid closure for large-eddy simulation A Dissertation Defense by Randall J. McDermott Department of Chemical Engineering The University of Utah

Outline • Motivation and background • Large-eddy simulation (LES) • One-dimensional turbulence (ODT)

• Coupling LES and ODT • Results • Serial • Parallel

• Future work

The C-SAFE ASCI problem • Center for the Simulation of Accidental Fires and Explosions • Accelerated Strategic Computing Initiative

Soot Chemistry

Air TURBULENCE! Products of combustion Radiation

Fuel http://flash.uchicago.edu/~zingale/rt_gallery/rt_gallery.html

Courtesy Schönbucher, Stuttgart

Navier-Stokes physics

Claude Louis Marie Henri Navier (1785-1836)

George Gabriel Stokes (1819-1903)

Outline • Motivation and background • Large-eddy simulation (LES) • One-dimensional turbulence (ODT)

• Coupling LES and ODT • Results • Serial • Parallel

• Future work

The symphony of turbulence

large eddies (bass) universal eddies (mid-range)

A.N. Kolmogorov (1903-1987) Reynolds number:

viscous eddies (treble)

Filtered Navier-Stokes

Filtered Navier-Stokes

Finite-volume LES

Volume-averaged velocity

Surface stress due to fluctuating velocity

Outline • Motivation and background • Large-eddy simulation (LES) • One-dimensional turbulence (ODT)

• Coupling LES and ODT • Results • Serial • Parallel

• Future work

“Perhaps the holy grail of turbulence is the statistical resolution of all scales…” “[As] example[s] we cite the linear eddy model (LEM) and one-dimensional turbulence (ODT), which can be used as SGS models in LES. These models fall short of the ideal in that they involve empirical prescriptions, and the computational work increases as a power of the Reynolds number (albeit a smaller power than DNS).” Stephen B. Pope. Ten questions concerning the large-eddy simulation of turbulent flows. New Journal of Physics, 6 (2004) 35.

What is ODT?

The triplet map

16

16

12

12

8

8

4

4

0

0 0

4

8

12 Y

16

0

4

8

12 Y

16

ODT physics: When, where, and how big are the eddies?

A comparison of LES and ODT LES • • • • • •

Dynamic (time dependent) 3d (Poor Re scaling) Large scales No empiricism Poor near walls Model sgs mixing

ODT • • • • • •

Dynamic 1d (Excellent Re scaling) Small scales Empirical parameters Excellent near-wall model Explicit sgs mixing

A marriage made in heaven!

Outline • Motivation and background • Large-eddy simulation (LES) • One-dimensional turbulence (ODT)

• Coupling LES and ODT • Results • Serial • Parallel

• Future work

Coupling the grids

Grid 1 example

Generating the sgs stress

The LESODT algorithm • Compute the LES sgs stress via ODT • Advance the LES equations • Adjust the ODT field

Reconstruction: Initial state

Reconstruction: Step 1

Reconstruction: Step 3

Reconstruction: Step 4

Outline • Motivation and background • Large-eddy simulation (LES) • One-dimensional turbulence (ODT)

• Coupling LES and ODT • Results • Serial • Parallel

• Future work

Relating 1d and 3d spectra

ODT spectra time increasing

LES 3d spectra

time increasing

LES 3d spectra

ODT spectra

energy flux

An “energy leak”

What causes the leak?

The spectral “dip”

Harlow and Welch implied filter classic 2nd order staggered scheme (1965)

sharp spectral filter normalized box filter CBC with Dynamic Smagorinsky

Serial scaling

Outline • Motivation and background • Large-eddy simulation (LES) • One-dimensional turbulence (ODT)

• Coupling LES and ODT • Results • Serial • Parallel

• Future work

A single periodic patch

Patch overlapping

Patch slitting error?

Multiprocessor results

Outline • Motivation and background • Large-eddy simulation (LES) • One-dimensional turbulence (ODT)

• Coupling LES and ODT • Results • Serial • Parallel

• Future work

A “to do” list • Fix energy leak: variance reconstruction • Adaptive ODT sample distribution • Reconcile isotropic vs. anisotropic dissipation • Understand diffusive splitting error • Dynamic ODT equilibrium analysis to nail down rate constant • Parallel algorithm verification • Near-wall validation (complex geometry)

Acknowledgements DOE (DE-FG02-97ER25308)

Sandia National Laboratories Combustion Research Facility

Thanks to my committee • • • • • •

Phil Smith Adel Sarofim Rod Schmidt Aaron Fogelson Steve Krueger Alan Kerstein

Contributors • • • •

Zhaosheng Gao (U. Utah) Tom Smith (Sandia) Steve de Bruyn Kops (U. Mass.) Stanislav Borodai (U. Utah)

• Many others: P. McMurtry, J. Klewicki, S. Kumar, Jeremy Thornock, James Sutherland, Bhushan P., Gautham K., Jennifer Spinti, Mohit Tandon, Niveditha K., David Lignell, Devin Yeates

Dedication

To my grandfather, Les Smith, who inspired my love of science

Questions? “I am an old man now, and when I die and go to heaven there are two matters on which I hope for enlightenment. One is quantum electrodynamics, and the other is the turbulent motion of fluids. And about the former I am rather optimistic.” ―Horace Lamb (1849 - 1934)

Toward one-dimensional turbulence subgrid closure for ...

http://flash.uchicago.edu/~zingale/rt_gallery/rt_gallery.html. Products of combustion. Air. Fuel ... Center for the Simulation of Accidental. Fires and Explosions.

2MB Sizes 3 Downloads 224 Views

Recommend Documents

No documents