||Uncertainty Quantification for Turbulent Mixing Simulations
||444, 5th International Conference of Numerical Modeling of Space Plasma Flows (ASTRONUM 2010)
||Kaman, T.; Glimm, J.; Sharp, D. H.
||We have achieved validation in the form of simulation-experiment agreement
for Rayleigh-Taylor turbulent mixing rates (known as α)
over the past decade. The problem was first posed sixty years ago.
Recent improvements in our simulation
technology allow sufficient precision
to distinguish between mixing rates for different
experiments. We explain the sensitivity and non-universality of the
mixing rate. These
play a role in the difficulties experienced by many others
in efforts to compare experiment with simulation.
We analyze the role of initial conditions, which were not
recorded for the classical experiments of Youngs et al. Reconstructed
initial conditions with error bars are given. The time evolution
of the long and short wave length portions of the instability are
analyzed. We show that
long wave length perturbations are strong at t = 0, but are quickly
overcome by the rapidly growing short wave length perturbations.
These conclusions, based solely
on experimental data analysis, are in agreement with results from
theoretical bubble merger models and
numerical simulation studies but disagree with models based
on superposition of modes.