Answers are grounded in this paper and cite their sources.
Using a formalism adapted to slow spatio-temporal evolutions of phase perturbations to a perfect roll pattern, we show that "zigzags" which relax in a diffusive way for the case with "no-slip" (or "rigid") boundaries are replaced by Busse's oscillatory mode for the case with "stress-free" boundaries. The diffusion coefficient D_⊥ for zigzags, which is determined analytically using a perturbation expansion, has the form ${D}_{{⊥}}={{{ξ}}₀²}{{{τ}}₀}[{q{-}{q}c}{{q}c}+{N(P)}{{R}₂(P)}{R{-}{R}ᶜ}{{R}ᶜ}],$ where $R [={R}ᶜ+{R}₂(P){{ε}}²+{⋯}]$ is the Rayleigh number, $P$ is the Prandtl number, ${{ξ}}₀$ is the coherence length, and ${{τ}}₀$ is the relaxation time; ${{{ξ}}₀²}{{{τ}}₀}={5.685}{(1.95+{P}^{{-}1})}$, $N(P)=0.166+23.04{P}^{{-}1}+6.196{P}^{{-}2}$, ${R}₂(P)=10.76{-}0.073{P}^{{-}1}+0.128{P}^{{-}2}$, and ${R}ᶜ=1707.8$ at the critical wave vector ${q}c=3.116$ for no-slip boundaries. The curvature is seen to induce a large-scale flow which tends to convect the roll pattern. This flow is forbidden for concentric rolls in cylindrical geometry. This leads to a new wavelength-selection criterion which no longer corresponds to D_⊥=0. The corrected criterion is calculated for both no-slip and stress-free boundaries and is seen to behave in a similar way at the limit P→0 in the two cases. Our findings are in quantitative agreement with earlier numerical results and also with more recent approaches using the amplitude-equation formalism.
No takes yet. Share an insight, caveat, or question.
Manneville et al. (1983) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: