Published online by Cambridge University Press: 26 October 2017
We perform an experimental study of turbulent Rayleigh–Bénard convection up to very high Rayleigh number, $10^{8}<Ra<10^{14}$, in a cylindrical aspect ratio one cell, 30 cm in height, filled with cryogenic helium gas. We monitor temperature fluctuations in the convective flow with four small (0.2 mm) sensors positioned in pairs 1.5 cm from the sidewalls and 2.5 cm vertically apart and symmetrically around the mid-height of the cell. Based on one-point and two-point correlations of the temperature fluctuations, we determine different types of Reynolds numbers, $\mathit{Re}$, associated with the large-scale circulation (LSC). We observe a transition between two types of $\mathit{Re}(\mathit{Ra})$ scaling around $\mathit{Ra}=10^{10}{-}10^{11}$, which is accompanied by a scaling change of the skewness of the probability distribution functions (PDFs) of the temperature fluctuations. The $\mathit{Re}(\mathit{Ra})$ dependencies measured near the sidewall at Prandtl number $\mathit{Pr}\sim 1$ are consistent with the $\mathit{Ra}^{4/9}\mathit{Pr}^{-2/3}$ scaling above the transition, while for $\mathit{Ra}<10^{10}$, the $\mathit{Re}(\mathit{Ra})$ dependencies are steeper. It seems likely that this change in $\mathit{Re}(\mathit{Ra})$ scaling is linked to the previously reported change in the Nusselt number $\mathit{Nu}(\mathit{Ra})$ scaling. This behaviour is in agreement with independent cryogenic laboratory experiments with $\mathit{Pr}\sim 1$, but markedly different from the $\mathit{Re}$ scaling obtained in water experiments ($\mathit{Pr}\sim 3.3{-}5.6$). We discuss the results in comparison with different versions of the Grossmann–Lohse theory.