Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-16T19:11:02.713Z Has data issue: false hasContentIssue false

On the universality of anomalous scaling exponents of structure functions in turbulent flows

Published online by Cambridge University Press:  05 January 2018

E.-W. Saw
Affiliation:
SPEC/IRAMIS/DSM, CEA, CNRS, University Paris-Saclay, CEA Saclay, 91191 Gif-sur-Yvette, FranceSchool of Atmospheric Sciences, Sun Yat-Sen University, Guangzhou, China
P. Debue
Affiliation:
SPEC/IRAMIS/DSM, CEA, CNRS, University Paris-Saclay, CEA Saclay, 91191 Gif-sur-Yvette, France
D. Kuzzay
Affiliation:
SPEC/IRAMIS/DSM, CEA, CNRS, University Paris-Saclay, CEA Saclay, 91191 Gif-sur-Yvette, France
F. Daviaud
Affiliation:
SPEC/IRAMIS/DSM, CEA, CNRS, University Paris-Saclay, CEA Saclay, 91191 Gif-sur-Yvette, France
B. Dubrulle*
Affiliation:
SPEC/IRAMIS/DSM, CEA, CNRS, University Paris-Saclay, CEA Saclay, 91191 Gif-sur-Yvette, France
*
†Email address for correspondence: berengere.dubrulle@cea.fr

Abstract

All previous experiments in open turbulent flows (e.g. downstream of grids, jets and the atmospheric boundary layer) have produced quantitatively consistent values for the scaling exponents of velocity structure functions (Anselmet et al., J. Fluid Mech., vol. 140, 1984, pp. 63–89; Stolovitzky et al., Phys. Rev. E, vol. 48 (5), 1993, R3217; Arneodo et al., Europhys. Lett., vol. 34 (6), 1996, p. 411). The only measurement of scaling exponents at high order (${>}6$) in closed turbulent flow (von Kármán swirling flow) using Taylor’s frozen flow hypothesis, however, produced scaling exponents that are significantly smaller, suggesting that the universality of these exponents is broken with respect to change of large scale geometry of the flow. Here, we report measurements of longitudinal structure functions of velocity in a von Kármán set-up without the use of the Taylor hypothesis. The measurements are made using stereo particle image velocimetry at four different ranges of spatial scales, in order to observe a combined inertial subrange spanning approximately one and a half orders of magnitude. We found scaling exponents (up to ninth order) that are consistent with values from open turbulent flows, suggesting that they might be in fact universal.

Information

Type
JFM Papers
Copyright
© 2018 Cambridge University Press 
Figure 0

Figure 1.Sketch of the apparatus. Velocity fields are observed in a two-dimensional area on a laser sheet centred on the centre of the experiment (equal distance from the impellers and on the axis of the cylindrical tank). Two cameras view the observation area at angles approximately $45^{\circ }$ from the laser sheet.

Figure 1

Table 1.Parameter space describing the cases considered in this paper. $\unicode[STIX]{x0394}x$ is the spatial resolution of our measurements; $\unicode[STIX]{x1D716}_{g}$ is the global non-dimensional energy dissipation (non-dimensionalized using the radius of tank ($R$) and $2\unicode[STIX]{x03C0}F$, where $F$ is the frequency of the impellers) measured through torques, while $\unicode[STIX]{x1D716}_{v}$ is the local non-dimensional energy dissipation estimated via the second-order structure function (see the text for information regarding how it is estimated); $\unicode[STIX]{x1D702}$ is the Kolmogorov scale (estimated using $\unicode[STIX]{x1D708}$ and $\unicode[STIX]{x1D716}_{v}$). Liquid temperature is $20\,^{\circ }\text{C}$ unless otherwise specified.

Figure 2

Figure 2.Longitudinal structure functions (up to ninth order) rescaled using the intermittent scaling relation. (a) Plots showing all data points from all cases (A though D). (b) Cropped structure functions where in each case, the segments of the data affected by measurement noise (small $r$) and by finite measurement volume (large $r$) are removed. (c) Same as (b) but only showing $S_{6}^{L}$ to $S_{9}^{L}$ (from bottom to top) and each curve is compensated using the global $\unicode[STIX]{x1D701}_{n}$ given in table 2. (d) Local slopes, $\unicode[STIX]{x0394}\text{log}_{10}(S_{n})/\unicode[STIX]{x0394}\text{log}_{10}(r)$ versus $r$ for case B. Solid lines from top to bottom are for $n=9,6,4$ respectively and they are smoothed by a running average method. The dashed line shows the respective un-smoothed data. The curve fitting range is selected as the intermediate part by removing the strongly varying or fluctuating parts at small and large $S_{3}$.

Figure 3

Figure 3.(a) Non-dimensionalized second-order structure function ($S_{2}/(\unicode[STIX]{x1D716}_{v}\unicode[STIX]{x1D708})^{1/2}$) from case A. Comparison with the Sirovich model. Inset: discrepancy between the two curves (ratio of $S_{2}$ to the model). Close agreement between the two implies an accurate estimate of $\unicode[STIX]{x1D716}_{v}$ (for case A). The experimental noise in $S_{2}$ is partially removed by subtracting a value of 0.2. (b) Third-order structure function compensated by power 1.

Figure 4

Table 2.Comparison of measured scaling exponents between this paper and previous experiments. Arneodo ESS: results from various open turbulent flows in Arneodo et al. (1996). This paper, ESS: results from this paper using ESS. This paper, global: results from this paper based on conjoined structure functions obtained at different Reynolds numbers. The values at highest orders marked with ‘$\ast$’ are likely unreliable due to stark inconsistency with local scalings of the corresponding structure functions (details in text). The values of $\unicode[STIX]{x1D701}_{3}$ are not shown since they are, by assumption of the methodologies, equal to unity in all cases.

Figure 5

Figure 4.(a) Comparing the consistency of the structure functions rescaled using K41 and the intermittent scaling relations. From top to bottom: blue circles show the eighth-order structure function ($S_{8}$) rescaled using the K41 scaling ($\unicode[STIX]{x1D701}_{8}=8/3\approx 2.667$); red circles show $S_{8}$ rescaled via the intermittent relation with $\unicode[STIX]{x1D701}_{8}=2.35$. Cyan dashed lines are best fits to each set of data in the range $r/\unicode[STIX]{x1D702}=2.4-2.6$. The intermittent case is found to show a higher general level of consistency (continuity) among the higher-order structure functions (the difference is insignificant at lower orders). (b) Local slopes, $\unicode[STIX]{x0394}\text{log}_{10}(S_{n})/\unicode[STIX]{x0394}\text{log}_{10}(S_{3})$ versus $S_{3}$ for case B. Solid lines from top to bottom are for $n=9,6,4$ respectively and they are smoothed by the running average method. The dashed line is the respective un-smoothed data. The ESS curve fitting range is selected as the intermediate part by removing the strongly varying or fluctuating parts at small and large $S_{3}$.

Figure 6

Table 3.Scaling exponents of the transversal structure functions of velocity. $\unicode[STIX]{x1D701}_{n}^{T}$ values that are not presented as ranges imply uncertainties of $\pm 0.005$, except the last column (Average) which shows simple averages of the mid-values of each row.

Figure 7

Figure 5.Statistical convergence of the structure functions. (a) Main figure and inset, we plot $C_{9}(v)/C_{9}(v_{0})$ versus $v/v_{0}$, where $C_{n}(v)=\int _{0}^{v}{v^{\prime }}^{n}P(v^{\prime })\,\text{d}z^{\prime }$, $v$ and $v^{\prime }$ are moduli of velocity differences, $P$ the PDF and $v_{0}$ is the value of $v$ where $P(v)$ first reaches a value of zero in our data. The curves (blue, gold, red, cyan) represent, from cases A to D respectively, the least converged points used to calculate $\unicode[STIX]{x1D701}_{n}$ (and retained in figure 2b,d). As the inset shows, $C_{9}(v)$ increase with $v$ before they start to saturate at large $v$. The main figure verifies that all data points used in our analysis of the scaling exponents do not vary more than 1 % even if additional sampling would increase $v_{0}$ by 10 %. (b) $|\unicode[STIX]{x1D6FF}u|^{9}P(\unicode[STIX]{x1D6FF}u)$ versus $\unicode[STIX]{x1D6FF}u$ corresponding to case B (gold) in the left panel, where $\unicode[STIX]{x1D6FF}u$ is the velocity difference.