On the wall-normal velocity variance in canonical wall-bounded turbulence
Pith reviewed 2026-05-18 11:25 UTC · model grok-4.3
The pith
Deviations in local shear stress explain most differences in wall-normal velocity variance across turbulent flows
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
Deviations in the local stress from the surface shear velocity U_τ account for almost all of the differences in wall-normal velocity variance observed across different canonical flows, including for plane Couette flow. Spectra along the spanwise wavenumber have a pronounced peak well described by the turbulent dissipation rate and the local shear stress. The dependence on local stress is attributed to wall-normal motions being predominately active per Townsend's attached eddy hypothesis. A semi-empirical fit matches the simulations across the lower half of the boundary layer and extrapolates to a value between 1.45 and 1.65 times the local shear stress in the high-Reynolds-number limit, with
What carries the argument
Local shear stress replacing surface U_τ as the scaling quantity for wall-normal velocity variance, because active wall-normal motions contribute directly to it
If this is right
- The semi-empirical fit reproduces the direct numerical simulation results across the lower half of the boundary layer.
- The extrapolated high-Reynolds-number limit is consistent with earlier predictions for the near-neutral atmospheric boundary layer.
- Small discrepancies in the variances remain due to dissimilar low-wavenumber contributions across flow configurations and wall-normal positions.
- Exact universality of the proportional constant is precluded by these discrepancies.
Where Pith is reading between the lines
- Turbulence closure models for wall-bounded flows may need to incorporate local rather than surface stress when predicting wall-normal velocity fluctuations.
- The remaining role of weak inactive wall-normal motions could be tested by isolating their spectral contribution in additional simulations.
- The same local-stress scaling might be examined in other statistics or in non-canonical geometries to check consistency.
Load-bearing premise
Wall-normal motions are predominately active per Townsend's attached eddy hypothesis under the assumption of constant turbulent shear stress.
What would settle it
A high-Reynolds-number simulation or experiment in one of the canonical flows showing wall-normal velocity variance clearly outside the 1.45–1.65 range relative to local shear stress, or showing that local-stress deviations no longer account for the variance differences between flows.
read the original abstract
The variance and spectra of wall-normal velocities are investigated for direct numerical simulations of turbulent flow in a channel, pipe, and zero-pressure-gradient boundary layer across a decade of friction Reynolds numbers. Spectra along the spanwise wavenumber have a pronounced peak well described by the turbulent dissipation rate and the local shear stress throughout the bottom half of the boundary layer. Deviations in the local stress from the surface shear velocity $U_\tau$ account for almost all of the differences in wall-normal velocity variance observed across different canonical flows, including for plane Couette flow. The dependence on the local stress is attributed to the fact that wall-normal motions are predominately `active' per Townsend's attached eddy hypothesis and directly contribute to the local shear stress, noting this hypothesis assumes simplified ideal conditions with constant turbulent shear stress. A semi-empirical fit applied to the Reynolds number dependence of the variance matches the simulations across the lower half of the boundary layer and aligns with observed values in the literature. The fit extrapolates to a value between 1.45 and 1.65 times the local shear stress in the high-Reynolds-number limit, consistent with previous predictions relative to $U_\tau$ including for the vertical velocity in the near-neutral atmospheric boundary layer. However, universality in the exact proportional constant is precluded by small discrepancies in the variances corresponding to dissimilarity in the low-wavenumber contributions across different flow configurations and wall-normal positions. We speculate the dissimilarity is due to relatively weak `inactive' wall-normal motions that are excluded from Townsend's original hypothesis.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper investigates the variance and spectra of wall-normal velocities in DNS of channel, pipe, and zero-pressure-gradient boundary layer flows across a decade of friction Reynolds numbers, along with comparisons to plane Couette flow. It claims that deviations of the local shear stress from the surface friction velocity U_τ account for nearly all observed differences in wall-normal velocity variance across these canonical configurations. Spectra along the spanwise wavenumber are shown to peak in a manner described by the turbulent dissipation rate and local shear stress in the bottom half of the layer. This dependence is attributed to wall-normal motions being predominantly 'active' per Townsend's attached-eddy hypothesis (noting the hypothesis's idealization of constant turbulent shear stress). A semi-empirical fit to the Reynolds-number dependence of the variance is presented that matches the simulations and extrapolates to a high-Re asymptotic value between 1.45 and 1.65 times the local shear stress, consistent with some prior predictions.
Significance. If the central claim holds, the work provides a unifying local-stress explanation for wall-normal velocity statistics that reconciles differences across channel, pipe, boundary-layer, and Couette flows, with potential implications for turbulence modeling and high-Re extrapolations relevant to atmospheric boundary layers. The DNS spectra and variance trends offer concrete support for the local-stress scaling. The manuscript also supplies a falsifiable semi-empirical prediction for the high-Re limit. However, the strength of these contributions is tempered by the need for fuller specification of the fit and clearer quantification of the active/inactive distinction outside the constant-stress idealization.
major comments (2)
- [Abstract and §4] Abstract and §4 (results on Reynolds-number dependence): the semi-empirical fit is load-bearing for the high-Re extrapolation claim (1.45–1.65 times local stress), yet its functional form, fitting procedure, and exact data-selection criteria (which wall-normal locations, which flows, which Re range) are not fully specified. This creates circularity because the fit is constructed to reproduce the simulated variance trends and is then invoked to predict the asymptotic constant.
- [§2 and §3.3] §2 (Townsend hypothesis discussion) and §3.3 (variance collapse analysis): the attribution of the local-stress scaling to 'active' motions relies on Townsend's attached-eddy hypothesis, which is derived under the idealization of constant turbulent shear stress. The DNS data span regions where total stress varies with y (linearly in channels, decreasing in boundary layers), and the paper does not quantify how much of the reported collapse survives when this y-dependence is removed from the normalization or how the residual inactive-motion discrepancies affect the claimed universality.
minor comments (2)
- [Figures 3–6] Figure captions and legends should explicitly list the wall-normal positions (e.g., y^+ or y/h ranges) and the precise definition of local shear stress used for each data set to improve reproducibility.
- [Notation section] Notation for local versus surface shear stress should be introduced once and used consistently; occasional mixing of symbols for U_τ and the local value can confuse readers.
Simulated Author's Rebuttal
We thank the referee for the careful reading and valuable comments on our manuscript. We have revised the paper to address the points raised regarding the specification of the semi-empirical fit and the discussion of Townsend's attached-eddy hypothesis. Our responses to the major comments are provided below.
read point-by-point responses
-
Referee: [Abstract and §4] Abstract and §4 (results on Reynolds-number dependence): the semi-empirical fit is load-bearing for the high-Re extrapolation claim (1.45–1.65 times local stress), yet its functional form, fitting procedure, and exact data-selection criteria (which wall-normal locations, which flows, which Re range) are not fully specified. This creates circularity because the fit is constructed to reproduce the simulated variance trends and is then invoked to predict the asymptotic constant.
Authors: We agree that additional details on the semi-empirical fit are necessary to strengthen the presentation. In the revised manuscript, we have expanded the description in §4 to include the explicit functional form (a logarithmic approach to the asymptotic constant), the fitting method using nonlinear least squares, and the precise data-selection criteria: wall-normal locations 0.15 < y/δ < 0.4 for channel, pipe, and boundary-layer flows at Re_τ > 1000. While the fit is calibrated to the DNS trends, the extrapolated range 1.45–1.65 is cross-checked against independent literature predictions for the high-Re limit, reducing the risk of circularity. A table of fitted coefficients and residuals has been added. revision: yes
-
Referee: [§2 and §3.3] §2 (Townsend hypothesis discussion) and §3.3 (variance collapse analysis): the attribution of the local-stress scaling to 'active' motions relies on Townsend's attached-eddy hypothesis, which is derived under the idealization of constant turbulent shear stress. The DNS data span regions where total stress varies with y (linearly in channels, decreasing in boundary layers), and the paper does not quantify how much of the reported collapse survives when this y-dependence is removed from the normalization or how the residual inactive-motion discrepancies affect the claimed universality.
Authors: We acknowledge that Townsend's hypothesis is derived under constant-stress idealization, which is already noted in the original §2. In the revision we have added a quantitative check in §3.3: the wall-normal variance normalized by local shear stress is replotted versus y/δ, demonstrating that the collapse across flows remains within 8 % in the lower half of the layer. Residual discrepancies at low wavenumbers are explicitly linked to weak inactive motions. A complete active/inactive decomposition lies outside the present DNS analysis and would require additional techniques; we have flagged this as a limitation and direction for future work while maintaining that the local-stress scaling accounts for the dominant observed differences. revision: partial
Circularity Check
Semi-empirical fit tuned to DNS variance data then extrapolated to high-Re asymptotic value
specific steps
-
fitted input called prediction
[Abstract]
"A semi-empirical fit applied to the Reynolds number dependence of the variance matches the simulations across the lower half of the boundary layer and aligns with observed values in the literature. The fit extrapolates to a value between 1.45 and 1.65 times the local shear stress in the high-Reynolds-number limit, consistent with previous predictions relative to $U_τ$ including for the vertical velocity in the near-neutral atmospheric boundary layer."
The semi-empirical fit is adjusted to reproduce the Re-dependence of wall-normal velocity variance seen in the DNS; the extrapolated high-Re constant (1.45–1.65) is therefore fixed by the same finite-Re simulation data it is offered as predicting or aligning with in the asymptotic limit.
full rationale
The paper's key result—that local stress deviations explain nearly all cross-flow differences in wall-normal velocity variance—rests on a semi-empirical fit to the observed Reynolds-number dependence of the variance in the DNS. This fit is constructed to reproduce the simulation data across the lower half of the layer and is then used to extrapolate the high-Re limit (1.45–1.65 times local stress). Because the asymptotic constant is determined by the same finite-Re data it is presented as predicting, the extrapolation step reduces to the input by construction. The attribution to Townsend's attached-eddy hypothesis supplies the physical motivation for local-stress scaling, but the hypothesis is invoked in regions where total stress is not constant, which is an assumption rather than a circular derivation. The observed collapse with local stress provides independent empirical content, so the circularity is partial and localized to the fit-extrapolation procedure.
Axiom & Free-Parameter Ledger
free parameters (1)
- asymptotic proportionality constant
axioms (1)
- domain assumption Townsend's attached eddy hypothesis assumes simplified ideal conditions with constant turbulent shear stress
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.