Dataset: Dataset: Developing Turbulent Secondary Flow over Streamwise-Aligned Roughness Strips

This dataset contains mean-flow statistics of incompressible turbulent channel flow over streamwise aligned ridges. The Reynolds number Re_b = 2 h u_b / nu, where h is the channel half height, u_b the bulk velocity and nu the kinematic viscosity, is constant at a value of Re_b = 4240, corresponding to a friction Reynolds number Re_tau ≅ 180 for smooth-wall channel flow.

The channel features streamwise-aligned roughness strips of various spanwise sizes: s/h = {1, 1.5, 2, 4}. Two configurations are considered: fully developed flow (streamwise-periodic domain) and developing flow (onset of the boundary condition at z=0, with a velocity recyling region at z=[-40h, -32h]), followed by a recovery region and a z=88h long region with roughness strips, followed by Neumann outlet boundary condition.

The data was generated using XCompact3D with the scheme by Lamballais et al. (2011), modified to support velocity recycling.

The rougness strip is modeled using the slip-length boundary condition from [2]; see that publication for validation.

The simulations were initialised by tiling a snapshot from a periodic, smooth-wall case at $\Re_\tau = 180$. After an initial transient of $T_{trans.} u_{\tau0} / \delta = 60$, statistics were collected for at least $T u_{\tau0} / \delta = 380$ large-eddy turnover times. The lack of a statistically homogeneous streamwise direction dictates a high sampling frequency: the decorrelation time of pointwise velocity statistics is $\mathcal{O}(t_v)$ where $t_v = \delta_v ^2 / \nu$, and run-time statistics are sampled every $0.38 t_{v0}$. To further improve collected statistics, they were phase-averaged in $z$ and $y$; however, the $z$ symmetry was not enforced. The deviation from this symmetry therefore indicates the statistical uncertainty of the obtained mean values.

Two validation steps were performed. First, the implementation of the velocity recycling was validated by asserting that the terms of the Reynolds stress transport equation within the recovery region coincide with those of a fully developed (no-slip) reference case. Second, the statistics at the end of the inhomogeneous section were compared against a periodic reference case with the same boundary condition and run at the same bulk flow rate. The grid resolution is identical between fully developed and developing cases.

[1] Lamballais, Eric, Véronique Fortuné, and Sylvain Laizet. "Straightforward high-order numerical dissipation via the viscous term for direct and large eddy simulation." Journal of Computational Physics 230.9 (2011): 3270-3275.
[2] Neuhauser, Jonathan, et al. "Simulation of turbulent flow over roughness strips." Journal of Fluid Mechanics 945 (2022): A14.

(Technical Remarks) The data are stored using Zarr [1] as Xarray [2] backend; the datatree layout is

• inlet
  • 1
  • 1.5
  • 2
  • 4
• periodic
  • 1
  • 1.5
  • 2
  • 4

The following fields are contained for each field in the dataset:

• 'lsw': Boundary condition

• Runtime-averaged:
  ['', '', '', '', '', '', '', '', '', '', '', '', '', '', '', '', "<p'dudx'>", "<p'dudy'>", "<p'dudz'>", "<p'dvdx'>", "<p'dvdy'>", "<p'dvdz'>", "<p'dwdx'>", "<p'dwdy'>", "<p'dwdz'>", '<pdudx>', '<pdudy>', '<pdudz>', '<pdvdx>', '<pdvdy>', '<pdvdz>', '<pdwdx>', '<pdwdy>', '<pdwdz>', '

  ', "<u'p'>", "<u'u'u'>", "<u'u'v'>", "<u'u'w'>", "<u'u'>", "<u'v'v'>", "<u'v'w'>", "<u'v'>", "<u'w'w'>", "<u'w'>", '<up>', '<uuu>', '<uuv>', '<uuw>', '<uu>', '<uvv>', '<uvw>', '<uv>', '<uww>', '<uw>', '', "<v'*p'>", "<v'u'>", "<v'v'v'>", "<v'v'w'>", "<v'v'>", "<v'w'w'>", "<v'w'>", '<vp>', '<vvv>', '<vvw>', '<vv>', '<vww>', '<vw>', '', "<w'*p'>", "<w'*u'>", "<w'v'>", "<w'w'w'>", "<w'w'>", '<wp>', '<www>', '<ww>', '']

• Post-computed fields (note that compact finite differences are used for the derivatives)
  ['Pressure_Strain_k', 'Pressure_Strain_uu', 'Pressure_Strain_uv', 'Pressure_Strain_uw', 'Pressure_Strain_vv', 'Pressure_Strain_vw', 'Pressure_Strain_ww', 'Pressure_Transport_k', 'Pressure_Transport_uu', 'Pressure_Transport_uv', 'Pressure_Transport_uw', 'Pressure_Transport_vv', 'Pressure_Transport_vw', 'Pressure_Transport_ww', 'Production_k', 'Production_uu', 'Production_uv', 'Production_uw', 'Production_vv', 'Production_vw', 'Production_ww', 'Turbulent_Transport_k', 'Turbulent_Transport_uu', 'Turbulent_Transport_uv', 'Turbulent_Transport_uw', 'Turbulent_Transport_vv', 'Turbulent_Transport_vw', 'Turbulent_Transport_ww', 'Viscous_Dissipation_k', 'Viscous_Dissipation_uu', 'Viscous_Dissipation_uv', 'Viscous_Dissipation_uw', 'Viscous_Dissipation_vv', 'Viscous_Dissipation_vw', 'Viscous_Dissipation_ww', 'Viscous_Transport_k', 'Viscous_Transport_uu', 'Viscous_Transport_uv', 'Viscous_Transport_uw', 'Viscous_Transport_vv', 'Viscous_Transport_vw', 'Viscous_Transport_ww']