Shark-Skin Riblets: Protrusion Height and Turbulent Drag Reduction (2D)

Solve the viscous flow along and across a row of riblets, read off the two protrusion heights and see how their difference becomes a drag reduction that peaks and then breaks down.

Switch to 3D
Controls

Presets

Parameters

Model checks

The science

Why ribs can reduce drag. Near a wall in a turbulent boundary layer, streamwise vortices drag fast fluid towards the wall and slow fluid away. Tiny ridges aligned with the flow (shark denticles, riblets) let the streamwise flow sink between them but obstruct the spanwise flow of the vortices, so the turbulent stress origin sits higher than the origin for the mean flow. In the viscous regime (spacing s+ = s uτ/ν below about 10 to 15) the flow around the ribs is a linear Stokes flow and everything is captured by two lengths, the protrusion heights of the virtual origins measured from the rib tips: h∥ for flow along the ribs and h⊥ for flow across them (Bechert and Bartenwerfer 1989, Luchini et al. 1991).

What is solved here. Along the ribs the velocity obeys ∇²u = 0 with u = 0 on every wall and a unit shear far above; the far field u = y − y0 gives h∥ = h − y0. Across the ribs the Stokes equations are solved in streamfunction-vorticity form (u = ∂ψ/∂y, ω = −∇²ψ, ∇²ω = 0) with no slip (Thom wall vorticity), unit shear above and zero mean pressure gradient, which gives h⊥. Both are finite-volume calculations on the periodic cell of one rib (thickness t, height h, spacing s) and are re-solved when h/s or t/s change. The tracers in the pictures are advected by these solved fields.

From protrusion heights to drag. The mean velocity profile is shifted by ΔU+ = (h∥ − h⊥)+ in the viscous regime. With the smooth-wall friction from the Karman-Schoenherr law (0.242/√Cf = log10(Re Cf)), the drag reduction at the same outer flow is DR = 1 − (U+/(U+ + ΔU+))² with U+ = √(2/Cf). Doubling the spacing in wall units doubles the benefit, until it breaks down.

Honest simplifications. The breakdown is not computed. Beyond a groove size ℓg+ = √Ag+ of about 10.7 spanwise Kelvin-Helmholtz rollers appear (Garcia-Mayoral and Jimenez 2011) and the benefit falls again; here that is an empirical quadratic anchored at the optimum ℓg+ = 10.7 (onset 6), followed by logarithmic roughness growth (1/κ) ln(ℓ/ℓ0) once the drag exceeds the smooth wall. The rib shapes are rectangular blades (real riblets and shark denticles are curved and three-dimensional; real shark skin also has compliance and mucus); the peak drag reduction of this model for blades is a few percent to about 8 percent while measured maxima for the best shapes are around 10 percent. The Thom wall treatment has a first-order error: a flat wall returns a spurious h⊥ of 0.25 cell, which is subtracted (check 4), and the grid-convergence check bounds the rest. The shift is applied at fixed outer conditions, changes of uτ by the riblets are ignored, and the boundary layer is assumed fully turbulent. The shark in the 3D view is only a context model and is not to scale: ribs of 50 to 100 micrometres on a metre-long animal.

-
-