How this instrument works
y⁺ counts how many viscous length-scales the centre of the first mesh cell sits from a solid wall. That viscous length, ν ⁄ u_τ, is set by the local wall shear stress τ_w — the drag the fluid exerts on the surface — combined with the fluid's kinematic viscosity ν. Divide the physical wall distance y by that length and the result carries no units at all: a cell 0.1 mm off the wall can read y⁺ ≈ 1 in a slow, viscous flow and y⁺ ≈ 50 in fast-moving air shearing hard against the same surface.
The friction velocity u_τ = √(τ_w ⁄ ρ) is what makes the formula shaped this way. Near a wall, a turbulent boundary layer organizes itself around one velocity scale set by the momentum flux into the surface, not the free-stream speed far above it — this is Prandtl's law-of-the-wall scaling, and it is why velocity profiles from wildly different flows collapse onto the same curve when plotted against y⁺. A turbulence model's near-wall behavior is keyed to this number precisely because it is universal in a way that raw mesh spacing in millimetres is not.
The catch is that τ_w is almost never known before a mesh exists. It is usually estimated from a flat-plate or pipe skin-friction correlation, then corrected once a solver has actually run and can report the true wall shear stress cell by cell. The y⁺ this calculator returns during mesh planning is therefore a target, not a certainty — a surface near separation, a shock, or a sharp geometry change can carry a wall shear stress far outside the flat-plate guess, so the same fixed cell height lands at a different y⁺ there than the correlation predicted.
- Enter the Distance from wall, first cell centre — the wall-normal gap between the wall and the centroid of the first computational cell.
- Enter the Wall shear stress, taken from a skin-friction correlation, a prior run, or the solver output at this location.
- Enter the Fluid density and the Kinematic viscosity, m²/s for the working fluid at the flow's temperature and pressure.
- Read y⁺ (dimensionless wall distance) and compare it against the range your turbulence model's near-wall treatment expects.
- Adjust the first-cell height and recompute until y⁺ lands in the band your solver's documentation calls for, rather than accepting whatever the initial mesh happened to produce.
Worked example — a first cell landing at y⁺ ≈ 7.4
A mesh engineer is checking the first cell of a RANS simulation in air: density 1.225 kg/m³, kinematic viscosity 1.5×10⁻⁵ m²/s, and a wall shear stress of 1.5 Pa pulled from a flat-plate skin-friction estimate at that station. The chosen cell centre sits 0.0001 m — 0.1 mm — off the wall. Friction velocity comes first: u_τ = √(1.5 ⁄ 1.225) = 1.1066 m/s.
Then y⁺ = 0.0001 × 1.1066 ⁄ 0.000015 = 7.377. That places the first cell inside the buffer layer, roughly y⁺ 5 to 30 — the zone every standard near-wall treatment is built to avoid, caught between the viscous sublayer a low-Reynolds-number model wants resolved directly (y⁺ ≈ 1) and the log-law region a wall function assumes (y⁺ 30 to 300). The fix is to halve the cell height toward y⁺ ≈ 1 or grow it toward y⁺ 30 or beyond, not leave it stranded in between.
Questions
What y⁺ value should I actually be aiming for?
It depends on the near-wall treatment your turbulence model uses. Low-Reynolds-number models such as SST k-ω resolving the viscous sublayer directly want y⁺ ≈ 1, with several cells below y⁺ = 5. Wall-function approaches assume the log-law and want y⁺ roughly 30 to 300. The buffer layer in between, y⁺ 5 to 30, is a poor target for either.
Why does the calculator need wall shear stress, which I don't know yet?
Because y⁺ estimation happens during mesh design, before a simulation has run. Pull an estimate from a flat-plate or pipe skin-friction correlation, or from a similar prior case, then recompute y⁺ using the actual wall shear stress the solver reports once the simulation converges, and refine the mesh if the two disagree.
How is friction velocity different from the free-stream velocity?
Friction velocity u_τ is derived from wall shear stress, not flow speed, and characterizes the intensity of turbulence right at the surface rather than the bulk velocity far from it. It is usually much smaller than the free-stream speed, but it — not the free-stream value — sets the near-wall length scale that y⁺ is measured against.
Does y⁺ depend on the overall boundary layer thickness?
No. y⁺ is purely local: it uses only the wall-normal distance of the first cell, the local wall shear stress, and the fluid's density and viscosity at that point. Boundary layer thickness and body-scale Reynolds number matter only indirectly, through the wall shear stress correlation used to estimate τ_w in the first place.
Can y⁺ differ across one continuous surface?
Yes, because wall shear stress varies along a surface — higher near a leading edge, lower where the flow is decelerating or near separation. A mesh with one fixed cell height therefore produces a range of y⁺ values across the same body, which is why meshing tools typically report a minimum and maximum y⁺ rather than a single number.
What happens if the first cell ends up in the buffer layer anyway?
Near-wall treatments built for either the viscous sublayer or the log-law region both lose accuracy there, since the buffer layer follows neither the linear viscous profile nor the logarithmic one cleanly. Most guidance is to adjust the first-cell height rather than accept a y⁺ of 5 to 30 and hope the solver's blending function compensates.