boundary layer

A boundary layer is a thin region where a solution changes extraordinarily fast to satisfy a condition the bulk approximation cannot meet. The picture comes from fluid flow: air streams smoothly past a wing almost as if frictionless, yet right at the surface the air must be at rest (it sticks to the wing). All of that rapid change from 'at rest at the wall' to 'free-stream speed' is squeezed into a microscopically thin sheet against the surface — the boundary layer.

Mathematically it is the fingerprint of a singular perturbation. When a small parameter multiplies the highest derivative, the reduced (outer) problem cannot satisfy all the boundary conditions. The remedy is a layer, of small width set by balancing the neglected highest-derivative term against the dominant remaining term, in which the previously discarded derivative becomes important again. Inside the layer one rescales the coordinate (stretching it by the appropriate power of the small parameter) so the rapid variation looks order-one; the result is an inner equation whose solution typically rises or falls exponentially across the layer, like 1 - e^(-x/epsilon). The layer's thickness, location, and shape are dictated by the equation, not chosen freely.

Boundary layers are central to aerodynamics and drag (Prandtl's 1904 insight that viscosity, however small, matters in a thin layer and governs skin friction and flow separation), to heat and mass transfer, to chemical reactions with fast and slow steps, and to control theory. They can also appear in the interior of a domain (interior or transition layers, shock-like structures) rather than at a boundary. The honest caveat: not every singular problem has a boundary layer at the obvious place; whether a layer forms, and at which end, depends on the signs of the coefficients, and diagnosing this correctly is the crux of the analysis.