AERODYNAMICS
Shape the section.
Discover the flow.
Start here to compare the upper and lower flow at zero incidence.
The Flow view shows parallel incoming guides and computed streamlines bending around the section. Choose Speed to find acceleration, Pressure to inspect pressure differences, or Vorticity to see local rotation.
Keep an outline, change the camber or angle, and compare the geometry. The saved outline is a visual reference; it does not retain a second flow solution.
At the same reference Reynolds number, the flow can change substantially with shape.
This is a two-dimensional, low-Mach numerical experiment with a finite domain. It does not predict full-scale aircraft performance, three-dimensional wingtip effects or a certified stall angle.
Explore the assumptions ↓A D2Q9 lattice Boltzmann solver approximates isothermal, weakly compressible viscous flow at low lattice Mach number. The obstacle is stationary, with a no-slip bounce-back boundary. The domain is 8c × 4c, with a prescribed inlet velocity, a pressure-relaxing characteristic outlet and impermeable free-slip top and bottom boundaries.
Re = U∞c / ν. Here c is the original reference chord, or original cylinder diameter. Editing does not change this reference scale. Increasing Re changes ν at fixed U∞ and c. No real-time seconds or airspeed in m/s are implied.
Dragging is a geometry design edit, not physical motion of an elastic body. After an edit, a numerical preparation accelerates the fluid and inlet together over two reference convective-time units. An equilibrium-difference momentum increment avoids suddenly imposing full-speed flow on the body. This is initialisation, not a model of a real fan starting. The increment is exactly zero afterwards. U∞ and Re denote the target conditions; tU∞/c includes preparation. Read developed fields after preparation; elapsed time does not certify convergence. Unsteady flow need not become steady.
Continuous lines follow the current computed velocity directions from evenly spaced inlet positions. The labelled uniform-inlet guide lies outside the calculated domain. Moving tracers are passive markers advected through the computed velocity field. Their trails represent sampled particle paths, not instantaneous streamlines or smoke streaklines. They carry no feedback into the solver. Speed is normalised by U∞. Pressure uses Cp = 2(ρ − 1)/(3U∞²), relative to target density 1 and the final target inflow speed. The characteristic outlet approaches that pressure reference; its instantaneous density need not equal 1. Vorticity is ωc/U∞, positive anticlockwise in the displayed physical plane.
Pressure and vorticity use linear colour scales, shared by the map and colour bar, with no spatial tint or image smoothing. Colours clip at their labelled limits; the point probe reports the unclipped value.
Curved surfaces and thin trailing edges are represented on a Cartesian grid. Small details may be unresolved. Lateral confinement affects the flow, especially for the cylinder. Use the finer grid to explore sensitivity; neither option is certified mesh independent. The model stops on non-finite states or excessive density/Mach excursions. Open boundaries use non-equilibrium extrapolation and a local one-dimensional characteristic outlet. These reduce the tested pressure ringing; they are not perfectly transparent to every wave or verified for arbitrary outlet backflow.
Lift and drag coefficients are deliberately withheld pending dedicated force verification and benchmark validation. Trailing-edge deflection is a smooth sealed morph, not a hinged, slotted or multi-element flap.
Methods: Guo et al.: non-equilibrium boundaries · Wissocq et al.: characteristic boundaries · NASA: boundary layers
Inlet–outlet flux difference alone is not a mass-conservation residual: density storage changes during start-up. The interior divergence diagnostic excludes wall-adjacent cells and is not a convergence certificate.
Waiting for the first flow update.