Donut lip vs none · 70 mm · 6S · hover — lip adds 620 g (46.2%).
Attached inflow
Layer A thrust · Layer B = picture
Honesty note. The flow picture is a planar 2D panel solve of the duct section, not an axisymmetric one. Planar flow over-predicts how far the duct's influence reaches and under-predicts the lip suction peak of the real ring. It drives the picture and relative trends only (which lip separates first, where the stagnation point sits, crossflow asymmetry). Thrust and power come from Layer A. The Cp map is scaled so its lip force matches Layer A's duct thrust at axial static. The 3D field is the 2D solve swept round the axis. Swirl is kinematic: the blade torque sets Vθ behind the rotor and rVθ is conserved downstream. Stator vanes turn it back toward axial (Carter's-rule deviation); their pressure recovery enters Layer A only as a difference from an 8-vane fan, which the calibration already describes. Air outside the jet only picks up rotation through a stylised mixing layer that spreads and dilutes the swirl downstream; there is no induced far-field rotation. The twin superposes the two EDFs' planar fields, each tapered as R/r beyond its radius so the neighbour's pull falls off like a real inlet's (1/r², not the planar 1/r); jet attraction and merging come from a stylised entrainment model. Duct length and fan station move the picture and the lip criterion (calibrated at 1.1 D / 0.44 D); in Layer A they only add [S] bore friction and a short-inlet distortion loss against that reference.
P/D 1.71 · read-only · 0.75·π·tan β75
—
σ = Aexit/AFSA
An inlet lip is the rounded edge at the front of an electric ducted fan's shroud. At static thrust the fan draws air from the sides as well as from ahead, and that air has to turn round the lip. A generous radius lets it turn without separating, and the suction it carries on the lip face becomes thrust on the duct itself. A sharp edge separates, shrinks the effective inlet area and throws that share away. This simulator compares no lip, a simple lip and a donut lip on presets from 50 to 120 mm and 4S to 12S. Thrust, current and power come from a 1D performance model. The flow picture is a planar 2D panel solve of the duct section swept round the axis, not an axisymmetric solution. Every number is model output, not a thrust-stand measurement.
Yes, in this model. The lip radius sets the inlet-loss coefficient K that the performance model reads from a table: 0.50 with no lip, 0.12 at r/D 0.08, 0.03 from r/D 0.20 up. The lip keeps √(1 − K/0.5) of its suction, and K also shrinks the effective inlet area as 1/√(1 + K). For the 70 mm 6S·12 preset at static, the solver gives 1963 g with a donut lip (tube r/D 0.15) and 1343 g with no lip, and a sharp inlet makes about 73 % of the thrust of a simple lip at r/D 0.10. These are model outputs, not thrust-stand measurements.
A donut lip is a full toroidal lip: a ring of round cross-section wrapped round the inlet and tangent to the bore. Here its tube radius r_t/D runs from 0.05 to 0.30 and counts as an equivalent lip radius of 1.4 × r_t/D in the inlet-loss table; that 1.4 wrap-around bonus is a stylised assumption, not a law. The ring also adds frontal drag, C_D = 0.10 + 1.5 (r_t/D)², so at 50 m/s the fattest donut (0.30) gives the lowest net thrust of the lips the self-test compares.
The model uses two criteria. At static the inlet-loss table decides: the lip counts as separated when K exceeds 0.12, which means no lip at all or a simple lip under r/D 0.08; in forward flight that K fades as (1 − V∞,ax/V_d)². In crossflow (lateral speed in hover, or incidence in forward flight) the planar panel solve decides: the pressure recovery ΔCp from the lip's suction peak to the fan face, scaled by κ = 0.16, is compared with ΔCp_crit = min(2.6, 1.2 + 6 r/D). The lip trips when that ratio passes 1 and is worse than without crossflow, and its loss is fully in by 2. The margin to separation readout is 1 minus that ratio.