WingHopper · Designer · Guide · Sections · Reef · Methods · Pricing · FAQ

How the numbers are computed

How the numbers are computed

Use the models to compare designs. Geometry is calculated from your wing; performance and strength depend on assumptions.

Speeds and fluid

Lift must balance weightLift ↑Weight ↓FlowArea S

Schematic, not a computed result. Lift must balance weight.

V = √(2W / (ρSCL)). W is rider + gear mass × gravity in newtons; S is planform area in m²; ρ is fluid density in kg/m³. The Fluid tab assumes CL 0.8 / 0.5 / 0.3 for takeoff / cruise / fast. These are app presets, not measured conditions. NASA lift equation.

Reynolds number is V × MAC / ν, with chord in metres and kinematic viscosity ν in m²/s. Seawater preset density is 1,025 kg/m³. The wing-class and section-advisor bands are app starting ranges.

Geometry

  • Area integrates the leading/trailing edges and rounded tip.
  • Aspect ratio = span² / area. Use the same length units.
  • Mean aerodynamic chord (MAC) = ∫c² dy / ∫c dy; area / span is the geometric mean chord.
  • Wetted area integrates section perimeters along the span.
  • The pitching-moment reference is the chord-weighted mean quarter-chord. It is a reference location, not a universal measured aerodynamic centre.

Two wing-flow models

ModelUsed forAssumes
Lifting-line (LLT)Fast Polar and shape scoresPlanar attached flow; section camber/twist included; no swept/nonplanar influence matrix
Vortex lattice (VLM)Performance, span loads and coupled twistIncompressible potential flow on the panelled camber surface; straight wake
Catalog 2D polarsSection comparisonfoil.tools data at stated Reynolds/transition settings; separate from the wing solvers

Both wing models omit viscous separation, free-surface effects, ventilation and cavitation. Displayed L/D adds a fixed CD 0.009 profile-drag allowance to induced drag. It does not include mast/fuselage drag or use each catalog section's viscous polar.

Solver details and stall screening

LLT uses Fourier circulation and thin-airfoil zero-lift angles from the sampled camber. Its polar sweeps −2° to 16° in 0.5° steps. Best L/D and CL max exclude points exceeding the assumed local stall angle. This cutoff is not a separation calculation.

VLM uses quarter-chord bound vortices and three-quarter-chord control points. Twist and dihedral enter the normals. Near-field Kutta–Joukowski integration supplies forces; moments are nose-up positive. The Performance sweep is −2° to 14° in 1° steps, with a sectional CL screening ceiling of 1.45.

Classical basis: NASA induced drag and MIT AVL primer. WingHopper runs its own solvers. XFOIL couples a panel method to a boundary-layer model; it can model transition and limited separation, unlike the wing solvers. XFOIL primer.

Supercavitating and blunt-base sections

A section with a trailing edge at least 2% of chord thick, or its maximum thickness aft of 75% chord, is outside the rounded-nose assumptions above. For those wings the CL 1.45 ceiling and the CD 0.009 allowance are replaced, and the numbers are labelled SC section estimate. Everything is wetted flow, before the section cavitates.

  • Lift ceiling: the highest strip may reach a sectional cl of 0.77 to 0.81. Those are two measured maxima for sharp-nosed sections: the NACA 0012 run sharp edge first (0.77, Re 1.8 million; Critzos, Heyson and Boswinkle, NACA TN 3361, 1955) and a 4.23% faired double wedge (0.814, Re 5.8 million; Rose and Altman, NACA TN 1934, 1949). Lift-off is shown as the speed range those two ceilings give. Both sections are symmetric, and TN 3361 found that leading-edge roughness lowers the peak, so a rough or chipped nose can stall below this range.
  • Why the wetted lift slope still applies up to that point: Wick (NACA TN 3221, 1954) found that the flat plate and the Rose and Altman double wedge (NACA TN 2172) follow thin-airfoil lift up to about 7.5°, despite the leading-edge separation bubble.
  • Drag: each strip carries turbulent friction with Hoerner's thickness form factor plus Hoerner's base drag for its own trailing-edge height (Fluid-Dynamic Drag, 1965, ch. 3 and 6), at the Reynolds number of the speed that carries you at that CL. Peak L/D is searched below the lower lift ceiling.
  • Regime speed sweep (Cavitation panel, SC sections only): at each speed every strip is trimmed with the lift law of its own regime, the wetted slope below inception, Acosta's partial cavity, or Tulin's supercavity with the Tulin and Burkart pressure-face camber term (DTMB C-638, 1955, as restated in Johnson, NACA RM L57I16 / NASA TR R-93). The chart shows lift-off, inception, the unstable 3/4 to 4/3 chord band (no number is reported there) and L/D through the regimes. In this linear theory the branch a level-flight wing lands on is fixed by its loading ratio W / (S (p∞ − pv)), not by speed; a rider-loaded foil sits far below the supercavity threshold, and the panel says what wing area would reach it. The supercavity law was checked against the Caltech flat-plate data of Parkin (1956): lift within 6% and drag within 8% for cavitation numbers up to 0.21 and incidence 8 to 13°, growing to about 10% by 0.3. Ventilation from the surface is not modelled.

Pump, carve and glide

  • Pump: minimum V / (L/D) below the assumed stall cutoff, a steady-power proxy. More area affects it through loading.
  • Carve: an aspect-ratio proxy. It omits actual roll dynamics and rider technique.
  • Glide: peak computed L/D.

The 0–10 bands rank shapes within this model, without ride-data calibration. They are not manufacturer ratings. The pump guide separates these scores from the uncalibrated heave-stroke power model.

Stabilizer trim

The wing studio fits a stabilizer lift line from two VLM solves, samples main-wing downwash at the tail and closes pitch moment about your entered CG. Tail induced drag uses a fitted parabolic polar. The Fluid tab's entered tail-load pairing is a separate fixed-lift estimate.

Assembly uses one influence matrix for both wings, balances total lift at fixed shim angle and reports the remaining pitch moment. Assembly assumptions.

Structure and load cases

Skins and core carry different loadsCompression skinCore shearTension skinUpward bending shown. Reverse load swaps tension and compression.

Schematic, not a computed result. Skins and core carry different loads.

The wing-studio design factor is max(load floor, min(wave gust, lift ceiling)), with a minimum floor of 1. The default user floor is 3.5. This is an app policy, not a universal marine safety factor.

The gust model uses a regular wave at 1.86 × significant height, an approximate maximum for about 1,000 Rayleigh-distributed waves. Duration and wave statistics matter. DNV-RP-C205.

Strength, flex and material assumptions

All design load is assigned to the front wing; tail download can increase it. VLM distributes spanwise loads. Sandwich bending assumes skin separation 0.85 × section thickness and effective flange width half the chord. Solid-section properties use integrated geometry. Deflection uses cantilever M/EI integration.

Material allowables and fatigue knockdowns are app design estimates, not coupon data for your laminate. Example values: woven carbon 150 MPa, 6061-T6 aluminium 60 MPa. Divinycell H80 core shear uses 0.95 MPa minimum, reduced separately to 0.475 MPa peak and 0.314 MPa high-cycle allowables. Corecell and ROHACELL entries lacking minimum data use an estimated 0.8 × typical.

Diab datasheets; Gurit Corecell M data. Supplier data supports the base properties; the app's knockdowns are additional assumptions.

The root bolt line subtracts hole strips from the integrated section and reports net bending stress plus idealised open-hole concentration. It uses an elliptical root-moment approximation. It does not solve clamp load, bearing, inserts or the complete joint.

Plies checks skin wrinkling and coupled load/twist at 24 spanwise stations. Divergence is the linear equilibrium limit. The displayed ratio compares the higher of 10 m/s and the operating-band top with predicted divergence speed. The model omits root flex, laminate bend/twist coupling and large rotations. Georgia Tech: divergence.

Not covered: water-entry slam, defects, skin fatigue counting, most joints and bondlines. More skin plies do not automatically fix core shear or a joint.

Auto laminate

Sizes a symmetric two-skin schedule for bending and wrinkling at 90% utilisation. One schedule describes one side. Core shear, bolt-line and divergence checks still need review.

ProcessAssumed fibre volume fraction Vf
Wet hand layup0.35
Vacuum bag0.45
Infusion0.50
Prepreg0.55

These fractions are process estimates. Measure your cured coupon. Cured ply thickness (mm) = cloth g/m² / (fibre density kg/m³ × Vf). Carbon uses 1,800 kg/m³; E-glass 2,550 kg/m³. Example: 200 g/m² carbon at Vf 0.35 gives 0.32 mm. Gurit: laminate equations and process effects.

How the schedule is sized

Strength and modulus scale with Vf / 0.40; thickness varies inversely with Vf. The skin-beam stress is M / (t × c/2 × h). A full-span ±45° torsion ply starts the schedule; laminate theory determines its axial contribution with the 0/90° stack.

The algorithm adds and trims full-span 0/90° plies against the worst bending and wrinkling station. At a solid tip, inner patterns stop where the section cannot hold the stack. The resulting schedule is a model-sized proposal, not a qualified layup. Schedule, seams and cut files.

Ply patterns and drape

Triangulated surface unfolding gives flat outlines and an in-plane shear estimate. The 15° warning and roughly 30–40° locking bands are app heuristics, not measured forming limits for your fabric. Fit patterns on the tool; review darts and splits without silently changing structural fibre direction.

Verification and references

Solver gates compare selected wings with classical lifting-line cases and symmetry/moment invariants. Geometry gates check saved-shape stability and export consistency. Numerical agreement in those cases does not establish accuracy for every wing or a physical build.

For a suspect result, send the design link and inputs through Feedback so the calculation can be reproduced.