Wingtip Device Study for the Helios M2 Wing
Why do airplanes have wingtips?
During flight, high pressure air from the bottom of the wing leaks to the top, which creates two undesirable effects:
Reduces pressure difference between the lower and upper wing surface (↓ Lift)
Leads to wing tip vortices (↑ induced drag).
Wingtips provide many benefits:
Reduces induced drag by weakening the wingtip vortex and moving it away from the wing surface.
Recovers some of the swirl energy as a forward force.
The greater the CL, the greater the benefit grows.
What it costs:
Extra wetted area (profile drag), higher root bending moment, and greater tip mass.
Wingtip Types:
Baseline:
Item | Value | Source |
|---|---|---|
Wing Span | 2.42m | Rohaan Vasa |
Aspect Ratio | 8.5 | Rohaan Vasa |
Planform Area | 0.691m^2 | Rohaan Vasa |
Airfoil | NACA 4412 | Jakob Boelling |
CL_max | 1.12 at 7 kg, 1.60 at 10 kg | Jakob Boelling |
CL_cruise | 0.28 at 7 kg, 0.40 at 10 kg | Jakob Boelling |
Re (@ cruise) | 300k | Jakob Boelling |
TL;DR - How much could a wingtip save us?
At our cruise lift coefficient (CL 0.28 at 7 kg, 0.40 at 10 kg), no wingtip device changes wing drag by more than about 0.17 N, roughly 4% of wing drag. The typical case is a few hundredths of a newton. That is inside the error of rule-of-thumb sizing, so the choice rests on structure, printability and risk, not on drag.
Recommendation: design the raked tip first and confirm the drag estimates in an XFLR5 vortex-lattice (VLM) comparison.
Wingtip Type | 7 kg net | 10 kg net |
|---|---|---|
Winglet | +0.7% | −1.5% |
Raked tip | −0.2% to +1.7% | −3.6% to −0.5% |
Hoerner | −0.5% to +1.4% | −4.2% to −1.1% |
Rounded | 0.0% to +1.9% | −3.5% to −0.3% |
Endplate | +3.7% to +4.8% | +1.2% to +3.0% |
Method:
Induced drag: CDi = CL^2 / (π · e · AR), with Oswald e = 0.9 (assumed) and π · e · AR = 24.03.
Case | Cruise CL | CDi | Induced share of wing drag |
|---|---|---|---|
7 kg | 0.28 | 0.00326 | 24% |
10 kg | 0.40 | 0.00666 | 39.5% |
Share = CDi / (CD0 + CDi), with CD0 = 0.01021 (XFLR5, NACA 4412).
Induced drag scales as 1 / span^2, so a device that adds effective span b' gives an induced-drag reduction of 1 − (b / b')^2.
Net change = (cost − saving) / (CD0 + CDi). A negative value means less drag.
Evaluating Various Wingtip Types:
Winglets:
As a winglet acts as a longer wing, its extra span is about 45% of its height (known as Kroo’s result [1]). Since induced drag also scales with 1/span^2, adding a winglet of height h on a wing of span b cuts induced drag by:
The wingtips will provide a new effective span of 2.42 + 0.135 = 2.555 m (considering the height of the winglet is 0.15m). Plugging into the equation above, this will result in a 10.3% reduction in induced drag.
Cost:
Extra profile drag caused by the winglet:
Area per winglet = height × average chord = 0.15 × (0.12 + 0.072) / 2 = 0.0144 m²
Both winglets = 0.0288 m²
As a fraction of wing area = 0.0288 / 0.691 ≈ 4.2%
Cost = 4.2% × CD0:
With CD0 = 0.01021 (Determined from XFLR5 NACA 4412 analysis): 0.0417 × 0.01021 ≈ 0.00043 (about 0.0004)
Case | Cruise CL | Induced share of wing drag | Best-case saving | Cost | Net change* |
|---|---|---|---|---|---|
7 kg | 0.28 | 24% | 0.00034 | 0.0004 | +0.7% |
10 kg | 0.40 | 39.5% | 0.00069 | 0.0004 | -1.5% |
*A minus sign means less drag, which is good
Raked Wingtips:
A raked tip is a swept, tapered extension beyond the current tip (1.210 m from the root). It adds span, so it follows the same 1 / span^2 rule.
Geometry (per side):
Length 0.12 m, root chord 0.200 m, tip chord 0.100 m
40 deg. leading-edge sweep (offset = 0.12 × tan 40 = 0.10 m), unswept trailing edge
New span = 2.42 + 2 × 0.12 = 2.66 m (+10%)
Added area = 0.036 m^2, so the area becomes 0.727 m^2 and AR ≈ 9.7
Saving. A swept, tapered tip carries less lift than a full-chord extension, so let k be the fraction of the added span that works.
k = 1 (best case): (2.42 / 2.66)^2 = 0.828, a 17.2% cut in induced drag
k = 0.5 (conservative assumption, not from a source): effective span 2.54 m, a 9.2% cut
Cost. 0.036 m^2 is 5.2% of the wing area, so cost = 0.052 × 0.01021 = 0.00053. The lower Re on the tip chord (0.1–0.2 m) makes this somewhat optimistic.
Case | Cruise CL | Induced share | Saving (k = 1) | Saving (k = 0.5) | Cost | Net (k = 1) | Net (k = 0.5) |
|---|---|---|---|---|---|---|---|
7 kg | 0.28 | 24% | 0.00056 | 0.00030 | 0.00053 | −0.2% | +1.7% |
10 kg | 0.40 | 39.5% | 0.00115 | 0.00061 | 0.00053 | −3.6% | −0.5% |
In newtons at 24 m/s: net change is −0.007 N to +0.057 N at 7 kg, and −0.151 N to −0.020 N at 10 kg.
Hoerner Wingtips:
A Hoerner tip has a drooped outer panel and a cut-off trailing edge.
Source: Hoerner, AF TR 5752 [2], reports +1–2% range and +1 ft/s climb for the best shape.
Geometry (per side): 0.12 m added span, droop 35° (the droop is another 0.084 m vertically), root chord 0.200 m, tip chord about 0.05 m.
Area: about 0.0183 m^2 per side including the droop, which is 5.3% of S, so cost = 0.053 × 0.01021 = 0.00054.
Saving: the same span effect as the round tip, plus about 2% more from the droop (−2.1%, my VLM estimate). That gives 18.7% at k = 1 and 11.0% at k = 0.5.
Case | CL | Saving (k=1) | Saving (k=0.5) | Cost | Net (k=1) | Net (k=0.5) |
|---|---|---|---|---|---|---|
7 kg | 0.28 | 0.00061 | 0.00036 | 0.00054 | −0.5% | +1.4% |
10 kg | 0.40 | 0.00125 | 0.00073 | 0.00054 | −4.2% | −1.1% |
Rounded/Elliptical Wingtips:
A rounded tip is a semi-elliptical cap added beyond the current tip. The chord shrinks to zero at the end, so the outer part carries less load than a raked tip.
Geometry (per side): 0.12 m long, root chord 0.200 m.
Area: 2 × (π/4 × 0.12 × 0.200) = 0.0377 m^2, which is 5.5% of S, so cost = 0.055 × 0.01021 = 0.00056.
Saving: it adds the same span, so the saving is the same as the raked tip: 17.2% at k = 1 and 9.2% at k = 0.5. Expect it to land near k = 0.5, because the tip loading falls off.
I found no small-UAV test of a round tip against a raked tip.
Case | CL | Saving (k=1) | Saving (k=0.5) | Cost | Net (k=1) | Net (k=0.5) |
|---|---|---|---|---|---|---|
7 kg | 0.28 | 0.00056 | 0.00030 | 0.00056 | 0.0% | +1.9% |
10 kg | 0.40 | 0.00115 | 0.00061 | 0.00056 | −3.5% | −0.3% |
Endplate Wingtips:
An endplate is a flat vertical plate at the tip. It adds almost no physical span, only a small effective span increase.
Geometry (per side): 0.10 m tall × 0.14 m chord = 0.014 m^2, so both plates are 0.028 m^2, which is 4.05% of S.
Cost: 0.0405 × 0.01021 = 0.00041. A flat plate and its wing junction add more than this, so my vortex-lattice-plus-junction estimate is 0.00078.
Saving: a low case of 4.0% and a high case of 8.6% (my vortex-lattice result).
Literature: plate parasite drag largely cancels the gain [4]. A tip fence gave −15.3% L/D at 0° on a NACA 4415 [5]. Note that it is one CFD result at 0°, so it does not show a general endplate penalty.
Case | CL | Saving (low) | Saving (high) | Cost | Net (low) | Net (high) |
|---|---|---|---|---|---|---|
7 kg | 0.28 | 0.00013 | 0.00028 | 0.00078 | +4.8% | +3.7% |
10 kg | 0.40 | 0.00027 | 0.00057 | 0.00078 | +3.0% | +1.2% |
Wind Tunnel Tests:
Study | Conditions | What they measured | Result |
|---|---|---|---|
Re about 390–430k; aspect ratio 7; SD7032 airfoil; best L/D at CL 0.55–0.64; winglets 15% of span tall | Two single winglets (tight corner and smoothly blended) and a raked tip, against a plain elliptical wing | Blended winglet: best L/D up 3.2%. Tight-corner winglet: up 1.3% at lower CL, then below the plain wing as CL rose. Raked tip: up 0.9%, lowest minimum drag. Winglets raised skin area 6–8% and minimum drag 1.2% | |
Re 400k; aspect ratio 8; rectangular wing; NACA 23015 | Six arrangements of three small winglets ("sails") at different tilts | Best L/D up 7.3%, but only above CL of about 0.45. At zero angle of attack L/D changed by −9% to +5%. Zero-lift drag rose from 0.032 to 0.035–0.040. The authors say the extra drag cancels the saving at low-lift cruise |
Findings:
Our cruise CL (0.28–0.40) is below the ~0.45 break-even where small winglets started to help in the São Paulo tests (rectangular wing, aspect ratio 8, NACA 23015) [3]. So expect no cruise gain from a winglet.
Gains are small even at higher lift. NASA's wind tunnel measured +3.2% for a winglet far taller (15% of span) than ours (6%); a CFD study reports more but is low confidence (see note).
The raked tip had the lowest minimum drag in the NASA test, which matters at our low cruise CL.
Note: A CFD study reports about +6% averaged over 0–12° for a blended winglet [5], but it is low confidence: lift varies about 5% across its meshes and it has no experimental validation. The same CFD study found the raked tip best when averaged across all angles of attack (+3.35% L/D, +3.2% at CL max), and that drag rose with winglets at 0–12° [5]. This supports the raked tip.
Selection: Raked tip
Factor | Raked tip | Winglet | Hoerner |
|---|---|---|---|
Drag at cruise CL 0.28–0.40 | Equal or slightly better | Equal or slightly worse | Equal or slightly better |
Lowest minimum drag in the NASA test | Yes [1] | No | Not tested |
Tip mass and vertical fin | None | Both | Droop adds a little |
Printing | Simplest (flat swept panel) | Harder (blend, cant, toe) | Harder (droop, cut-off trailing edge) |
Added span | +0.24 m | None | +0.24 m |
Rule-of-thumb sizing (NASA [1]):
Setting | Rule | Value for Helios M2 |
|---|---|---|
Rake start | Outboard of the aileron end (1.150 m) | 1.210 m (current tip), so no aileron conflict |
Rake length | 5–10% of semi-span | 0.12 m (range 0.06–0.12 m) |
Root chord | Match the wing tip | 0.200 m |
Tip chord | About 0.5 × root chord | 0.100 m |
LE sweep | 30–45° | 40° |
Trailing edge | Straight, keeps the hinge line clean | Unswept |
Dihedral | Start flat | 0° |
Twist | Continue the wing washout | No extra twist |
Blend | Rounded, no sharp corner (the NASA tight-corner winglet lost L/D as CL rose [1]) | Smooth leading-edge radius |
Airfoil | Same as wing | NACA 4412; 12 mm thick at 0.1 m chord (check print wall thickness) |
Condition: our team must accept +0.24 m of span (transport, launch limits, etc.). If not, we should use the winglet or the plain tip.
The differences between devices are within the method's error. Hoerner beats the raked tip by only 0.5–0.7 percentage points, about 0.03 N, and part of its advantage is my own VLM estimate. The choice rests on structure, printability and risk.
Validation plan:
XFLR5 VLM runs of the baseline, raked tip, plain extension, winglet and Hoerner tip at CL 0.28 and 0.40, Re about 330k. Compare drag in newtons, not percent.
Estimate root bending moment and tip mass from the SolidWorks mass properties, and check print wall thickness at the tip (12 mm at a 0.1 m chord).
Optional: one Fluent run of baseline vs raked tip.
If built: a flight-test A/B at constant speed, comparing battery power or glide ratio. State what the test could and could not resolve, since differences of 0.1 N will not be visible.
Design Changes | Feedback From |
|---|---|
-Change self tapped m3 hole to 2.8mm -Adding more clearance above gps -Updating slots for indicator lights |
Jakob Boelling |
Sources:
[4] NASA TP-1020