Investigate if extra drag is correct #14

Activity

1. 1-Bart-1

   assigned

   1-Bort-1

   on Oct 6, 2026
2. 1-Bort-1

   added

   agent:queuedAgent task stateAgent task state

   on Oct 6, 2026
3. 1-Bort-1 commented on Oct 6, 2026

   

   1-Bort-1

   on Oct 6, 2026

   Last edited by 1-Bort-1

   Member

   More actions

   Answered, no repo change: the wing drag pitches the kite up into a better trim (more force at a fixed depower), and at equal AoA it costs about 35 % of the force. Figures and numbers are in the comments below.
4. 1-Bort-1

   added

   agent:runningAgent task stateAgent task state

   and removed

   agent:queuedAgent task stateAgent task state

   on Oct 6, 2026
5. 1-Bort-1 commented on Oct 6, 2026

   

   1-Bort-1

   on Oct 6, 2026

   Member

   More actions

   It doesn't double them, and what it does do is physical. I flew simple_fig8.jl (fixed length, default wind, phase-4 means) on all three projects with wing_drag_coeff 0, 0.007 and 0.03. Drag 0.007 raises the tether force by 2–4 %, and 0.03 by 4.5–13 %:

   project	F [N] (drag 0 / 0.007 / 0.03)	v_app [m/s]	kite speed [m/s]	L/D wing	L/D system	span-mean AoA [°]
   150 m	5013 / 5107 / 5242	24.97 / 24.57 / 23.28
   200 m	4825 / 4961 / 5168	24.89 / 24.61 / 23.47	23.95 / 23.65 / 22.54	8.14 / 7.66 / 6.47	4.32 / 4.27 / 4.08	4.58 / 4.88 / 5.83
   300 m	4177 / 4360 / 4723	23.56 / 23.49 / 22.85	22.56 / 22.48 / 21.88	8.43 / 7.91 / 6.66	3.96 / 3.93 / 3.82	4.52 / 4.82 / 5.74

   All nine runs pass all 8 success criteria. Which run showed you the doubling — project, wind speed, and which force? I couldn't find a setup here that does it.

   Why more drag gives more force. The drag does slow the kite: v_app is 6 % lower at 0.03 (200 m), so q is 11 % lower. But the force coefficient F/(q·A) rises from 0.726 to 0.874 (+20 %), and that wins. The chain:

   1. The point drag acts on the 20 wing nodes, and they sit about 10 m above the KCU along the bridle. At 200 m and 0.03, the added drag is 185 N (½ρv²·0.03·17.5 m² gives 178 N, so the force itself is right). Its pitch moment about the KCU changes by −1833 N·m, a lever arm of 9.9 m.
   2. The kite is a pendulum on its bridle, so it trims until the resultant passes through the KCU again. The whole kite pitches back, rigidly: the wing shape in the body frame moves by at most 9 mm. Relative to the wing, the apparent wind tilts from 9.80° to 11.45°.
   3. The AoA rises everywhere along the span by the same amount (span mean 4.58° → 5.83°), and C_L rises from 0.730 to 0.876.

   A real kite does the same: drag acting at the wing has to be balanced about the bridle point, and the only way is a higher AoA. So the sign is correct, and so is the drop in L/D. Wing L/D falls by 21 %, system L/D by only 6 %, because the tether drag dominates; that 6 % is what the kite speed follows. Where the drag sits along the chord hardly matters, since the lever arm is the 10 m bridle height and not the chord.

   

   

   

   In the 3D view the two shapes overlap to within 9 mm. Where only the light line shows, at the tips, that is a draw-order artifact, not a shorter wing.

   How it was measured

   • The 150/200/300 m projects, default wind, 90 s runs, turn-rate law and course loop unchanged (identified for 0.03). Means over phase 4 only.
   • L/D wing is var_15 and L/D system is var_16, both from V3Kite's step!. C_L and per-panel AoA come from the VSM solution, sampled every step through a probe copy of the script. The body frame is built from the geometry: z from the KCU to the wing centroid, y from tip to tip, x pointing from TE to LE.
   • With drag 0 the wing nodes still log 67 N of "drag". That is the bridle-line drag the segments hand to their attachment points (drag_in), not wing drag, so I subtracted the drag-0 run to isolate the wing drag's share.
   • Nothing in the repo was changed. Found on the way: the tracked Manifest-v1.13.toml.default no longer matches Project.toml (DocStringExtensions is missing from the package's deps), so a fresh install fails to load the package. That is Re-resolve Manifest-v1.13.toml.default: SimpleKiteControllers' entry... #15.
6. 1-Bort-1

   added

   agent:waiting-humanAgent task stateAgent task state

   agent:queuedAgent task stateAgent task state

   agent:runningAgent task stateAgent task state

   agent:steeredAgent task stateAgent task state

   and removed

   agent:runningAgent task stateAgent task state

   agent:waiting-humanAgent task stateAgent task state

   agent:queuedAgent task stateAgent task state

   on Oct 6, 2026
7. OpenSourceAWE

   deleted a comment from 1-Bort-1 on Oct 6, 2026
8. 1-Bort-1 commented on Oct 6, 2026

   

   1-Bort-1

   on Oct 6, 2026

   Member

   More actions

   Yes: the drag moves the kite to a better trim, and at equal trim it costs about 35 % of the force. Same 200 m panels (power and reel-out speed taken out), now with depower swept at drag 0 and the drag-0.03 run dashed:

   

   run	F [N]	v_app [m/s]	kite speed [m/s]	L/D wing	L/D system	span-mean AoA [°]
   drag 0, depower 0.23	8332	28.93	28.19	8.67	4.95	5.99
   drag 0, depower 0.25	6417	26.97	26.14	8.48	4.65	5.24
   drag 0, depower 0.27	4825	24.89	23.95	8.14	4.32	4.58
   drag 0, depower 0.29	3545	22.66	21.67	7.70	3.98	4.06
   drag 0, depower 0.31	2602	20.50	19.41	7.26	3.69	3.67
   drag 0.03, depower 0.27	5168	23.47	22.54	6.47	4.08	5.83

   • What the drag does: it pitches the kite up by about as much as 0.035 less depower. At depower 0.27 its AoA (5.83°) sits between the drag-free runs at depower 0.25 and 0.23.
   • At equal trim, the drag costs force. Interpolated to an AoA of 5.83°, the drag-free kite pulls about 7.9 kN, against 5.2 kN with drag 0.03 (about −35 %). Its system L/D is about 4.9 against 4.08, and its apparent wind about 28.5 against 23.5 m/s. So the drag behaves as drag should, once the trim is the same.
   • Why the force still rises at a fixed depower: the drag-free kite at depower 0.27 flies at an AoA (4.6°) where both C_L and L/D still grow with AoA. Over the whole sweep, less depower raises the force and both L/D values. Anything that pitches it up gains more C_L than the drag costs, and the drag's pitch-up wins at a fixed depower.
   • How sensitive depower is: each 0.02 of depower moves the tether force by 25–30 % at this wind, more than drag 0.03 does. Comparing drag values at the same depower therefore compares two different trims. Matching the force or the AoA, which is what the depower conversion identification does, is the fair comparison.

   All six runs pass all 8 success criteria (200 m, v_wind 5 m/s, same pattern f8_a 30°, f8_b 12°).
9. 1-Bort-1

   added

   agent:waiting-humanAgent task stateAgent task state

   agent:steeredAgent task stateAgent task state

   and removed

   agent:steeredAgent task stateAgent task state

   agent:waiting-humanAgent task stateAgent task state

   on Oct 6, 2026