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

   Queued as RobustKiteControllers.jl-14.
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

   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. 28 remaining items

   Load more
8. 1-Bort-1

   added

   agent:queuedAgent task stateAgent task state

   agent:runningAgent task stateAgent task state

   agent:doneAgent task stateAgent task state

   agent:blockedAgent task stateAgent task state

   and removed

   agent:runningAgent task stateAgent task state

   agent:queuedAgent task stateAgent task state

   agent:doneAgent task stateAgent task state

   on Oct 23, 2026
9. 1-Bort-1 commented on Oct 23, 2026

   

   1-Bort-1

   on Oct 23, 2026

   Member

   More actions

   Blocked. turned back 3 times in the last hour. The last of them: its session ended blocked, so the pre-push gate refused: uncommitted changes: LocalPreferences.toml
10. 1-Bort-1

    added

    agent:doneAgent task stateAgent task state

    and removed

    agent:blockedAgent task stateAgent task state

    on Oct 23, 2026