Repository navigation
ELAC: the high-AoA protection acts on an alpha lagged by 2 s, and keeps swinging after a fast entry #11009
Unanswered
bochen2029-pixel
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
In
fbw-a32nx/src/wasm/fbw_a320/src/model/ElacComputer.cpp(master, 66021c2), the ELAC passes theangle of attack it takes from the ADRs through a first-order lag before anything uses it:
(lines 545–546).
LagFilter_C1is 0.5 (ElacComputer_data.cpp, line 103), andElacComputer_LagFilter(lines 132–149) is the bilinear form of C1 / (s + C1), so
rtb_Yfollows alpha with a time constant of2 s. The high-AoA protection acts on that lagged alpha. It engages when
rtb_Yexceeds α prot (line 1161),and stays engaged until the stick is pushed forward — past about 3 % of its travel for half a second while
rtb_Yis under α max, or past half its travel at once — or, below 200 ft radio height, untilrtb_Yfalls 2° under α prot (lines 1157–1165). The pitch normal law receives
rtb_YasIn_alpha_deg(theLawMDLOBJ5.stepcall from line 1435,&rtb_Yon line 1438), lags it again by 0.625 s(
PitchNormalLaw.cpp, line 1874,LagFilter1_C1= 1.6), and regulates the difference between that andthe alpha the stick asks for, with the difference's rate.
We run the ELAC, SEC, FAC and FCDC unmodified, outside MSFS, against JSBSim airframes — an A320 model of
our own and JSBSim's 737 — and log the ELAC's own outputs every step. From level flight at 150 kt and
8,000 ft in CONF FULL (α prot 10.6°, α max 13.0°), with full back stick held for 20 s, our A320's true
alpha passes α prot 1.6 s before the lagged one does, and the protection engages only then, with the
true alpha at 17°. It pushes on a lagged alpha still near 11° while the true alpha falls to −4°, then
pulls on a lagged alpha near 1° while the true alpha rises to 20°; the swing repeats about every 9 s,
and is still between 5° and 17° when the stick is let go. JSBSim's 737 overshoots to 19.5°, swings down
to 2.7° and back to 14.3°, and settles at α max.
Whether an airframe keeps swinging is partly its own: its short period and its pitch response near the
stall. Fitting our A320's pitch response to cited data — MIL-F-8785C's short-period limits, published
elevator-hold measurements and NASA's GTM post-stall data — every setting inside those bands still
swings back to 16° to 20°, the first overshoot 15.3° to 16.7°, wherever the fit puts the airframe. So we
changed
LagFilter_C1alone, to 5 (a 0.2 s lag), in a copy of the same sources, and flew the samemanoeuvres:
On our A320, then, the lag is what keeps the protection swinging: with it at 0.2 s the same pull settles
toward α max within the hold, where with 2 s it still swings between 5° and 17° when the stick is let
go. The lag is not what makes the first overshoot, which stays near 17° though the protection engages
1.4 s sooner — the fit found it at 15.3° to 16.7° whatever the airframe — and a shorter lag is not free:
on the 737 it lowered the first overshoot and drew out the swings after it. Entered slowly, the
protection holds α max more closely with the shorter lag on both.
We don't know whether the lag is meant to be this long — the vanes' noise may call for a filter — nor
whether MSFS's flight model, which the laws are tuned against, swings as ours does; 0.2 s was a
diagnostic, not a proposal. What we would ask: whether the protection should engage on, and regulate,
an alpha with less lag, or with some phase advance — an alpha-rate or pitch-rate term — so that after a
fast entry it does not keep acting on where the aircraft was.
All reactions