Airliner horizontal stabilizer aero data?

[seanmcleod70]

Does anyone have any references to aero data for horizontal stabilizers of airliners?

Looking through all of the airliners in the JSBSim repo (737, 787, A320, MD11) I couldn't find one that models the horizontal stabilizer in terms of lift, drag and pitching moment. All of them simply model an elevator as opposed to modelling both.

[seanmcleod70]

@Octal450 I noticed that in your MD-11 repo at - https://github.com/Octal450/MD-11/blob/master/FDE/Config/md-11-aerodynamics.xml that you have separate aero data for the horizontal stabilizer and elevators. Where did you manage to find this sort of aero data?

[Octal450]

@seanmcleod70 The data is computed. The horizontal stabilizer is the coefficient for the stab portion + faired elevators, and the elevator portion is the delta from that faired position to deflected.

All my airplanes there use computed data for this, not any data available sorry.

Most FG simulate the whole thing as just elevators... which is WRONG.

Kind Regards,
Josh

[seanmcleod70]

@Octal450 can you share your computation formula/algorithm that you're using?

I also came across the A320 that you contribute to which also models the stabilizer and elevator separately.

https://github.com/legoboyvdlp/A320-family/blob/625c97cbb0d5e485547d4a8ee77a66ab16553b96/FDE/Config/a320-aerodynamics.xml#L707-L738

I did come across a public Boeing report commissioned by NASA from 1970 for the 747-100 which has aero data on the stabilizer and elevator.

• The Simulation of a Large Jet Transport Aircraft. Volume 1 - Mathematical Model
• The Simulation of a Jumbo Jet Transport Aircraft. Volume 2: Modeling Data

@Gijsrooy has taken this Boeing data and created a JSBSim FDM from it:

https://github.com/Gijsrooy/747-400/blob/e4634103d91cc85e58b0a55e9986573ee9e73130/747-400.xml#L4514-L4585

[Octal450]

It's part of my giant spreadsheet stuff, not one single thing. It all together. I can't tell you how accurate/inaccurate it is, because I don't have any of that data. Let me look through what Gijs' numbers are and see.

I highly doubt it's 100% right. My goal with FDE's is to match the published numbers best I can, and match expected "feel" as closely to I can.

Every one of my flight models has each surface modelling completely independently, except for spoilers which I combine from the FCS where they are separate into the aero.

Kind Regards,
Josh

[seanmcleod70]

One of the reasons I'm interested in trying to get somewhat accurate aero data for the stabilizer and elevator is to try and model some aircraft accidents in JSBSim which involved horizontal stabilizer trim issues, for example:

• Boeing MCAS accidents
• A320 crash into the Mediterranean

In particular where the stabilizer is producing a moment opposite to what the elevator is producing.

Are there combinations where the stabilizer over powers the elevator acting in the opposite direction?

The A320 crash investigation report mentions that the stabilizer ended up at -11deg (maximum range for electric input) so giving a large nose up moment, with no automatic trimming given they ended up in direct law. And that a maximum nose down input from the sidestick to the elevator couldn't overcome the pitch up moment, although there was also a pitch up moment from the engines which had been advanced to TO/GA thrust. It's not clear if the elevator moment could've overcome the stabilizer moment if the thrust had been reduced.

In the Ethiopian MCAS accident the engines were also at a high thrust setting, so producing a pitch up moment, but the combination of that plus the elevator input couldn't overcome the nose down pitch moment from the stabilizer.

Potentially useful snippet from the A320 crash investigation in terms of comparing what you have for CL vs alpha for the landing configuration.

the calculated angle of attack was between 21.7 and 23.2 degrees, very close
to the theoretical angle of attack for initiating the warning in landing configuration

[seanmcleod70]

Here is another example of a combination of stabilizer trim plus engine thrust causing an issue.

Report on 737 Approach Incident

They don't mention the particular stabilizer angle, presumably in the typical range used for trim on a landing approach, i.e. not at the maximum stop.

The trimmed position of the stabiliser, combined with the selection of maximum thrust, overwhelmed the available elevator authority.

Only once they reduced the thrust from maximum thrust to GA thrust were they able to pitch down.

A reduction in thrust lever position to a go-around (GA) thrust setting occurred 40 seconds after the go-around was initiated, which allowed sufficient nose‑down elevator authority to control the pitch-up couple.

But not before reaching a maximum pitch angle of 44 deg and slowing down to 82kt!

[Octal450]

Yes, trim can absolutely overpower elevator. Much bigger surface.

In FG most people sum and clip elevator and stab but unless you have a flying stabilizer like fighters or L1011, then that's a big wrong. I simulate the stab totally separately how it should be.

Kind Regards,
Josh

[seanmcleod70]

Much bigger surface.

Not completely as simple as that.

Take a look at your A320 aero data for the stabilizer and elevator.

Moment Direction	Stabilizer Stop	$C_{m_{\delta s}}$	Elevator Stop	$C_{m_{\delta e}}$
Nose Up	-13.5deg	0.557	-30deg	0.361
Nose Down	+4deg	-0.165	+17deg	-0.205

So if the stabilizer is moved to the stop for pitching the aircraft nose up, then yes the maximum elevator nose down input is smaller, -0.205 versus 0.557.

But if the stabilizer is moved to the stop for pitching the aircraft nose down, then there is enough elevator power to overcome it, 0.361 versus -0.165.

Need to do the same analysis on the Boeing published data for the 747-100. The Boeing data is a bit more involved since it takes into account density altitude, Mach and downwash based on main wing's alpha and flap position.

[Octal450]

Max deflection anu is 20degrees limited by the FBW.

Also I think the number is arbitrarily high in the A320. I dont remember why. Maybe I need to fix that.

Kind Regards,
Josh

[seanmcleod70]

Quick comparison with the elevator only approach of the 737 in the JSBSim repo.

Elevator range is symmetrical at 0.3rad (17.2deg).

jsbsim/aircraft/737/737.xml

Lines 770 to 784 in 2006648

	<function name="aero/coefficient/Cmde">
	<description>Pitch_moment_due_to_elevator</description>
	<product>
	<property>aero/qbar-psf</property>
	<property>metrics/Sw-sqft</property>
	<property>metrics/cbarw-ft</property>
	<property>fcs/elevator-pos-rad</property>
	<table>
	<independentVar>velocities/mach</independentVar>
	<tableData>
	0.0 -1.20
	2.0 -0.30
	</tableData>
	</table>
	</product>

There is a Mach dependence.

Mach	$C_{m_{\delta e}}$	737 Nose Down	A320 Nose Down	737 Nose Up	A320 Nose Up
0	-1.20	-0.36	-0.37	+0.36	+0.918
0.4	-1.02	-0.31	-0.37	+0.31	+0.918
0.8	-0.84	-0.252	-0.37	+0.252	+0.918

So the total nose down authority is quite similar, and I'm assuming that the pitching moment of inertia of the B737 and A320 are quite similar. However the total nose up authority is about 3 times more for the A320.

[seanmcleod70]

So some figures from the 747-100.

Assuming density altitude of 0ft, Mach = 0, k_factor = 1 (down wash etc.). I've combined the inboard and outboard elevators.

Moment Direction	Stabilizer Stop	$C_{m_{\delta s}}$	Elevator Stop	$C_{m_{\delta e}}$
Nose Up	-12deg	0.636	-25,-23deg	0.6675
Nose Down	+3deg	-0.159	+15,+17deg	-0.4485

So much like the A320 case, a maximum nose up moment from the stabilizer can't be overcome by nose down elevator, but a maximum nose down moment from the stabilizer can be overcome by nose up elevator.

So does the 737 not match this pattern given the MCAS accidents? Especially since the thrust from the low slung engines would also have produced a nose up pitch moment.

[seanmcleod70]

It's part of my giant spreadsheet stuff

@Octal450, you might find the Trim Envelope tool I mention in the following comment - #988 (comment) potentially useful as a way to visualize your models.

[Octal450]

I uhh, don't have any answers for you. I am sorry. Could it be that the force required at that speed to pull the elevator all the stop was too much?

I wonder if there is an NTSB report that might clarify.

Kind Regards,
Josh

[seanmcleod70]

@Octal450 out of interest, do you make use of the JSBSim trim routines for your aircraft like the MD-11 and A320 when setting them up in FlightGear?

[Octal450]

Hi,
I'm afraid I don't know what you mean by JSBSim trim routines.

Kind Regards,
Josh

[seanmcleod70]

The trim routines within JSBSim, accessed via simulation/do_simple_trim like in this example:

jsbsim/examples/python/TrimEnvelope.py

Lines 59 to 75 in 6b81abf

	# Iterate over a range of speeds and for each speed a range of flight path angles (gamma)
	# and check whether a trim point is possible
	for speed in range(min_speed, max_speed, 10):
	for gamma in range(min_gamma, max_gamma, 1):
	fdm['ic/h-sl-ft'] = altitude
	fdm['ic/vc-kts'] = speed
	fdm['ic/gamma-deg'] = gamma
	# Initialize the aircraft with initial conditions
	fdm.run_ic()
	# Trim
	try:
	fdm['simulation/do_simple_trim'] = 1
	results.append((fdm['velocities/vc-kts'], fdm['aero/alpha-deg'], gamma, fdm['fcs/throttle-cmd-norm[0]']))
	except jsbsim.TrimFailureError:
	pass # Ignore trim failures

I'm not really familiar with FlightGear, so I'm not sure if they implement their own trim routines or whether they offer a pass-through mechanism that invokes the trim routines supplied by JSBSim.

So for example, say you wanted to setup your MD-11 on approach at some specific location at say 5nm from the runway at 1,500ft in landing config in terms of flaps, gear etc. and with some specific weight, specific cg, possibly specific atmosphere etc. at some speed and say flight path angle of 3 deg with everything in trim. How would you go about doing that?

[Octal450]

FlightGear does some trim thing on load up. I don't do anything with it unless it reports failure to trim which usually means some invalid configuration.

Starting in air is not supported currently, that messes with the systems. FlightGear can position planes there, so at some point I plan to handle it.

TL:DR, this type of trimming is done transparently to aircraft developers in FlightGear

Kind Regards,
Josh

[seanmcleod70]

FlightGear does some trim thing on load up

I'm assuming in this 'load up' case you're talking about starting on the ground, in which case FlightGear is more than likely invoking the ground-trim routine of JSBSim, which is simpler to solve unless you have some issue with the gear.

Whereas trim in the air would typically be full-trim.

jsbsim/src/initialization/FGTrim.h

Lines 67 to 68 in 6b81abf

	typedef enum { tLongitudinal=0, tFull, tGround, tPullup,
	tCustom, tTurn, tNone } TrimMode;

this type of trimming is done transparently to aircraft developers in FlightGear

However, that's the thing, it's not completely transparent to the aircraft developer. If an aircraft developer wants their aircraft to be able to be trimmed via the JSBSim trim routines then there are some requirements that they need to be aware of. Which is why I was asking if your aircraft are trimmable, in the air in particular.

You've mentioned one potential issue already, i.e. non-trivial systems. But even excluding non-trivial systems, even an aircraft with no systems has some basic requirements if the JSBSim trim routines are going to work.

For example here are the axes and the controls that the full trim routine works with.

jsbsim/src/initialization/FGTrim.cpp

Lines 834 to 846 in 6b81abf

	case tFull:
	if (debug_lvl > 0) {
	FGLogging log(fdmex->GetLogger(), LogLevel::DEBUG);
	log << " Full Trim\n";
	}
	TrimAxes.push_back(FGTrimAxis(fdmex,&fgic,tWdot,tAlpha));
	TrimAxes.push_back(FGTrimAxis(fdmex,&fgic,tUdot,tThrottle ));
	TrimAxes.push_back(FGTrimAxis(fdmex,&fgic,tQdot,tPitchTrim ));
	//TrimAxes.push_back(FGTrimAxis(fdmex,&fgic,tHmgt,tBeta ));
	TrimAxes.push_back(FGTrimAxis(fdmex,&fgic,tVdot,tPhi ));
	TrimAxes.push_back(FGTrimAxis(fdmex,&fgic,tPdot,tAileron ));
	TrimAxes.push_back(FGTrimAxis(fdmex,&fgic,tRdot,tRudder ));
	break;

Take the pitch trim, it means that the aircraft needs to have an implementation such that fcs/pitch-trim-cmd-norm will respond during the trim routine to adjust the pitch moment, i.e. to either move the stabilizer or elevator so that the trim routine can find a value that zeroes out the pitching moment for the particular trim conditions.

[Octal450]

Ahh I see.
Well I don't use fcs/pitch-trim-cmd-norm etc. My flight control system takes the input directly from flightgear and computes each surface position.

even if I used that, the hydraulics/etc will not be powered when JSBSim tries to in-air-trim and thus the surface won't move.

So this doesn't really work for things like that.

When I eventually implement states I'll probably have it spawn with the autopilot engaged which will put it into trim. At any in-flight state, the trim values will not differ TOO much from the initial condition.

Kind Regards,
Josh