Break wing inflight (+ gears on ground) due to excessive forces

wlbragg commented 9 years ago

This is one I really want to do. We have the modeled effect and the consequent FDM effect in place. I need the force calculations from the FDM?

wlbragg commented 9 years ago

I just noticed #123 is this the same or tied to this new issue I created?

dany93 commented 9 years ago

Inflight, wing(s) can break because an excessive speed or by over-g's (excessive load factor). Both due to an excessive force in reality. But in FG not the same properties. /accelerations/n-z-cg-fpf_sec /fdm/jsbsim/accelerations/Nz, pilot-gdamped also accelerations along y and x axis I do not know if you need all of them. I should say that it is enough. Force limits are are difficult to estimate.

onox commented 9 years ago

You could take a look at how the f-14a implements this.

wlbragg commented 9 years ago

I'll do that, thanks.

wlbragg commented 9 years ago

Dany added some new FDM effects tied to these forces. We are currently testing them using a non-branched source. So far so good, they may need some fine tuning. It's a real sinking feeling to have a wing just quit working.

wlbragg commented 9 years ago

Wow, Dany, does our existing FDM Juanvvc:master have the same spin effects when you loose the flight envelope as you have in dany93? The one wing "damage", INCREDIBLE. I tried to limp along back to the airport and screwed around till I got into a spin at low altitude, oops! This is really nice. Again my question, does our existing Juanvvc:master FDM have the same spin effects when you loose the flight envelope as you have in dany93 FDM?

I'm ready to merge Dany's FDM into master along with this new FDM damage model. It is incredible gang. Any comments anyone?

dany93 commented 9 years ago

does our existing Juanvvc:master FDM have the same spin effects when you loose the flight envelope as you have in dany93 FDM?

Yes. it is the same for the "spin and stall" effects. I have not changed the general FDM other than for damages on left or /and right wings. Easy to test: fly with the FDM from juanvvc (no damage effects), cut the throttle, hold the stick back, up to stalls. Worsen it with more or less rudder (see the slip ball). I wonder if you do not mix two effects:

one which has existed for a long time, the spin and stall effects, with can give a spin roll on stall, depending on your flight asymmetry (that is the original feature of this FDM, from the beginning you had it). That is when you are close to the edge or out of the flight envelope (high angles of attack),
this very new damage effect, which is independent of the stall and spin (although a partly damaged wing can lead you to a stall and spin, as it seems to have happened to you).

Did I understand well and answer your question? mmh... pardon, what does "incredible gang" means?....

wlbragg commented 9 years ago

"incredible gang", means I really like your changes and also I really like the original spin modeling and I am telling our entire gang, working on this aircraft, that it is incredible work.:) Even though that spin was original, that is the first time I ever experienced "getting out of the flight envelope", other than meeting the ground, can you believe that! Yes, I believe I left the envelope because of the addition of the wing damage. I wish it happened at greater altitude so I would have had a chance to try to pull out of it, (or drop out of it).

"Did I understand well and answer your question?" Yes

wlbragg commented 9 years ago

Dany, I had to put your new FDM damage model on hold for a few days. I have to get the MP working first, I think it needs to take priority. I'm already sold on your changes though and want to replace any existing FDM damage modeling with it where appropriate (where the new and old overlap). I hope to get back to it by early next week. The MP is a big bite to swallow.

dany93 commented 9 years ago

OK, thanks for the information, I had more or less understood that.

When you have made your mind and if you agree for the new controls, I can clean the physics.nas from the conditional FDMs and the previous useless properties. Also if you like, update the "original" FDM.

I will let you for the damages on the ground.

wlbragg commented 9 years ago

"I can clean the physics.nas from the conditional FDMs and the previous useless properties. Also if you like, update the "original" FDM."

Please do, thank you!

dany93 commented 9 years ago

I'll do it as soon as I can. This night if I can. If not, not sure for sunday... Sorry.

dany93 commented 9 years ago

Done. http://www.mediafire.com/download/3tbxwr9upunmu4j/C172p-original-FDM-damage-20150510.7z

Main file: "original" FDM with wing-damage/left-wing, /right-wing properties.

  <fdm>
    <jsbsim>
        <wing-damage>
            <left-wing type="double">0</left-wing>
            <right-wing type="double">0</right-wing>
        </wing-damage>
    </jsbsim>
  </fdm>

I have also included the physics.nas file that I have used to test it. In this, I have deleted (commented) the conditions in FDMs to be as close as possible as it will be. That's a working file, of course it needs cleaning. However, I think it can help. It is more complete and better done than the previous one. If you start from one, please take this one.

When you have made your modifications in physics.nas, we will have to delete the

        <wing>
            <broken-one type="double">0</broken-one>
            <broken-both type="double">0</broken-both>
        </wing>

in c172p-detailed-set.xml (Not before, we would have nasal errors)

Still to observe if it works well for every features, I guess you have noticed that testing is not at all easy (almost stressful, you have to be quick on Ctrl-U and recovering...)

V > 158 kt and flight asymmetry ==> one wing damaged (can still fly, degraded) V > 180 kt and flight asymmetry ==> one wing broken V > 200 kt ==> both wings broken

g > 4.4 and flight asymmetry ==> one wing damaged (can still fly, degraded) g > 5.5 and flight asymmetry ==> one wing broken g > 6.6 both wings broken

at which we might add: (??) V > 180 kt and symmetric flight ==> both wings damaged (degraded lift)

Also connected to https://github.com/Juanvvc/c172p-detailed/commit/a8b98ccbcbe41df1c6f85528bf6d1f654c7220fd#commitcomment-11078203

wlbragg commented 9 years ago

"at which we might add: (??) V > 180 kt and symmetric flight ==> both wings damaged (degraded lift)"

I like the idea...

dany93 commented 9 years ago

==> On the "To do" list (when it is stabilized, I will pull from juanvvc, modify and do a Pull Request)

dany93 commented 9 years ago

V > 180 kt and symmetric flight ==> both wings damaged (degraded lift)"

Done (currently only on my disk).

Moreover, prior to adding this into my branch and doing a Pull request, I' d like to have your opinion about some form changes in physics.nas.

only one value for vne (no longer vne2), other values calculated,
vne and max-g above these thresholds calculated by a factor (idea: max-max = 150% of specifs in terms of forces).

(I have taken max-positive-g for "Normal" category, which lowers it: 3.8 instead of 4.4)

Which gives, for the main changed lines:

# Specifications Vne=158 KIAS,
# Normal category: max positive-g=3.8, max negative-g=1.52,
# Utility category: max positive-g=4.4, max negative-g=1.76,
# structure holds at least 150% max g's
props.globals.initNode("limits/max-positive-g", 3.8, "DOUBLE");
props.globals.initNode("limits/max-negative-g", -1.52, "DOUBLE");

#Over-speed damages  
    if (getprop("velocities/airspeed-kt") > (getprop("limits/vne") * 1.14)) # 180 KIAS
    {
        if (roll_moment > -4000 and roll_moment < 4000 and getprop("/fdm/jsbsim/wing-damage/left-wing") == 0 and getprop("/fdm/jsbsim/wing-damage/right-wing") == 0)
        {
            setprop("/fdm/jsbsim/wing-damage/left-wing", 0.3);
            setprop("/fdm/jsbsim/wing-damage/right-wing", 0.3);
            screen.log.write("Overspeed. Both wing DAMAGED ");
            gui.popupTip("Overspeed. Both wing DAMAGED", 5);
        }

    if (getprop("velocities/airspeed-kt") > (getprop("limits/vne") * 1.225)) # Vne x sqrt(1.5) ~ 200 KIAS

#Over-g damages
    if (g > (max_positive * 1.25))
    if (g > (max_positive * 1.5))

Also,

You have seen my warning red messages on the screen for damage information. I've changed them for Popup messages (more visible).
I have tested a landing with both wings damaged: the engine stops when I am just above the ground. It comes from your line nr 235:

    if(getprop("position/altitude-agl-m") < 10 and (getprop("/fdm/jsbsim/crash") or getprop("/fdm/jsbsim/wing-damage/left-wing") or getprop("/fdm/jsbsim/wing-damage/right-wing")))
        killengine();

Another criteria should be chosen for "killing" the engine (at least one wing broken? nose wheel broken? very strong g's?).

Also if you prefer, the engine can be killed thanks to the (new) Float chamber, Systems/Fuel.xml, with

fdm/jsbsim/propulsion/tank[2]/priority = 0

Can you give me your opinion and, possibly, your agreement on these changes, for a Pull request?

wlbragg commented 9 years ago

I good with all of it. I'll rework the kill-engine condition for nose gear broke and wing "broken". Plus I'll give it some more thought as to the original reason I kill it.

dany93 commented 9 years ago

Right, thanks wlbragg. Also, see my Pull request comments https://github.com/Juanvvc/c172p-detailed/pull/197#issuecomment-104210991

onox commented 9 years ago

This can be closed?

dany93 commented 9 years ago

Yes, IMO.

onox commented 9 years ago

I noticed though it is very hard to break the gears while keeping the wings not broken. Should we open a new issue for this?

dany93 commented 9 years ago

True.

In physics.nas, lines 227 and 231 (break threshold for each gear)

     if(getprop(gears~"unit[1]/compression-ft") > 0.49 or getprop(gears~"unit[1]/broken"))
     if(getprop(gears~"unit[2]/compression-ft") > 0.49 or getprop(gears~"unit[2]/broken"))

A break threshold (compression-ft) of 0.25 seems enough in practice (to be tested at full load, which I have done). Even a bit less (0.22 ?) might be enough.

Not obvious from the calculation: FDM: spring coefficient 5400 LBS/FT for each main gear; 0.25 ft compression corresponds to 5400 x 0.25 =1350 lbs per gear. If we assume both gears at touching: 2700 lbs for 0.25 ft compression. Which, with at full load = 2400 lbs, gives 2700 / 2400 = 1.125 g. Not much... should be easily attained... But I guess the damping helps to avoid reaching the compression threshold in such a short time. Or is the compression threshold a good criteria? Wouldn't the g's be a better one for the forces on the gears (damping included)? But the compression has the advantage of "measuring" the stress separately on each gear... Better IMO. Decrease the damping in order to reach the compression threshold on a hard landing?

Anyway, if the result is plausible in practice, it can be considered as acceptable...

I let you think about it (no errors?) and do the tests.

onox commented 9 years ago

Decrease the damping in order to reach the compression threshold on a hard landing?

I think we should decrease the damping instead of the compression-ft. Do you agree?

If I add a co-pilot and some baggage (2200 lbs gross weight), it becomes much easier to break some gear. But I think that even with just the pilot, a user should be able to break the gear when the aircraft is deliberately dived into the ground :stuck_out_tongue_closed_eyes:

wkitty42 commented 9 years ago

hey now! I resemble that comment :laughing: but I'm actually trying to land decently... not bend and break things... but I'm getting better :wink:

dany93 commented 9 years ago

I think we should decrease the damping instead of the compression-ft. Do you agree?

Yes, I think that would be be a good thing. I've not tested it, I had this idea just at the end, after my small calculations.

tigert commented 9 years ago

The nose gear compression to break it is a bit problematic, as I think the gear can sustain a compression to the end, if the force is straight, the nose gear arm is facing a bit forward.

What I am trying to test is the porpoise, where you land with excessive speed and often the aircraft starts leaping wiht increasing amplitude if you don't immediately arrest the movement by pulling back on the wheel and maybe add a bit of throttle to smoothen the next touchdown. If you do the very common mistake of pushing when the aircraft bounces (because you want to land! natural instinct) it will increase the size of the next bounce and eventually this leads to collapsing the nose gear.

This is the very common case when the gear breaks and it would be nice to be able to model that. I tried to test this by reducing the needed compression value a bit, and got it somewhat there, since I think it is a bit too strong initially, but then the problem is, that it also breaks if you put full throttle with parking brake on and push the stick forward. This compresses the gear too, and it collapses too. This is a bit unexpected, even though you usually don't hold brakes and push forward anyway. But maybe the compression ratio is a bit wrong parameter to check alone for gear collapsing. Side loads would be interesting to check too, like on crosswind landings with nasty bounces, it could collapse. But I don't know if those are available that easily from the FDM?

//T

onox commented 9 years ago

Should we perhaps compute the force on the nose gear by computing pitching moment divided by distance from center to nose gear? (Or can we get the force directly from JSBSim?)

dany93 commented 9 years ago

What do you mean by pitching moment (which property)? fdm/jsbsim/moments ? I have not found the forces on gears directly from jsbsim. A rigorous way should be (if I understand well) Force = (compression-ft) x (spring_coeff) + (compression-velocity-fps) x (damping_coeff) I have not tested it.

I'm struggling with this breaking threshold for the main gears. I would prefer using properties directly available than swamping the (slow) nasal with these calculus for such a secondary behavior. But that's trial and error and compromises.... Maybe the above calculus will be the best solution.... If it is so simple, not having this done by JSBSim is weird.... JSBSim would do it much faster than nasal. Unless using switches, but we still add xml lines just for this....

onox commented 9 years ago

I meant the moment around the y-axis. M = F * d.

dany93 commented 9 years ago

I understood, but which property do you take for this moment?

onox commented 9 years ago

Perhaps /fdm/jsbsim/moments/m-gear-lbsft?

dany93 commented 9 years ago

That was my first question here, with this possibility. https://github.com/Juanvvc/c172p-detailed/issues/134#issuecomment-107653548. I don't understand well what this moment is. From which axis? To counteract which moment when stopped on the ground? If we have to do a calculus, I prefer Force = (compression-ft) x (spring_coeff) + (compression-velocity-fps) x (damping_coeff) It is direct, easy to understand and rigorous. It can easily be applied separately on each gear. I will try this, to see if I've understood well and if it works.

onox commented 9 years ago

Ok, your equation seems better.

dany93 commented 9 years ago

I have a test file (working file). http://www.mediafire.com/download/wud4yvfdtbp80hq/C172-gear-damage-force-20150602.7z

Only for the principles, kind of feasibility, to get your opinion.

FDM: only the damping for main gears is changed (400 instead of 1600 LBS/FT/SEC, lines 192 and 207). Does not seem very important, but I think 400 is better.
physics.nas: TEST FILE For replacement. See lines 319 - 337. Important: MUST be the same coefficients as spring_coeff and damping_coeff in the FDM for each gear.

Instead of breaking the gears and having to repair each time, I have made the forces on gears printed in the console above a threshold, which can be considered as the future breaking-threshold. By lowering the threshold, it can be used as a tool to measure these forces rather than doing trial and error. The first values to be fiddled are the threshold forces. For the nose, 1350 lbs is necessary to not break it on strong and long braking. It "breaks" on a strong nose contact. For the main gears, I hesitate between 1500 and 2000 lbs. It must hold at full load and for crosswind landing.

More difficult to break at default loading, but if FG works well, the principle seems realistic.

Some more complication in the nasal file because of the calculation lines, but after that the breaking criteria is obvious and the same as IRL.

Promising IMHO.

@tigert, if you can? Also, I have found a global gear-Y-force in the FDM. It might be used to break the gear in crab landing or bad crosswind landing. I've not still tested.

onox commented 9 years ago

@dany93 Just create a branch in your own repository:

git checkout -b gear-changes
Make the changes and commit them
git push -u origin gear-changes

dany93 commented 9 years ago

Done. https://github.com/dany93/c172p-detailed/tree/gear-damage

dany93 commented 9 years ago

~~How can I recover the content I had in my branch "master" before these changes? I've done git checkout master then git pull but, in my working directory, the files that I see on my disc inside my github folder are still with my last changes (physics.nas).~~ Sorry. I was mistaken by the recent file-dates but the file contents are the initial ones. OK

onox commented 9 years ago

@dany93 Thanks. I'll test this tomorrow, it's night time currently.

dany93 commented 9 years ago

I have added a "print" in the console for the side force on gears (along Y-axis). tigert wrote:

Side loads would be interesting to check too, like on crosswind landings with nasty bounces

@tigert: what do you think we should set as the breaking threshold for this side force? I'd say between 700 and 1000 lbs. It must hold crosswind landings at full load (not easy).

dany93 commented 9 years ago

@wlbragg I have not understood well your links between the damages on wings and the 3D representation. As well as the damage names, the "var" in physics.nas

What I have seen:

wing(s) slightly folded upwards with a high diedra angle outboard of the struts. I'd say that's right for wings "damaged" (still able to fly); Typically partial damage inflight.
wings folded downwards, touching the ground. Good for a very hard landing. Can be considered as "broken" on the ground.
What I have not seen: one or both wings really broken inflight (half of them lacking outboard of the struts, to extend what you have already done). Possible with not too much work?

Will you take care of this when the breaking thresholds on the ground are set? If I have to do it, I will need the properties or nasal variables in connection with the 3D results. And to agree, on the ground, which gear damage (or over-g's on the ground) makes which 3D wing damage.

onox commented 9 years ago

When you define a variable for the first time, you should add var so that it is a local variable. When you later assign a new value to it, you don't need to use var again.

dany93 commented 9 years ago

Yes, but this is a working file. I did not want to write everywhere. In the "good" version, the variables will be defined as var = at the top of the file, like those for wing damages. Inside the loop, only the name of the variable will be called.

onox commented 9 years ago

Ok, I'll start testing your branch in an hour or so.

onox commented 9 years ago

@dany93 I have tested your branch. The forces are nicely printed in the console. I think gear should break > 2000. Can you implement this in the current version? (Your master branch is 32 commit behind) We can later tune the 3 numbers (the forces) if necessary.

dany93 commented 9 years ago

OK, the important was your agreement on the principle. I update my master branch, introduce the gear breaking instead of the printing, and I do a pull request. (Tomorrow, now that's for me that it is late and I do not like make things in a hurry)

wkitty42 commented 9 years ago

I was just playing with the current version and testing if the wings would fall off from overspeed... this is not with @dany93 stuff you have been talking about in recent posts... I got up to ~10000 and threw her into a dive... from chase or heli view, I saw "wing damage" reported but couldn't see anything... ok... then I saw the wings break outboard of the struts and the wings broken report... I kept on going looking to see if they'd come off but they didn't... it was a dance try to recover and she dived into the ground only to bounce some 2000 back up into the air :laughing: I did get into some spins but I don't know enough about how to recover from them... is quite freaky seeing everything spinning around from inside the cockpit and not knowing which way is which... I hope I never experience it in RL...

onox commented 9 years ago

The wings can be heavily damaged, but cannot be separated. That would be cool though :smile:

dany93 commented 9 years ago

The wings are broken "in the FDM", but the 3D wing representation do not completely break them, it is a matter of animation. This is the subject of my question to wlbragg in this message https://github.com/Juanvvc/c172p-detailed/issues/134#issuecomment-108431906.

I this situation, the "Repair" option is miraculous. If you are too close to the ground, Ctrl-u and some touch (pull gently) can help. No parachute is simulated... Even with "Repair", try slowing down the aircraft before repair (by strongly pulling the yoke until being below the bad speed) or the wings will break again. Stressful, isn't it? :smile:

wkitty42 commented 9 years ago

hahaha, i played the CTRL-U card numerous times, yes... as for slowing down, a lot of times after the breakage i was close to stall speed...

speaking of which, i haven't heard the stall horn in a while... i have to have my engine sounds turned down to hear the radios clearly... is the stall sound in the same ""voice"" as the engine sound? yeah, i now... probably another issue ;)

wlbragg commented 9 years ago

dany, "What I have not seen: one or both wings really broken inflight " I think we can do that, you want from the support strut to the tip "gone" broken off, yes?

c172p-team / c172p

Break wing inflight (+ gears on ground) due to excessive forces #134