Wing flexing in 1.7.1

[Azimech]

Just now, Gargamel said:

Understandable, but at this point, we have identified an issue, and we need to narrow down it's root cause.   Only way the devs can get around to fixing it. 

Just a reply that I find it strange I can't like more posts per day with 7.5k rep. This like was aimed at you.

[Klapaucius]

1 hour ago, Gargamel said:

Try a clean install and then build the plane from scratch.  See if it still wobbles.   It might be the importing of existing designs, or it might be in the new code.

I just did a clean install.  The plane still flexed, but...I went back to the SPH and simply pulled it in half--the whole thing is built around an I-beam.  I then just reattached those two halves and the problem was 80% solved. I did need to add an extra set of struts on the front, but that is a vast improvement.  Once I did that, I had no flex at all.

So your suggestion about rebuilding has merit.  

[HebaruSan]

The video in the OP reminds me of the pre-autostrut, post-U4-wheel days. A heavy fuselage attached to wings attached to landing gear used to just break in half on launch, with the wheels acting like fulcrums and the wings like levers. Landing gear autostruts were introduced to work around that.

[MR L A]

2 hours ago, Klapaucius said:

3 hours ago, MR L A said:

Not sure why you would need struts for a launch vehicle these days anyway... rigid attachment all the way. Unless your designs are more than a little unorthodox... 

Rigid attachment generally does not work well on planes.  Heavier ones need a bit of give on landing or they just shatter. I almost never use rigid attachment for that reason.

See bold, lol. He was talking about a launch vehicle, not a plane.

[Klapaucius]

1 hour ago, MR L A said:

See bold, lol. He was talking about a launch vehicle, not a plane.

Well, that could include a spaceplane...Was not sure.

[Yakuzi]

Thanks for the craft file @Klapaucius. In KSP 1.7.1, I could reproduce the wing flexing behaviour as you reported in the OP. No flexing in 1.6.1 or 1.7.0, so it seems this issue was introduced in 1.7.1.

There was no mention of any autostrut adjustments in the changelog (see below), so it seems to be a bug. I've upvoted the associated report on the bugtracker that was linked earlier in this thread. Then again, it might be a separate issue, as 5thHM said he experienced bug #22733 prior to 1.7.1.  Edit: nevermind, read that wrong.

@JPLRepo, any idea on what's going on?

Quote

 

ChangeLog:
=================================== v1.7.1 ============================================================
+++ Improvements
* Implemented Cargo part category and Inventory processing.
* Implemented PAW UI drop-down grouping functionality.
* Implemented scrollbars to PAW UI windows.
* Added a new type for PAW fields, a double slider to set ranges with a min and max values
* Kerbal Portraits now available on EVA.
* Implemented Axis groups.
* Implemented Action/Axis group override sets (if 'Activate Action Sets' is enabled)
* New game setting - Shadow casting for CBs.
* Add and apply High Gee settings to wheel suspension.

+++ Localization

+++ Parts
* Fix Bobcat, Wolfhound, Kodiak, Poodle, Terrier, Spark, Flea, Hammer SRBs not recognising thrust obstructions.
* Fix Twitch, Spark ISP engine value.
* Fix Flea and Hammer white variant using incorrect texture

+++ Bugfixes
* Fix for maneuver mode UI being interactable while the Pause Menu is active
* Fix for maneuver nodes collapsing to delete mode if moused-over quickly after switching nodes with the maneuver mode UI.
* Fix maneuver node data being copied across nodes in some circumstances when using the maneuver node editor.
* Fix NRE when root part is delete and fuel overlay is turned on.
* Fix Menu Navigation when in the Game Difficulty > Advanced menu.
* Fixes to wheel suspension spring and damper settings.
* Fix wheel damage on docking, undocking and loading on high Gee CBs.
* Fix wheel friction.
* Fix engineer report showing incorrect mass for vessels containing parts with variable part module mass.
* Fix simply clicking on maneuver node resetting the node burn.
* Fix science experiment overwrite dialog text.

+++ Mods
* Added GameSettings.SaveSettingsOnNextGameSave method to flag the settings dirty and trigger a settings save next time the game is saved.
* New GameEvents.onPartActionUIShown when PAW window is shown.
* PartLoader no longer clears modules and resources from preconfigured parts.
* Contracts - contract definitions Nodes are now processed additively. Before subsequent Contract nodes encountered on loading would override each other.
* Experience Trait nodes can now handle multiple Effect nodes in different cfg files.

 

Edited June 2, 2019 by Yakuzi

[HebaruSan]

7 minutes ago, Yakuzi said:

associated report on the bugtracker 

Interesting. @5thHorseman is right, the old behavior was a bug, the new behavior is a fix.

[dok_377]

59 minutes ago, HebaruSan said:

Interesting. @5thHorseman is right, the old behavior was a bug, the new behavior is a fix.

Autostruts worked this way straight from 2016. If that's not how they meant to work, it took WAY too long to "fix" it. Even though it seems logical now (no magical struts through the air), we need a word from Squad on this one, just to be sure. 

Edited June 3, 2019 by dok_377

[Klapaucius]

50 minutes ago, HebaruSan said:

Interesting. @5thHorseman is right, the old behavior was a bug, the new behavior is a fix.

The new DLC parts are set up so they can mirror each other. There is a setting to override symmetry to reverse direction on a rotor, for example. I wonder if this is an unintended consequence of that.

 

[OHara]

46 minutes ago, HebaruSan said:

if 5thHorseman is right, the old behavior was a bug, the new behavior is a fix.

Well, he only said he would have thought it was a bug, for "autostrut:grandparent" to connect each symmetry-related part both to its grandparent and also across grandparents of the symmetry-related parts. 
That does make it seem plausible that the old behavior was an accident, but an accident we might want to keep.

The stock craft "Albatross" shows the effect very well, with its long wings made of short segments, presumably to demonstrate the flex. 
In version 1.3.1 the wings flexed nicely, looking like I would expect them to as a glider pilot. 
In version 1.6.0 someone at Squad added auto-struts-to-grandparent to the outer wing segments of that craft, making the whole wing extremely stiff.  It seems that autostruts constrain the rotation of the linked parts, so the strut to the symmetry-grandparent prevented the wingtips from rotating upwards at all, and beams don't bend nearly as far if the rotations of their ends are clamped.
In version 1.7.1 the cross-aircraft auto-struts disappear, and the wing is just a little stiffer than the original, as struts-to-grandparent would imply.

I think people who want realistic flex generally avoid auto-strut, so don't care how it works (except that we want to turn off the forced auto-strut of wheels and legs).

People trying to make sci-fi vessels or simulations with the limited menu of KSP parts just want a rigid vessel, so magical struts through the air are no problem.  If the old behavior of autostruts gets in the way of robotics, users can turn it off and use real struts that give more control.

[Klapaucius]

44 minutes ago, OHara said:

I think people who want realistic flex generally avoid auto-strut, so don't care how it works (except that we want to turn off the forced auto-strut of wheels and legs).

People trying to make sci-fi vessels or simulations with the limited menu of KSP parts just want a rigid vessel, so magical struts through the air are no problem.  If the old behavior of autostruts gets in the way of robotics, users can turn it off and use real struts that give more control.

From a gameplay standpoint (irregardless of what is happening in the background) I think of autostruts simply as adding stiffening to the aircraft without the complete inflexibility that comes from rigid attachment.  It is not, to me, a magic invisible strut even if is coded that way.   I don't really buy the "realistic flex" point of view, because we are working with set parts, and in the real world we'd be able to re-engineer those parts to adjust the flex.

Also, in KSP, you have to compensate for the fact that you cannot necessarily "weld" one part to another. For example, if I build a triangle out of I-beams, one of those corners will be loose unless I strut it somehow.  In the real world, we would not have to work around that. We have very few reinforcing tools at our disposal.

 

ON ANOTHER NOTE:

I've gone through and tested some of my other large planes and so far, no issues. I don't know if it is that particular plane or whether the reinstall did the trick.  But so far, so good.  

 

Thank you all for pursuing this along with me.

Edited June 3, 2019 by Klapaucius

[OHara]

7 hours ago, Klapaucius said:

Feel free to download and test it [Gogol on kerbalx].

Wow! 36 tonnes surface-attached on either side of a short I-beam.  That is how we used to try to make flapping wing craft.  Version 1.7.0 does cross the autostruts and the craft is stiff.

Even though I argued that we probably want the former behaviour, the reason for the change might be a stubborn reason, so auto-strutters might need augment with real struts.  They are heavy at 50kg, but since version 1.2 they have zero drag.

Moving just one pair of real struts so they cross the centreline makes Gogol stiff again.

[HebaruSan]

Don't cry because it's over. Smile because it happened.

[Klapaucius]

2 hours ago, OHara said:

Wow! 36 tonnes surface-attached on either side of a short I-beam.  That is how we used to try to make flapping wing craft.  Version 1.7.0 does cross the autostruts and the craft is stiff.

Even though I argued that we probably want the former behaviour, the reason for the change might be a stubborn reason, so auto-strutters might need augment with real struts.  They are heavy at 50kg, but since version 1.2 they have zero drag.

Moving just one pair of real struts so they cross the centreline makes Gogol stiff again.

It would be nice if there was some sort of weld function. While this is an out there aircraft, the fact is, if it were built in the real world that wing would be a single reinforced unit.  The only reason for the I-beam is you need to start with a base part, and wings won't cut it.

 

PS. How do you get the strut diagram?

Edited June 3, 2019 by Klapaucius

[OHara]

There was a mod to 'weld' parts, now being revived by users with some success.

Strut diagrams are under Alt-F12 (on Windows, otherwise Mod-F12) : Physics : Visualize Autostruts

[Azimech]

11 hours ago, Klapaucius said:

It would be nice if there was some sort of weld function. While this is an out there aircraft, the fact is, if it were built in the real world that wing would be a single reinforced unit.  The only reason for the I-beam is you need to start with a base part, and wings won't cut it.

 

PS. How do you get the strut diagram?

I think a weld function would make KSP into X-Plane.