1.11 Part Mass Bug?

[tehmattguy]

Windows, Stock + Breaking Ground, fresh install. I suspect this also affects non-DLC games. 

I've been observing some strange behavior related to low-mass parts in version 1.11. Certain (low-mass, physics enabled) parts seem much heavier than usual, despite no changes in mass or CoM being shown in the VAB.  Crafts that include these parts suffer from a significant increase in weight and a loss of delta-V.  Robotic parts sag as if burdened with extra weight. Some of my crafts from 1.10 have become unplayable as of this release. 

So far I've tested the effect using grip pads, empty dumpling and baguette tanks, and robotic parts with various physicsless parts attached.  Grip pads, small robotic parts, and empty dumpling tanks are affected whereas empty baguette tanks  are not. I think this is because they're below some mass threshold and are not physicsless parts. Small physicsless parts are not affected on their own but do become affected if attached to robotic parts.

Below are some screenshots of the tests I've performed. 

Rocket delta-V test:

Spoiler

 

Experiments done with a simple test rocket with various parts attached. The rocket is Alt-F12'd to orbit and performs a maneuver based on its stated delta-V. Meant to show discrepancies in listed mass and dV. 

Standard test rocket with no other parts attached. 1032m/s stated, 0.7m/s lost (insignificant / no loss).

Two grip pads attached. 1029m/s stated, 34.6m/s lost.

Empty baguette tanks. 850m/s stated, 0.0m/s lost.

Empty dumpling tanks. 868m/s stated, 14.7m/s lost.

Alligator hinges + a bunch of flags. 863m/s stated, 58.7m/s lost.

These tests were also performed in 1.10 with insignificant/ no delta-V loss in all cases.

 

Scale test:

Spoiler

 

Made a simple scale for a little extra flair. Seemingly affected parts are weighed against unaffected parts to find discrepancies in part mass. Tested in both 1.11 and 1.10.

5 grip pads vs 2 communotrons. Both sides should weigh 0.01t.

1.10: Perfectly balanced as all things should be 

1.11: Grip pads are much heavier. Scale shifts pretty quickly after loading.

2 empty dumpling tanks vs 1 empty baguette tank. 0.028t vs 0.034t.

1.10: Baguette is heavier, and the scale agrees.

1.11: Dumplings heavier for some reason.

Micronode vs Alligator hinge with communotrons attached. Both sides should weigh 0.15t.

1.10: Fairly balanced, though the scale favors the alligator hinge.

1.11: Alligator hinge definitely heavier.

 

Edit:

I did more testing and it seems like the game treats any physics-enabled part under 0.03t to weigh minimum of 0.03t.

The effect is pretty consistent. Parts like empty oscar-B tanks, empty dumpling tanks, small grip pads, and small robotic parts all weigh exactly 0.03t according to the scale.

I plugged these new mass numbers into a delta-v calculator for my test rocket. The delta-v numbers I got were incredibly accurate taking the 0.03t per part into consideration.

More screenshots below:

Spoiler

 

Another round of experiments to hopefully clarify the 0.03t issue.

Using my scale again, a grip pad is shown to balance against 6 communotrons, weighing 0.03t. All physics-enabled parts listed under 0.03t also show this effect. 

The effect also applies to physics-less parts, but only if they're directly attached to a robotic part.

The scale balances out when adding 0.03t per physics-less part.

Parts like fuel tanks that fall below 0.03t by draining will stop losing mass as soon as they hit 0.03t. An empty dumpling tank has the same 0.03t mass as a partially fueled tank. 

Going back to my initial delta-v test, the discrepancies can be explained if we account for the 0.03t added by each affected part. Using an external delta-v calculator gives us 995 m/s, a difference of 34m/s from 1029m/s as shown in-game.

With this setup the game estimates our delta-v to be 868.0m/s. However in this situation each individual communotron, being attached directly to a robotic part, will add 0.03t to the rocket. External delta-v calculations put us at 828.1m/s, a difference of 39.9m/s as shown in-game. 

 

 

1.11.1 Update:

As of this update partRBMassMin has been reduced from 0.03 to 0.002, which apparently gives parts a new minimum mass of 0.002t. This seems to have corrected the mass values for every physics-enabled part listed at or over 0.002t.

So from what I can tell most of the part mass issues introduced in version 1.11 have been resolved.

However, any physics-enabled part that weighs less than 0.002t is still affected, and will have an effective mass of 0.002t. The only stock part I know of that fits this description is the Tiny Nosecone, which should weigh 0.001t yet measures as 0.002t on the scale and in delta-v tests. 

Tiny Nosecone (0.001t) balances two Cubic Octagonal Struts (0.001t each)

While testing for physics-less parts I found that attaching one directly to a robotic part gives it a minimum mass of 0.0065t. Interestingly this mass isn't shown in the editor but is accounted for in-flight through delta-v readouts. I also found this effect to be present in version 1.10.1, so I don't know whether it's intentional or not. 

A Communotron 16, Cubic Octagonal Strut, and an RFP-0 Flag (0.0065t total) balances against a single Cubic Octagonal Strut when it's attached directly to a robotic part. 

 

Bug report: https://bugs.kerbalspaceprogram.com/issues/26898

Edited January 30, 2021 by tehmattguy
Updated for 1.11.1 patch

[AlamoVampire]

I saw a video on YT about 1.11 having new mass related tweaks. I think what you are seeing is related to this. As an example of what I saw in the video they built a scale of sorts (similar to yours) with 2 identical space plane crew cabins, with 1 having 0 kerbals and the other having 4. They unlocked the hinge or pivot they used to free the scale and the side with 4 kerbals dropped. They transfered the crew over and then that side moved. So, I think its intended behavior and not a bug. Ive yet to play 1.11 myself as Im awaiting the mod makers updates to the mods I use, so I can only go off that video. I wanna say it may have been a video by Matt Lowne but I am NOT 100% sure.

172712192020

[tehmattguy]

1 hour ago, AlamoVampire said:

They transfered the crew over and then that side moved. So, I think its intended behavior and not a bug.

Oh yeah, I'm familiar with this part of the update but my issue doesn't have anything to do with crew mass (probably). 

What I'm seeing is that specific parts or part combinations act much heavier as of 1.11. For example, the small grip pads are listed at 0.002t in the editor, but in flight they act as if they're actually 0.03t. Another oddity is that physicsless parts seem to gain mass if they're attached to a robotic part. This change in mass isn't listed anywhere, nor accounted for in the game's delta-v calculations. This leads me to believe the effect is unintentional. 

Edited December 20, 2020 by tehmattguy

[AlamoVampire]

@tehmattguy I was right about it being Matt Lownes video. I want to still say its likely intended behavior based off his video. Why do I say this? At about 2 minutes and 5-10-ish seconds in he mentions removing and or swapping eva gear or not wearing any backpacks to save mass. I would likely imagine parts having different effects on mass as your op describes to be a logical extension. As to dV and the like i cannot know for sure as again I can only go off his video as I await mods catching up. 
231412192020

[tehmattguy]

@AlamoVampire No doubt they've changed how part mass is handled but the problem is just that the effect can be pretty extreme and is not recorded anywhere in-game. If my scale experiment is to believed, grip pads now act ~15x heavier than they used to be.  Some of my rockets, which use a lot of grip pads and robotic parts, are losing hundreds of m/s delta-v just from this effect.  My Duna lander in particular loses 600m/s by the time it reaches orbit. This is in spite of the game telling me that the rocket's mass and delta-v is exactly the same as before. That's the whole reason I bring up this issue anyways, and why I think it's unintentional. 

[Snark]

[EDIT] Never mind what I said here, I missed seeing that the OP already did exactly this.   Redundant suggestion moved to spoiler so as not to clutter people's eyeballs unnecessarily.

Spoiler

You know what would be a really interesting repro case for demonstrating the bug very conclusively?

Build a balance.

We actually have the tools for it!

1. Make a craft that is anchored to the launch pad (e.g. with launch clamps), that has a rotor with horizontal axis that's unmotorized and has the damping set to zero.
2. Make a couple of long, really lightweight arms that project from either side of the rotor hub (cubic octagonal struts would be good for this).
3. At the end of each arm, put another unmotorized, free-pivoting rotor, with a lightweight arm dangling straight down from it.
4. Everything up to this point is symmetric, so if you were to launch this, it should just sit there on the pad with nothing moving.
5. Now the test.  On one arm, attach some bug-free parts whose mass adds up to some known quantity, like 0.05 tons or whatever.
6. On the other arm, attach a bunch of the bugged parts, such that their mass is supposed to add up to the same number.

If you're right, and the bugged parts are actually behaving as though they have significantly higher mass, the the idea is that they'll be heavier and pull that end of the scale down.

I wonder if this would work?  Has anyone tried it?

 

Edited December 20, 2020 by Snark
oops

[OHara]

It looks like it affects mostly small parts, and the discrepancy can be in either direction.

The small 2-kg grip pad balances the 10kg battery
The 20-kg spider balances 10kg battery

So far the affected parts all have the new of ModuleCargoPart in their cfg files. The *.cfg files show the same mass as the game UI.  It looks like something is causing the physics engine to see a different mass than shown in the configuration and in the UI.

It doesn't seem to need any DLC parts (although they do build a really nice balance).  A pair of parts tipping teeter-totter (a.k.a. see-saw) the wrong way shows the problem.  The simplest I can show is three 20-kg Spiders balancing an 80-kg Twitch.

Do you have a login to enter bug reports? Do you have a super-simple example to help narrow the problem? [Edit: never mind.  I just saw your amended top post.  I'll confirm the report with my simplest example.]

6 minutes ago, Snark said:

I wonder if this would work?  Has anyone tried it?

The OP did so; see the top post.

Edited December 20, 2020 by OHara
saw the bug report linked in the top post

[Snark]

3 minutes ago, OHara said:

The OP did so; see the top post.

Whoops-- missed seeing that one additional spoiler.  Never mind. 

[AlamoVampire]

@tehmattguy i see where youre pointing. Just guessing that perhaps the behavior is linked somehow through some aspect of crew mass and the various backpacks and the link is being a baby kraken causing the odd behavior? Just a guess coming from pure speculation ofc.

075912202020

Edited December 20, 2020 by AlamoVampire

[Corona688]

Kudos for the high level of documentation you put into this.  All we need now is the craft file so we can test your scale ourselves.

The one I see on the bug tracker is not your scale, but someone else's stock one.

Edited December 22, 2020 by Corona688

[Darkherring]

I have also encountered strange mass issue in 1.11.

The mass in VAB and in flight look absolutely different. In VAB, the reading is consistent with KER, but in flight not anymore

VAB:

 

Flight:

 

This is data from the same stage in VAB and flight. It's an unmanned probe

[tehmattguy]

On 12/22/2020 at 10:03 AM, Corona688 said:

Kudos for the high level of documentation you put into this.  All we need now is the craft file so we can test your scale ourselves.

The one I see on the bug tracker is not your scale, but someone else's stock one.

Thanks! Also, the craft file for the scale has just been added to the bug tracker page. 

[Stamp20]

I think this bug is caused by a new addition in Physics.cfg.

"partRBMassMin = 0.03 // Minimum mass that a parts RigidBody can have - If this is too small then PhysX will not behave when it is dropped as a vessel, this is the default if minimumRBMass is not defined in the part cfg."

 

[OHara]

On 12/23/2020 at 9:06 PM, Stamp20 said:

a new addition in Physics.cfg. "partRBMassMin = 0.03"

That observation suggests a workaround, decreasing partRBMassMin to the lowest mass of any 'physics-significant' part (the ones whose placement shifts the CoM).

That workaround clears the problem for me, but it might have side effects, so I posted it as a Module-Manager patch at the end of the bug report.

[Sovetskysoyuz]

On 12/27/2020 at 2:01 PM, OHara said:

That observation suggests a workaround, decreasing partRBMassMin to the lowest mass of any 'physics-significant' part (the ones whose placement shifts the CoM).

That workaround clears the problem for me, but it might have side effects, so I posted it as a Module-Manager patch at the end of the bug report.

I tried the workaround ModuleManager patch that you posted, but it didn't seem to have an effect in-game for me. Directly modifying Physics.cfg did the job though.

[OHara]

17 hours ago, Sovetskysoyuz said:

it didn't seem to have an effect in-game for me. Directly modifying Physics.cfg did the job though.

Thanks.  I've changed the notes on the bug-reports.

I didn't notice that, because I was using Module Manager's alt-F11 hotkey to load a patch without restarting KSP, which did have effect for some reason.

[Arrowstar]

Hey everyone, just wanted to add that this is definitely a bug that also occurs with the kOS CPU parts.  Hopefully we get a fix soon...

[Stone Blue]

On 12/31/2020 at 11:10 PM, Arrowstar said:

Hey everyone, just wanted to add that this is definitely a bug that also occurs with the kOS CPU parts.  Hopefully we get a fix soon...

Yup.. i expect VERY MANY modded parts will be affected...
First have to get it all figured out/working in pure stock, first, so devs know *exactly* what they have to fix/change/update.

[jlcarneiro]

On 12/23/2020 at 1:09 PM, Darkherring said:

[...]

[...]

Thanks for your effort!

On another note, what are those two buttons near the communication button? Those with what seems to be a pulse sign...

[Darkherring]

23 hours ago, jlcarneiro said:

Thanks for your effort!

On another note, what are those two buttons near the communication button? Those with what seems to be a pulse sign...

It's from Kerbalism

[dprostock]

Hello!
I don't know if it's going to be a mass problem, but today I conducted a team test on Muna, and I found that some kerbals can use the RCS backpack and others can't. 
I couldn't understand the problem, but reading the thread...

Spoiler

 

[Davi SDF]

On 1/21/2021 at 5:30 PM, dprostock said:

Hello!
I don't know if it's going to be a mass problem, but today I conducted a team test on Muna, and I found that some kerbals can use the RCS backpack and others can't. 
I couldn't understand the problem, but reading the thread...

  Hide contents

 

Omg, we found weak kerbals!

[dprostock]

1 hour ago, Davi SDF said:

Omg, we found weak kerbals!

I would say that the Kerbals are not to blame, is the erroneous application of the gravitational constant with a development of backpack loaded with gasified water in the last update.
Anyone would say to you, "Things don't slip on the surface anymore!!!"

[Arrowstar]

Anyone know if this bug is resolved or mitigated in KSP 1.11.1?  I didn't see anything in the release notes but I also don't have a chance to fire up KSP right now.

[m4ti140]

Danny is going to have his way with this one... I expect ships with mass effect cores.