Something doesn't seem right here....

[Levelord]

I've been doing some aero testing for the difference between radial attchments, parts joined by attachment points, part offset of attachment points and their effects on aerodynamics. I've seemed to have stumbled on something unexpected.

What I have here are 2 identical crafts where one has its boosters centrally attached to an adapter with attachment nodes, and the other where the boosters are also attached to attachment nodes, but have been radially displaced by the offest tools.

Initial predictions would be that the one with the centrally placed stack to travel further, faster... Or at the very least both rockets will travel the same. However, by a strange stroke of oddity, the one with the displaced stack flew consistently higher and faster with each repeated test.

I have no explanation for this, or why the game would decide that displaced fuselages attached to nodes would make a more aerodynamic craft.... If anyone can shed some light on this, I would very much like to know. I will attach the craft I have in this post for others to test here: https://dl.dropboxusercontent.com/u/16683090/Untitled%20Space%20Craft.craft

EDIT: Mystery solved on page 3

Edited May 23, 2015 by Levelord

[GoSlash27]

Levelord,

Could the difference be in which rocket you're focusing on? Debris follows different aero rules than focused craft.

Best,

-Slashy

[WinkAllKerb'']

node hierarchy, root order, unity and calculation may be could get an impact ... dunno (same craft root reordered may be worth a try)

[Levelord]

Levelord,

Could the difference be in which rocket you're focusing on? Debris follows different aero rules than focused craft.

Best,

-Slashy

I've actually considered this by switching crafts at launch, but the results were the same...

node hierarchy' date=' root order, unity and calculation may be could get an impact ... dunno (same craft root reordered may be worth a try)[/quote']

Probably... I have no idea though... It's really weird...

However, someone might find this a useful feature to take advantage of, now if only I can think of a use for this....

[Xannari Ferrows]

Objects that are not connected to the root part register as different crafts, therefore have their own TWRs and aerodynamic properties. However, because these parts are still linked together [courtesy of the offset tool] as a craft rather than multiple ships, I suspect the game is handling this by averaging each "ship's" aerodynamic coefficients, as opposed to the one whole part. It's weird, I know, but it's the only thing I can think of.

[WinkAllKerb'']

yup or what about some kind of drafting simulation related weirdness at some point also ? Because offset craft has low chance not to broke in game // in reality due to node attachement

Edited May 23, 2015 by WinkAllKerb''

[GoSlash27]

I've actually considered this by switching crafts at launch, but the results were the same..

Verry interesting... *scratches chin*

I wonder what would happen if the parts were offset just one pixel from their original position. In theory, couldn't you make a low drag version of a craft that's virtually indistinguishable from the original?

Best,

-Slashy

Edited May 23, 2015 by GoSlash27

[WinkAllKerb'']

Verry interesting... *scratches chin*

I wonder what would happen if the parts were offset just one pixel from their original position. In theory, couldn't you make a low drag version of a craft that's virtually indistinguishable from the original?

Best,

-Slashy

was thinking the same with different offset %

[SpaceChipmunk]

Ok this may be a strange question. Were both ships in the same position each launch? Or have you tried with the offset one on the left and the non-offset one on the right? Wonder if it could maybe be something to do with unusually modelled effects from the planets spin or the angle of the launchpad???

[Levelord]

Verry interesting... *scratches chin*

I wonder what would happen if the parts were offset just one pixel from their original position. In theory, couldn't you make a low drag version of a craft that's virtually indistinguishable from the original?

Best,

-Slashy

An interesting thought would be to offset an object out and then offset it back in again, you could in theory, make the craft have less drag for the same look.... *shrug*...

Needs moar testing..... TO THE VAB!

[Xannari Ferrows]

Isn't this just vabulous?

[Red Iron Crown]

That's a curious result indeed. I see you have MJ installed, is there a chance that "Limit overheats" is checked? That will throttle engines even without autopilot engaged, and it looks like the narrower design would overheat more quickly due to engine proximity.

[Levelord]

That's a curious result indeed. I see you have MJ installed, is there a chance that "Limit overheats" is checked? That will throttle engines even without autopilot engaged, and it looks like the narrower design would overheat more quickly due to engine proximity.

Negative. MJ was not used in these tests. I also actively switch crafts to check if they are at full throttle, to which they indeed are.

The craft is available for you to test and see for yourself too

[Red Iron Crown]

Not at my KSP machine for another few hours, a shame since this is an interesting observation that merits more testing. What happens to performance if you offset the stacks inward? I.e. clip the tanks and engines together somewhat or entirely?

[sal_vager]

I tried to test this contraption of yours but it turns out many parts are attached by their base node on one rocket, and their top node on the other, so I'll rebuild one rocket and test that with the boosters in-line and offset.

[Deddly]

Is it possible that the craft with the engines offset is more stable and so uses less thrust vectoring? Try it without SAS, maybe?

[Levelord]

Not at my KSP machine for another few hours, a shame since this is an interesting observation that merits more testing. What happens to performance if you offset the stacks inward? I.e. clip the tanks and engines together somewhat or entirely?

Sure thing. The right hand rocket is the one with inwards displaced fuselages.

Same result though....

[Deddly]

You have a strut connecting the two rockets at the top. Please could you post the results without the strut?

[GusTurbo]

This is a very interesting bug. I wonder what's going on here.

[WinkAllKerb'']

You have a strut connecting the two rockets at the top. Please could you post the results without the strut?

?

Edited May 23, 2015 by WinkAllKerb''

[sal_vager]

I think the bug is in how KSP handles drag on strangely built craft, as comparing the same craft with only parts offset between launches shows no appreciable difference in performance.

As you can see here with these altitude tests with a reproduction of Levelords rocket, minus any inverted construction.

Javascript is disabled. View full album

Just in case it's re-routing...

Javascript is disabled. View full album

Nope.

Edited May 23, 2015 by sal_vager

[Rhomphaia]

Tried separate launches of a similar craft alternating whether or not the Tanks were offset. offset or not they performed the same.

If you take your craft and offset both, one still outperforms the other. Rebuild it with neither offset and again, different performance between craft. Switch which one is the offset one and now the non offset one pulls ahead.

Must be to do with the separation rather than the offset I would imagine that a quicksave made after they had separated would show a different tree structure for each craft.

My guess would be that the one that gets the focus after separation gets the probe core as its root, while the other gets the leg (not sure if engines can be root, but the tank can) this may result in different drag calculations across the quadcoupler?

Edited May 23, 2015 by Rhomphaia

[Levelord]

MYSTERY SOLVED

You have a strut connecting the two rockets at the top. Please could you post the results without the strut?

I just had the same thought too. I made a new testing craft with mirrored rockets called the Strut Tester. You will be interested in these results...

This mirrored configuration has the basic rockets supported by launch clamps instead of struts to maintain stiffness.

During launch:

As expected, they both break the sound barrier and have identical performance.

We have mirrored vessels with a SINGLE strut connector. The parent craft being the left rocket.

Launch state:

The differences are incredible. The parent craft holding the strut connector is significantly slower than the right craft.

We now have a configuration where the LEFT rocket is a parent of one strut connector, while the RIGHT rocket is also a parent of a strut connector. 2 strut connectors in total.

Launch state:

Remarkable. With both rockets as parents, the rockets perform identically.

We've now determined that the strut connectors are causing the differences in the tests, but how big of a difference do struts matter on crafts? We compare the strut with placing a Advanced Inline Stabilizer on the right rocket to weigh it down. The left rocket is a parent to the strut connector. The Advanced Inline Stabilizer weighs 0.1 tonnes. It is twice the weight of a strut connector which is 0.05 tonnes. The right rocket on a whole weighs 0.05 tonnes more.

Launch state:

As observed, the left rocket carrying the strut has an initial gain in speed and height.

The striking and most surprising issue occurs when the heavier rocket suddenly gain speed and outperforms the supposedly lighter rocket and zooms to a higher apoapsis.

Conclusion:

We have a EAS-4 Strut Connector that weighs 0.05 tonnes which are usually strapped to the outside of rockets to make them stable.

Compared to adding a Advanced Inline Stabilizer which weighs twice as much in an inline configuration at 0.1 tonnes, strutted rockets experience a significant drag disadvantage compared to unstrutted rockets carrying heavier parts.

My findings suggests that excessive use of struts in your rockets will actually slow them down. Struts actually impose a significant aerodynamic drag on the craft. I will be removing all struts from my aircraft and redesigning them to require less to see how they impact SSTO performance.

Test craft used is provided here: https://dl.dropboxusercontent.com/u/16683090/KSP/Strut%20tester.craft

Edited May 23, 2015 by Levelord

[WinkAllKerb'']

wut and clap clap ; )

[JellyCubes]

Lesson learned from this thread: When experimenting with the drag model, make sure the ships you are testing are actually identical.