Magic Propulsion! - A Study of KSP Collision and Reaction Wheel Physics

[Helmetman]

Thanks for sharing the actualities towards this 'loophole'

I like to be informed about the internal faculties so I can better categorize bogus propulsion drives from propulsion drives.

One things remains however.........

And that is that this is fun!!!!!
And I never had fun in my life whereby others defined me as being the cheater when I had fun with whatever I had fun with at the time until present.

Don't start being the first guy categorizing myself as a cheater because I had fun with experimenting a loophole that you stereotypically pre categorized as cheating.
Not saying this to anyone particular. But I am saying this to the pre categorizers of supposed cheaters when doing stuff like this. 

Ksp made, faked or failed at it the way it is. Be happy we have a atmospheric hocus pucus drive rather then to debunk it because it's scientifically mweeeeeeeeehh....
As MajorJim said, this isn't Martin Schweicher's Orbiter. Like it better? Go there now or install any mod that takes swift work of this shenanigans.
 

Edited October 2, 2017 by Helmetman

[regex]

24 minutes ago, Majorjim! said:

Why do you even play KSP?

It's a pretty good modding platform, also kind of fun by itself.

Quote

The entire concept is unrealistic.. Perhaps you would be happier playing orbiter or something of the like.

I agreed with you, this isn't an "exploit" at all, it derives from a failed point of realism. Whether that's good or bad isn't really part of my statement so I'm not sure why you're attacking my playing this game. I see no reason why that little point of failed realism can't be modded out, and I frequently do so because I find it to be pretty insane, but at this point in the development cycle it's also insane to think that's going to change in the base game, or be worried that it will change.

Edited October 2, 2017 by regex

[Shadowmage]

22 hours ago, Geschosskopf said:

Then there's the rover wheels.  Ever notice how the EC they draw falls off the faster they go?  That leads to the exact same huge input in power out vs. power in that stock electric props have.

Yeah, the stock electric motor modeling for the wheels is... well.. mostly non-existent.  They -do- decrease their torque output at higher RPM (alongside the ec input), but it is nothing like a real electric motor efficiency curve (as they fall to zero EC @ zero torque @ max RPM; a real engine would still have a current draw in that range).

22 hours ago, Geschosskopf said:

I guess the underlying issue is that the 1 EC/sec = 1 kW ~ 1 hp standard is good for non-moving parts (batteries and the like) but the EC requirements for moving parts (rover wheels, reaction wheels, drills, etc.) is very low compared to doing the same thing by burning LFO.

No kidding.  I did a several day study of the stock EC use/density for various purposes, trying to determine what exactly 1EC meant (for KSPWheel).  Many mods would posit it to be 1EC = 1KJ (1EC/s = 1kw); but the stock rover wheels draw far too little power. 1/65th of the power they should if 1ec = 1kj.  In the end, I had to settle for a 'magic' value of 1ec = 65kj for KSPWheel, simply so the parts would still be usable from a power-draw perspective in an otherwise stock balanced system.

KSPWheel motor efficiency curve  (red = current draw, dkblue = torque output, green = power output (hp), ltblue = efficiency; abstract engine, plotted as % input/output/rpm):

 

I didn't even look at reaction wheels; though not surprising that they are a bit off when considering input/output power.  I know that solar panels, APUs/fuel cells, and of course motorized wheels are all completely out-of-whack.  Some more than others (by orders of magnitude).

[EpicSpaceTroll139]

 

Update: a further experiment has shown reaction wheel torque to indeed be constant, therefore the efficiency should increase linearly with speed. I believe the apparent torque curve was due to stretching of the spider engines from their joints.

I believe something interesting I could do is find out the moment of inertia of various parts, which could be useful for players who create various mechanical contraptions. Might be especially useful for that clock project I put on hold.

 

Edit: Computed MOI about roll axis for a couple of parts:

Flat Octo probe core: 40.59 kg*m^2

Small Reaction Wheel: 45.07 kg*m^2

Edited October 4, 2017 by EpicSpaceTroll139

[RedPandaz]

Is there any way to use this physics exploit for a space-capable engine?

[EpicSpaceTroll139]

1 minute ago, RedPandaz said:

Is there any way to use this physics exploit for a space-capable engine?

 

As far as I'm aware, no. There is no way to convert rotation produced by the reaction wheels into thrust in absence of some kind of exhaust. You could spin a wheel and decouple masses from the rim, but I'm pretty sure that you would always get a better exhaust velocity and thus efficiency from a regular rocket engine.

[RedPandaz]

Just now, EpicSpaceTroll139 said:

 

As far as I'm aware, no. There is no way to convert rotation produced by the reaction wheels into thrust in absence of some kind of exhaust. You could spin a wheel and decouple masses from the rim, but I'm pretty sure that you would always get a better exhaust velocity and thus efficiency from a regular rocket engine.

 

aww

3 minutes ago, RedPandaz said:

 

aww

Are there any space-capable exploit engines that you know of?

 

[RoadRunnerAerospace]

1 hour ago, RedPandaz said:

 

aww

Are there any space-capable exploit engines that you know of?

 

Kraken drive.

[Zeiss Ikon]

There's a "drive" that pumps ore (or other mass, but ore has the highest density of any transferrable resource) back and forth between rotating tanks.  Scott Manley demonstrated such a "reactionless drive", using KOS to control the transfers,  and was able to actually lift off the surface of Kerbin with it.  There's a short series of videos on his channel showing his initial demonstration of the effect, then various iterations of development of the drive.  I'm not sure if this is the same as the "Kraken drive", but as far as I know, it isn't any more prone to attract Kraken attacks than any other design that takes advantage of the lack of collision between parts of the same vessel.

[Geschosskopf]

3 hours ago, Zeiss Ikon said:

 I'm not sure if this is the same as the "Kraken drive", but as far as I know, it isn't any more prone to attract Kraken attacks than any other design that takes advantage of the lack of collision between parts of the same vessel.

The best and most reliable "Kraken" drive was based on lander legs.  It was kinda the same as the Ladder Drive in that it relied on the violation of Newton's Laws caused by the anti-clipping mechanics.  However, instead of needing to constantly have a Kerbal on a ladder head-butt some fixed object, it used the suspension system of the lander legs against a fixed object.  The fixed object was positioned slightly closer to the legs' point of attachment than the full stretch of the suspension.  The usual arrangement was a central 1.25m tank with 4-6 legs radially attached, with a 1.25m - 2.5m adapter plate on the bottom of the stack.

Once you got in space (it didn't work very well in the atmosphere) with the legs folded, you deployed the legs.  Until recently, when you deployed lander legs, they always went to full stretch of the suspension to begin with.  This caused them to clip into the adapter plate, which caused the legs to move "up" in relation to the plate, thus dragging the central core and the adapter plate with them despite Newton's 3rd Law.  When the legs moved, even though their relative to position to the rest of the craft hadn't changed, the cycle repeated.  Full stretch, recoil, repeat.  Thus, you got a rapid series ofreactionless forces applied to the whole thing and it accelerated.  Problem was, the legs didn't all recoil at the same instant so the force was asymmetric and the craft kinda spiraled around in a semi-random direction.  But it quickly build up very high speed so you could maybe go somewhere intended with it, with a bit of luck and finagling.  To stop the drive, you just retracted the legs, turned the ship as desired, then lowered the legs again.

I don't think this works anymore, though.  The same thing that made this type of drive work also caused landers and bases sitting on legs to jump off the ground on physics load, so a recent update altered leg suspension behavior to stop "base jumping".  That probably nixed this short of "Kraken" drive, although I haven't bothered to test this.

The ladder drive still works, however, and this allows precise control.  It's useful (and arguably realistic) to have Kerbal-powered vehicles (bicycles, kayaks, etc.).  Even small Kerbal-powered airplanes.  I even made a challenge for this, which lead to @Chemp circumnavigating Kerbin both by air and bicycle.  However, there's nothing to stop you from also using this to go to other planets.  Once you get pointed, upwards, you can escape the atmosphere and reach interplanetary velocities.  You can even reach orbit from the surface of Eve no problem.  All because Newton's 3rd Law is being ignored.

To make this work, you need a Kerbal, a short section of ladder only slightly longer than a Kerbal, and some fixed objects on both ends of the ladder for the Kerbal to collide with while climbing in either direction (so you have a reverse gear/thrust reverser/retro rocket).  You also need a small probe core (for directional control in air or space, plus maneuver nodes) and something to power it, and you need a command chair if you plan on going to other planets (so you can save and warp while coasting).  For a ground vehicle, you of course need wheels (use something with a high speed tolerance because you can REALLY get moving).  For a plane or SSTO, you need wings.  Control surfaces aren't necessary if you've got the small reaction wheel.

So, put the Kerbal on the ladder and climb him into the obstruction in front.  Keep holding down W and the vehicle will start to move (if it isn't too heavy).  If it'has wings, it will eventually take off.  The Kerbal and the craft are separate entities so you have to use ] and [ to switch back and forth between throttle (the Kerbal) and steering (the probe core).  Then just go wherever you want.  Of course, you can't need life support or be bothered about space radiation.  And I see now there's a mod called "Wwwwww" or something that basically holds down the W key for you, so you no longer have to get carpal tunnel from that.

All in all, the ladder drive's only practical and realistic use is for small Kerbal-movers.  Sure, you can get 1 to a few Kerbals anywhere in the solar system as well, but the payload is pretty low and of course you don't care about realism or physics.  Still, it's fun once in a while to "climb a stairway to heaven", just for larfs.

 

[RedPandaz]

1 hour ago, Geschosskopf said:

 

I don't think this works anymore, though.  The same thing that made this type of drive work also caused landers and bases sitting on legs to jump off the ground on physics load, so a recent update altered leg suspension behavior to stop "base jumping".  That probably nixed this short of "Kraken" drive, although I haven't bothered to test this.

 

1

My bases still jump tho

Edited October 20, 2017 by RedPandaz

[Geschosskopf]

19 minutes ago, RedPandaz said:

My bases still jump tho

The lander legs wasn't / isn't the only cause of base-jumping.  They were just the main culprit.  There's still the problem of clipping into the ground on physics load, and the same anti-clipping thing hates that and moves the base so it won't clip.  This cans cause jumps, especially if you make your base all 1 big assembly of connected modules, whether by docking or KAS pipes.

The underlying cause is that except for the Flats of Minmus, there is no smooth terrain anywhere in KSP, even at KSC itself, and especially if you have high terrain detail.  There are always little lumps, bumps, and minor slopes over short distances.  The base is draped over these terrain irregularities and all its joints bend slightly to conform with them due to gravity.  Trouble is, each time you load physics on the base, it gets placed in a slightly different position on the ground than it was last time you looked at it, no doubt due to some floating point issue.  However, all the base parts remember how they were in relation to each other, so now the base doesn't fit the ground, or it does but its bottom is slightly below the surface this time.  Either way, parts of it clip into the ground and get rejected.  Once clear of the ground, the parts then conform to the terrain that's now under them.  The more pieces you have connected over a wider area of ground, the more ground clipping happens, so the more force is involved and the more severe the jumps.  It's even worse if you use one of the mods that allows anchoring parts to the ground---the REALLY exacerbates this sort of thing.

So the only real way to avoid base jumping is to 1) set all the modules directly on the ground, no legs or wheels, 2) don't anchor anything to the ground, and 3) do NOT connect the modules together in 1 big thing but use a mod that alllows "wireless" resource transfer.between disconnected parts provided they're close together.  When you do this, each time you load physics on the base, the individual modules will shift and rotate very slightly in relation to each other, so over time you'll gradually lose any neat alignment you started with, but modules rarely, if ever, jump, so the base will last a very long time  The only issue is aesthetic---the base will always look like a shanty town and this will get worse as time passes.  However, when you see how much the individual modules move, you'll realize how much "phantom force" would have built up in your base had it all been 1 big thing, until it finally jumped or parts of it simply exploded for no apparent reason.

Edited October 20, 2017 by Geschosskopf

[RedPandaz]

Just now, Geschosskopf said:

The lander legs wasn't / isn't the only cause of base-jumping.  They were just the main culprit.  There's still the problem of clipping into the ground on physics load, and the same anti-clipping thing hates that and moves the base so it won't clip.  This cans cause jumps, especially if you make your base all 1 big assembly of connected modules, whether by docking or KAS pipes.

The underlying cause is that except for the Flats of Minmus, there is no smooth terrain anywhere in KSP, even at KSC itself, and especially if you have high terrain detail.  There are always little lumps, bumps, and minor slopes over short distances.  The base is draped over these terrain irregularities and all its joints bend slightly to conform with them due to gravity.  Trouble is, each time you load physics on the base, it gets placed in a slightly different position on the ground that it was last time you looked at it, no doubt due to some floating point issue.  However, all the base parts remember how they were in relation to each other, so now the base doesn't fit the ground, or it does but its bottom is slightly below the surface this time.  Either way, parts of it clip into the ground and get rejected.  Once clear of the ground, the parts then conform to the terrain that's now under them.  The more pieces you have connected over a wider area of ground, the more ground clipping happens, so the more force is involved and the more severe the jumps.  It's even worse if you use one of the mods that allows anchoring parts to the ground---the REALLY exacerbates this sort of thing.

So the only real way to avoid base jumping is to 1) set all the modules directly on the ground, no legs or wheels, 2) don't anchor anything to the ground, and 3) do NOT connect the modules together in 1 big thing but use a mod that alllows "wireless" resource transfer.between disconnected parts provided they're close together.  When you do this, each time you load physics on the base, the individual modules will shift and rotate very slightly in relation to each other, so over time you'll gradually lose any neat alignment you started with, but modules rarely, if ever, jump, so the base will last a very long time  The only issue is aesthetic---the base will always look like a shanty town and this will get worse as time passes.  However, when you see how much the individual modules move, you'll realize how much "phantom force" would have built up in your base had it all been 1 big thing, until it finally jumped or parts of it simply exploded for no apparent reason.

My base is on the minmus flats, but uses the very large rover wheels

[Geschosskopf]

Just now, RedPandaz said:

My base is on the minmus flats, but uses the very large rover wheels

It's a bad idea to mount bases on anything with suspensions, wheels or legs.  That's just asking for trouble.  You can maybe lock the suspension but that doesn't always help, at least over the long run.  Wheels and legs are the same thing as far as Unity is concerned, only wheels are worse because the only difference is they can slide along the ground instead of trying to sit there (any tire rotation is just animation, basically an illusion).  Both have been rather problematic ever since KSP went to Unity 5 or whatever a couple years ago.  I mean, even today, if a Kerbal touches a stationary wheel, he still goes ragdoll, so you know there's something flaky about them.

[RedPandaz]

12 hours ago, Geschosskopf said:

It's a bad idea to mount bases on anything with suspensions, wheels or legs.  That's just asking for trouble.  You can maybe lock the suspension but that doesn't always help, at least over the long run.  Wheels and legs are the same thing as far as Unity is concerned, only wheels are worse because the only difference is they can slide along the ground instead of trying to sit there (any tire rotation is just animation, basically an illusion).  Both have been rather problematic ever since KSP went to Unity 5 or whatever a couple years ago.  I mean, even today, if a Kerbal touches a stationary wheel, he still goes ragdoll, so you know there's something flaky about them.

So you're suggesting using a seperate rover to move base parts?

I think I'm gonna just make space stations around Mun and Minmus

[Geschosskopf]

31 minutes ago, RedPandaz said:

So you're suggesting using a seperate rover to move base parts?

I think I'm gonna just make space stations around Mun and Minmus

The way to make a base as Kraken-resistant as possible is to make it out of unattached modules that share resources "wirelessly", and these modules should just sit on the ground themselves, no wheels or legs.  There are several mods these days that allow "wireless" resource transfers over short distances, and they also allow scrapping or destroying parts.  So if you want to position the modules in some semblance of order (at least to start with---they won't stay there), land them with wheels and drive them into formation.  The scrap the wheels so they sit on the ground.

Stations aren't substitutes for bases because stations only consume, they do not mine, resources.  So if you have a station, you still need a base to keep the station supplied.

[RedPandaz]

On 10/20/2017 at 11:27 AM, Geschosskopf said:

The way to make a base as Kraken-resistant as possible is to make it out of unattached modules that share resources "wirelessly", and these modules should just sit on the ground themselves, no wheels or legs.  There are several mods these days that allow "wireless" resource transfers over short distances, and they also allow scrapping or destroying parts.  So if you want to position the modules in some semblance of order (at least to start with---they won't stay there), land them with wheels and drive them into formation.  The scrap the wheels so they sit on the ground.

Stations aren't substitutes for bases because stations only consume, they do not mine, resources.  So if you have a station, you still need a base to keep the station supplied.

Also I am having the problem that I am horrible at orbital rendezvous

[Geschosskopf]

10 hours ago, RedPandaz said:

Also I am having the problem that I am horrible at orbital rendezvous

That's something easily fixed, but best discussed in a different thread.

[DAL59]

On 9/29/2017 at 4:43 PM, Geschosskopf said:

they hit the arbitrary max set by the game to keep itself from crashing.  That's why stock props based on free-spinning reaction wheels are exploits--the game simply lacks the math to keep reaction wheels from spinning so fast.

That's probably a good thing...