Scientific mistakes in KSP

[Mortimer Kerman]

Hi, I have an idea.

Can you all report here the scientific mistakes made in KSP (like the SOI system or radiations) to make a non exhaustive list?

(Sorry for mistakes, i'm french.)

Edited December 29, 2019 by Trolol29

[MR L A]

As much as I love KSP... this list you want us to write could be as endless as real world science is itself. It's only a simulation of basic orbital mechanics at the end of the day :) But if you want major "mistakes" (I'd argue most are for gameplay reasons, not genuine mistakes) then we could start with;
 

- Unrealistic aerodynamic simulation

- Unrealistic 2-body gravity physics

- Sun is the wrong colour/temperature for its size/mass

- Kerbin is stupidly dense for its size 

- Kerbal fuel is not very dense at all

- No radiation

- Kerbin atmosphere just suddenly stops (not realistic at all)

- Kraken drives

- No real need to worry about satellites/stations overheating without radiators as they do in real life

The list could go on forever... especially if you start saying things like "KSP is inaccurate because it doesn't simulate neutrinos" or "I can't smash electrons together" or even as simple as "I can can crash a 1000t vehicle into the ground at supersonic speeds but there is no crater.. not even a scratch". They all sound pretty inaccurate scientifically which makes them "mistakes", I suppose.

[Mortimer Kerman]

5 minutes ago, MR L A said:

Autant j'aime KSP ... cette liste que vous voulez que nous écrivions pourrait être aussi interminable que la science du monde réel elle-même. Ce n'est qu'une simulation de la mécanique orbitale de base à la fin de la journée Mais si vous voulez des "erreurs" majeures (je dirais que la plupart sont pour des raisons de gameplay, pas de véritables erreurs) alors nous pourrions commencer;
 

- Simulation aérodynamique irréaliste

- Physique gravitationnelle irréaliste à 2 corps

- Le soleil n'est pas la bonne couleur / température pour sa taille / masse

- Kerbin est stupidement dense pour sa taille 

- Le kerbal n'est pas très dense du tout

- Pas de rayonnement

- L'atmosphère de Kerbin s'arrête soudainement (pas du tout réaliste)

- Entraînements Kraken

- Pas vraiment besoin de s'inquiéter de la surchauffe des satellites / stations sans radiateurs comme ils le font dans la vraie vie

La liste pourrait continuer indéfiniment ... surtout si vous commencez à dire des choses comme "KSP est inexact car il ne simule pas les neutrinos" ou "je ne peux pas briser les électrons ensemble" ou même aussi simple que "je peux planter un 1000t véhicule dans le sol à des vitesses supersoniques mais il n'y a pas de cratère .. pas même une égratignure ". Ils semblent tous assez inexacts sur le plan scientifique, ce qui en fait des "erreurs", je suppose.

Basically, I wanted to talk about a list of details that squad neglected to avoid turning the game into a headache simulator.

[Athen]

I don't think it's that realistic that Laythe has liquid water on its surface that far away from the sun.

Then there are a bunch of problems with engines & fuel:

1. Real engines have a limited amount of start-ups.
2. Cryogenic propellants boil off.
3. Not all engines can use the same fuel.
4. We don't have to worry about the fuel floating around in the tanks rather than entering the engine, and thus don't need ullage motors.

Edited December 28, 2019 by Athen

[MR L A]

3 minutes ago, Trolol29 said:

turning the game into a headache simulator.

aye, that it would be  Also you'd need one hell of a computer to simulate it all!

9 minutes ago, Athen said:

I don't think it's that realistic that Laythe has liquid water on its surface that far away from the sun.

 

Not sure how this works in regards to Kerbol's temperature etc. but I think Jool is actually closer to its star than Earth is to the sun... if I'm looking at the numbers properly  

KSP has such a dinky solar system lol

[Athen]

14 minutes ago, MR L A said:

...but I think Jool is actually closer to its star than Earth is to the sun... if I'm looking at the numbers properly  

It is closer, but because the sun is smaller, it probably emits less heat, which would put the optimal spot for life somewhere closer to the sun. Idk if squad actually did the math to put kerbin and jool in accurate spots, but judging by the fact that  the next planet over is literally an icy waste land, we can assume that it's pretty cold in the Jool system.

Edited December 28, 2019 by Athen

[MR L A]

1 minute ago, Athen said:

It is closer, but because the sun is smaller, it also emits heat at a smaller range. Idk if squad actually did the math to put kerbin and jool in accurate spots, but judging by the fact that  the next planet over is literally an icy waste land, we can assume that it's pretty cold in the Jool system.

If Squad didn't, somebody in the forums somewhere definitely did  I saw a post about it a long long time ago  Yeah, not somewhere I'd want to go without quite a large coat.

Thinking about it, does Laythe have water or is it some other liquid? Explodium seas or something I've seen in the forums? Lower freezing point? 

But I agree, in reality, I think virtually every substance would be pretty darn frozen on laythe

[Mortimer Kerman]

22 minutes ago, MR L A said:

aye, that it would be  Also you'd need one hell of a computer to simulate it all!

Not sure how this works in regards to Kerbol's temperature etc. but I think Jool is actually closer to its star than Earth is to the sun... if I'm looking at the numbers properly  

KSP has such a dinky solar system lol

Yes, Eeloo is a bit closer to Kerbol than Earth to Sun.

[MR L A]

22 minutes ago, Trolol29 said:

Yes, Eeloo is a bit closer to Kerbol than Earth to Sun.

Really??? Wow, I didn't realise that one  thanks!

[Guest]

1 hour ago, MR L A said:

aye, that it would be  Also you'd need one hell of a computer to simulate it all!

Not true. People play KSP with real-sized solar system, n-body physics and realistic engine plumes. It does require a beefy computer, but nothing exceptional (and some of the problems are because KSP just isn't too well optimized). They also don't make the game all that much more of a chore. The biggest thing is probably RSS, and that's because large planets make launches take 10 minutes or so (mostly flown in real time). Limited ignitions and ullage are surprisingly simple to work around.

Of the less obvious things, solar panel color is wrong. IRL, they can be blue, but for space-rated panels, black is more usual (some Russian ones are blue, but it's nothing like KSP's blue). The devs made them look like an artist's impression of a solar panel and not the real thing.

[MR L A]

6 hours ago, Dragon01 said:

Not true. People play KSP with real-sized solar system, n-body physics and realistic engine plumes. It does require a beefy computer, but nothing exceptional (and some of the problems are because KSP just isn't too well optimized). They also don't make the game all that much more of a chore. 

we were talking about simulating EVERYTHING... as in everything I said like particle smashing, realistic environmental destruction, PLUS all the things I didn't mention that KSP doesn't do...

Think you jumped in on that one without reading what we were saying properly  

[kerbiloid]

The science is ok. The universe differs.

[KerikBalm]

17 hours ago, Trolol29 said:

(Sorry for mistakes, I'm french (How to say "Non-exhaustif"?).)

As an English speaker living in Genève, I must say that there are many, many things that I do not like about the French language... however one thing I do like is how similar many words are between French and English.

The translation is "non-exhaustive"   https://www.dictionary.com/browse/non-exhaustive

Anyway, most of the major mistakes come down to the small scale. Assuming high density doesn't fix it. Kerbin's escape velocity is too low to hold on to an atmosphere for instance.

If we assume that 1 kerbal meter = 10 human meters, then 800 kerbal Isp would be 8000 human Isp, which is ridiculous.

2 body physics isn't a big problem in most cases (if a 3rd body is tiny in comparison), but the on-rails and SOI system really becomes inaccurate around Jool.

Also, the system itself doesn't seem like it could form naturally.

[Mortimer Kerman]

6 hours ago, KerikBalm said:

En tant que anglophone vivant à Genève, je dois dire qu'il y a beaucoup, beaucoup de choses que je n'aime pas dans la langue française ... mais une chose que j'aime est la similitude des mots entre le français et l'anglais.

La traduction est "non exhaustive" https://www.dictionary.com/browse/non-exhaustive  

Thanks

[KerbolExplorer]

The Jool system would colapse

I don't remember what would happen but as far as i can remember some moons would fly out other would crash or have their orbits altered

[Zeiss Ikon]

On 12/28/2019 at 10:17 AM, Athen said:

Then there are a bunch of problems with engines & fuel:

1. Real engines have a limited amount of start-ups.
2. Cryogenic propellants boil off.
3. Not all engines can use the same fuel.
4. We don't have to worry about the fuel floating around in the tanks rather than entering the engine, and thus don't need ullage motors.

 

On 12/28/2019 at 12:03 PM, Dragon01 said:

Not true. People play KSP with real-sized solar system, n-body physics and realistic engine plumes. It does require a beefy computer, but nothing exceptional (and some of the problems are because KSP just isn't too well optimized). They also don't make the game all that much more of a chore. The biggest thing is probably RSS, and that's because large planets make launches take 10 minutes or so (mostly flown in real time). Limited ignitions and ullage are surprisingly simple to work around.

Yep, we have Real Solar System and Realism Overhaul (which latter includes RealFuels and RealPlume to make the engines more realistic), and some of us use Principia to handle n-body physics and allow things like chaotic orbits, perturbation, and so forth.  And no, it doesn't really need a particularly "beefy" computer.  I have one that's quite a bit less performance than a top line gaming machine (AMD FX8350 with 4.1 GHz turbo, nVidia GTX750, some kind of SSD, and 16 GB RAM) and it actually performs better with all the mods for 1.6.1 RSS/RO/Principia than it does in stock 1.8.1.  I have to presume this is because the mods I'm using are more efficient code than the stock game.

I just came from launching a rocket with a WAC Corporal engine (pressure fed aniline/furfuryl fuel, red fuming nitric acid oxidizer) boosted with a V-2 analog (RD-100 burning 75% ethanol/water with liquid oxygen and high test peroxide to drive the pumps).  The booster has 70 second rated burn time and a single ignition; the upper stage 47 seconds rated burn and single ignition (plus I had to manage ullage, and spin up the rocket at the same time, so the upper stage would ignite and wouldn't tumble at 32+ km staging altitude).

The main difference is that KSP is a video game -- Realism Overhaul is a simulation.  Landing on the Mun in KSP is a challenge.  Landing (softly) on the Moon in RO is, well, a "moonshot" level project, or at least a simulation of one.

[Guest]

RSS is also a game, just a difficult one. With the full suite, it approaches Orbiter in physics accuracy, but it's still not quite there, in many areas. The very fact that rockets can be assembled lego-like is a pretty serious break from reality. Another example is complete disregard for just how difficult is it to reliably orient and locate yourself in space. Falcon BMS is a simulation (and quite an addictive one), FlightGear is a simulation, Reentry is a simulation, KSP with RSS/RO is a game. To make the latter into something deserving to be called a sim, it'd need to, well, simulate an awful lot of things. For example:
1. Thermal control. Insolation will heat your spacecraft on one side only, and your avionics will generate heat as well. Propellant tanks need to be kept at a specific temperature. KSP heat system is a sorry joke, and would have to be replaced to implement realistic radiators.
2. Attitude determination. A complex topic, Apollo had an IMU (3D gyrocompass) and two 3-axis gyros. Other spacecraft used star trackers (Apollo had a telescope for that). And that's just attitude. Good luck figuring out your velocity vector without ground control tracking you on radar (though having your own radar is also an option). Satnav can provide that, but you need to launch that first. Values you get also have errors, something completely absent from KSP.
3. For that matter, ground control had a major role in every space mission. All you can do without them is maintain orbit or break it (spacecraft typically have some emergency deorbit routine in case of comms failure).
4. Power. Spacecraft need continuous power for life support, instruments, computers, experiments... Typically, they have a primary DC system with AC inverters connected to it, for thing which need that kind of power. All this is split between several electrical buses. KSP's power system is rather laughable in comparison.
5. Tanks. Propellant tanks need to be pressurized (even for non-pressure fed engines), which RO doesn't simulate. KSP is also not capable of simulating multiple tanks on one part. Tank pressure is also important for crossfeed, as it determines flow direction (unless you have a dedicated pump).
6. Transients. Real RCS thrusters have minimum length of a single firing, and the thrust profile in such case is nontrivial. Real rocket engines don't instantly turn on and off like KSP ones (unless they run on ClF3, which, for all its problems, gave a very smooth ignition). Everyone knows that SRBs have thrust curves, the truth is, liquid engines have them, too. 
7. Cockpits. The real Apollo had half a million switches for a reason. You can hide that behind a touchscreen, or get away with cramming them on one of those neat command and signal tables the Russians use (and even then, those for more complex spacecraft are gigantic), but there's a lot of things that the crew can control, and usually has to at some point. 

Now, Orbiter doesn't have most of these things, either (well, not most spacecraft made for it, anyway), but it is only a physics sim, not an engineering one. If you want to simulate all the perils of spacecraft engineering, KSP doesn't even come close. I'm hoping for KSP2 to make some of that possible (particularly tanks, power and the thermal system).

[radonek]

• No light lag. Yes, Kerbol system is smaller, but still.
• Kerbals can handle pretty much any G-forces I throw at them, regardless of direction. Sometimes with a smile. They can also pop out to space in soft pressure suits without any prebreathing.
• Real rocket engines are not so easy to throttle as in KSP. For instance, Apollo lander descent engine can only by throttled up to 55%, above that it can only run at full blast due to thermal issues. Also, ablatively cooled engines tend to change their properties as they burn. Some booster engines can't be  throttled at all, while other are kinda regulated by changing fuel mixture. And some solids can be throttled down by pressure venting. Fine-grained and perfectly linear throttle of KSP engines is engineering dream. Also, shutting down an engine is not trivial matter (unless you want it blown to pieces) and adds to transients mentioned @Dragon01 . And all of this is probably trivial compared to issues of nuclear propulsion.
• In KSP it is quite possible to land on an engine bell. In real life, you really, really do not want this to happen.
• KSP gyros never saturates.
• No alignment is needed for directional antenas. Not a big thing for simple probes, but major issue for relays.
• No comms blackout during aerobreaking/reentry.
• Real life RTG power sources decay.
• KSP models perfectly spherical gravity field. Real planets are not ideal spheres of uniform density.
• No space weather whatsoever. No solar storms, radiation belts, light pressure etc.
• ISRU
• I harbor doubts about longterm stability of Gilly orbit. 
• Real atmospheric effects are not only way more complicated then in KSP, but sometimes even more complicated then  real rocket science, as witnessed by Skylab.
• KSP does math in finite precision (I think collision physics is only done in floats and celestial mechanics in double, but I may be wrong) so even correct math produce inaccurate results.
• Timewarp essentialy zeroes any rotation (and probably introduces a lot of other issues).
• Due to physics simulation in discreet time quantums, objects can actually pass through each other instead of colliding in some situations.
• Whole "physics bubble" thing.

Just for the record, I understand reasons behind most of these and I actually think Squad did a pretty good job designing the game.

 

[severedsolo]

11 minutes ago, radonek said:

No comms blackout during aerobreaking/reentry.

This is in the game if you turn it on. It was added in 1.2.

[DunaManiac]

1. The reaction wheels are far too powerful than in real life.

2. Engines fire up instantly, while in reality, engines take time to spool up before launch.

3. Command pods are too small, unless children are manning the spacecraft (I guess that would explain Jeb's manic smile)

4. The way probes are controlled, in the sense that it's like somebody with a joystick is at Mission Control maneuvering the spacecraft into orbit. kOS is more realistic in this manner.

5. Size/Density of the planets and solar system.

6. Engines are way more efficient than in real life.

etc.

Edited January 1, 2020 by DunaManiac

[Zeiss Ikon]

1 hour ago, radonek said:

• No light lag. Yes, Kerbol system is smaller, but still.
• Kerbals can handle pretty much any G-forces I throw at them, regardless of direction. Sometimes with a smile. They can also pop out to space in soft pressure suits without any prebreathing.
• Real rocket engines are not so easy to throttle as in KSP. For instance, Apollo lander descent engine can only by throttled up to 55%, above that it can only run at full blast due to thermal issues. Also, ablatively cooled engines tend to change their properties as they burn. Some booster engines can't be  throttled at all, while other are kinda regulated by changing fuel mixture. And some solids can be throttled down by pressure venting. Fine-grained and perfectly linear throttle of KSP engines is engineering dream. Also, shutting down an engine is not trivial matter (unless you want it blown to pieces) and adds to transients mentioned @Dragon01 . And all of this is probably trivial compared to issues of nuclear propulsion.
• In KSP it is quite possible to land on an engine bell. In real life, you really, really do not want this to happen.
• KSP gyros never saturates.
• No alignment is needed for directional antenas. Not a big thing for simple probes, but major issue for relays.
• No comms blackout during aerobreaking/reentry.
• Real life RTG power sources decay.
• KSP models perfectly spherical gravity field. Real planets are not ideal spheres of uniform density.
• No space weather whatsoever. No solar storms, radiation belts, light pressure etc.
• ISRU
• I harbor doubts about longterm stability of Gilly orbit. 
• Real atmospheric effects are not only way more complicated then in KSP, but sometimes even more complicated then  real rocket science, as witnessed by Skylab.
• KSP does math in finite precision (I think collision physics is only done in floats and celestial mechanics in double, but I may be wrong) so even correct math produce inaccurate results.
• Timewarp essentialy zeroes any rotation (and probably introduces a lot of other issues).
• Due to physics simulation in discreet time quantums, objects can actually pass through each other instead of colliding in some situations.
• Whole "physics bubble" thing.

Just for the record, I understand reasons behind most of these and I actually think Squad did a pretty good job designing the game.

 

Signal lag is an option with RemoteTech mod.  G force limits are an option in the stock game (that defaults "on" in Moderate and Hard modes) -- for parts as well as crew.  RO covers unthrottleable engines with limited ignitions via RealFuels and parts configs.  Reentry blackout, as noted in another reply, is an option in the stock game (only makes a difference if "coms required for control" is active).  Principia "fixes" gravity, to the limits of what a PC can process in real time, at least (and includes mods to fix issues with the stock solar system -- like making one of Jool's moons retrograde so they won't eject one another).  There are mods for decaying orbits, even beyond RSS extending the Earth's atmosphere to 140 km.  Persistent Rotation mod carries angular momentum through a time warp.  Most of the rest is necessary to make the game one that can run in a single-threaded program on a real-world Windows/Mac/Linux box.

I agree, Squad did a lot of stuff right -- and top of the list was making the game open for modifications.  Out of the box, it's great fun and will keep a new player busy and learning for weeks or months.  Add mods and it can satisfy someone who's been learning about space science for fifty years -- and still run on a computer that old guy can afford.

[DunaManiac]

24 minutes ago, VoidSquid said:

1. A revelation. Nobody ever mentioned this.

2. Real life: see 1.

3. Kerbals are humans, just in disguise.

4. I never heard about about any drones or other vessels being controlled remotely. Never.

5. Absolutely, how could they miss that the planetary system in KSP isn't exactly like our solar system?

6. See 5. 

 

It's not that we have a "model universe", adapted for people to play, have fun? It's not that there are several mods making the game more "realistic"?

I was only listing scientific discrepencies, nowhere did I say that I wished this would be in game. I like the KSP system the way it is, (although I do enjoy playing in RSS/RO sometimes). Anyone can admit that there are flaws in KSP. This isn't a thread about realism features to be added in game.

[VoidSquid]

24 minutes ago, DunaManiac said:

This isn't a thread about realism features

Seems I'm lost here then, what is this thread about?

[paul23]

My biggest gripe, apart from what mods like far/kopernicus fix, is the fact that ksp considers kerbal a perfect sphere. Being an oblated sphere adds a lot of fun to the mix, from having to do station keeping, so actually using it to change argument of periapsis "for free".

[kerbiloid]

Has somebody already mentioned the icy Minmus at same distance from the sun as the liquid water Kerbin?

Edited January 2, 2020 by kerbiloid