Spacecraft Design Evolution in KSP

[septemberWaves]

This is a thread intended to prompt discussion about the odd design quirks that many players tend to independently develop because of the restrictions that KSP imposes.

Something I find quite interesting is how spacecraft evolve to suit certain requirements. Some players, such as myself, try to make our spacecraft look quite realistic, but this often ends up not being optimal for the game (most of the time adding extra parts to make a design look realistic makes it less efficient). Other players, however, design spacecraft to work well in the unique environment of KSP.

The latter has the interesting effect where players who do it have ended up mostly-independently evolving a very specific appearance for spacecraft. End-game spacecraft designed to suit KSP with little regard for realistic appearances tend to be huge nuclear-propelled motherships composed almost exclusively of Mk3 liquid fuel tanks. Such spacecraft almost always include a small lander crammed tightly into a cargo bay, which has a Mk1 lander can as the command pod and is fueled by an abundance of Oscar-B tanks surrounded by the orange toroidal tanks. Obviously there's a lot going on here, and what makes it particularly intriguing is the fact that so many players seem to make the same design choices separately.

Edited March 14, 2017 by eloquentJane

[DoctorDavinci]

I think this has to do with ones approach to the problem one is trying to solve.

Many want to replicate RL missions or do things with a realistic flair, and then there are those who like to go all out with their space exploration and create gigantic monstrosities ... I fall into this category for the most part

In my case form always follows function ... for example: 

 

 

[Fearless Son]

I am not content with perfect efficiency in my spacecraft, I feel the need for some kind of aesthetic appeal too.  They do not need to replicate real missions, or fictional ones for that matter, nor do I want something that is terribly inefficient in the context of the KSP simulation itself, but I do want craft with an interesting and plausible looking profile.  For my really long-range missions, I tend to have very minimal integrated fuel tanks, but feed them through much larger fuel tanks attached via docking ports.  Makes it easier to assemble and refuel in orbit, and I can drop the extra tanks en-route to save dry mass.  This necessitates clearance on certain sides of the ship, often little booms or nacelles to mount them on easily.  Multiple points of attachment on each tank are needed to ensure the tanks are aligned properly so we do not worry about unbalanced mass that might throw off the ship during maneuvers.  

[MaxwellsDemon]

The optimum design for me is a mix of form and function.  It's entirely possible to put together something that works great, but if I can't stand to look at it on the screen, it's no good for me.

I'm usually going for a particular look, based on an IRL vessel or unbuilt concept; seldom have I cut entirely loose from what's actually been done (or thought about).  I have a lot of respect for the designers who can come up with something completely bizarre-looking that still works-- but it's not my personal style.

At first, radial components tended to be stuck on all over, but I hated the look.  I'm much more apt to stick them in a service bay or hide them by offsetting them partly into or below the surface of the larger structure.

Over time, I've tended toward the modular approach, designing "the perfect capsule" or family of boosters for a particular tech level or stage of the game, and then reusing those building blocks in later designs.  It just feels right to me.  

YMMV...

[Curveball Anders]

35 minutes ago, MaxwellsDemon said:

The optimum design for me is a mix of form and function.  It's entirely possible to put together something that works great, but if I can't stand to look at it on the screen, it's no good for me.

Much the same for me.

Something that gets the job done, preferably in a cost efficient way, and looks like it at least might work.

[fourfa]

2 hours ago, eloquentJane said:

This is a thread intended to prompt discussion about the odd design quirks that many players tend to independently develop because of the restrictions that KSP imposes.

Something I find quite interesting is how spacecraft evolve to suit certain requirements. Some players, such as myself, try to make our spacecraft look quite realistic, but this often ends up not being optimal for the game (most of the time adding extra parts to make a design look realistic makes it less efficient). Other players, however, design spacecraft to work well in the unique environment of KSP.

The latter has the interesting effect where players who do it have ended up mostly-independently evolving a very specific appearance for spacecraft. End-game spacecraft designed to suit KSP with little regard for realistic appearances tend to be huge nuclear-propelled motherships composed almost exclusively of Mk3 liquid fuel tanks. Such spacecraft almost always include a small lander crammed tightly into a cargo bay, which has a Mk1 lander can as the command pod and is fueled by an abundance of Oscar-B tanks surrounded by the orange toroidal tanks. Obviously there's a lot going on here, and what makes it particularly intriguing is the fact that so many players seem to make the same design choices separately.

This is very interesting and well-stated.  There is definitely a lot of convergent evolution if you maximize on efficiency with stock parts. Is this the lander you speak of?

LOL.  The clipped stack of Oscar-B and Round-8s has more than double the fuel per volume density of regular tanks - some might consider this more than a little cheaty.  Not to mention the part count is kind of pointlessly high, and the HIGHLY complex collider meshes on all those Round-8 has to be really dragging frame rate down when in a complex scene with a big mothership.

One way I really like to add some spice to the monotony of efficient stock designs - fuel switching mods.  All of a sudden you can put oxidizer in radial RCS tanks, LFO in your NCS adapters, and pure LF in the Mk2 tapers.  No new parts, no gameplay changes - but really opens up the aesthetic space.

 

[septemberWaves]

Just now, fourfa said:

Is this the lander you speak of?

Landers that resemble that are exactly what I'm talking about. That command pod is popular because it's the least massive option (except for the external command seat but that then limits you to a single flag per landing). The tank setup is popular because it's the best way to fit a huge amount of delta-v into a Mk3 cargo bay. There's fairly obvious reasons why the design works, I just find it quite interesting that it seems to be a fairly unanimous decision that a setup like this is the best.

 

1 minute ago, fourfa said:

The clipped stack of Oscar-B and Round-8s has more than double the fuel per volume density of regular tanks - some might consider this more than a little cheaty.

It's not really cheaty, the toroidal tank is empty in the middle so there's no point refusing to fill that space.

[A_name]

I would say one of the most common staples of KSP design is the (over)use of side boosters, especially designs that incorporate side boosters attaching to other side boosters (Yo, I heard you like boosters…), something you’d never see in real life.

Other examples of design choices that would only show up in KSP:

Landers without legs. Who needs them when you have a perfectly fine engine to set down on? (For light landers anyway.)

Landers using aircraft landing gear instead of legs (for heavier designs).

Rockets with un-aerodynamic parts sticking out all over the place.

Rockets with huge fins on the bottom, a.k.a. I’ll be damned if my rocket flips again.

Crafts with way more RCS fuel than you’ll ever need.

Asparagus monstrosities.

Edited March 14, 2017 by A_name

[Tyko]

Oscar-Bs definitely create an opportunity for KSP designs to go far astray of "RL" designs. Their capacity is so much higher for a given size tank that I'm often tempted to use them everywhere. It doesn't make sense that a stack of Oscar-Bs will have more capacity than a single larger tank of the same overall dimensions.

Another case of KSP driving design involves reaction wheels vs thruster choices. In RL some sort of RCS is really required if your planning on a lot of maneuvers. In KSP you can build craft with just reaction wheels and never worry about them saturating.

[fourfa]

1 hour ago, eloquentJane said:

It's not really cheaty, the toroidal tank is empty in the middle so there's no point refusing to fill that space.

Perhaps the original sin is the fact that the Round-8 attach nodes are set so far inboard - a stack of them obviously clips together and requires magic quantum entangled fuel densification headcanon.  But space the nodes out so there's no visual clipping, and it's still denser than the regular tanks.  Linearizing that kind of thing (along with a lot of others... Mk2s, NCS tank, RCS mass density...) would be an interesting mod project and could change that 'default aesthetic' you're writing about.  But I recall a phase of 1.2 pre (IIRC) where RoverDude linearized the RCS values, and everyone exploded because it might break their existing ships.  I guess it's hard to change things.

Mind you, I am enthusiastic about clipping for aesthetic reasons alone.  But it's interesting to see how people have different thresholds for what they think is OK, and how it changes over time

Edited March 14, 2017 by fourfa
typo

[Leafbaron]

I think it stems from the fact that by nature, humans are problem solvers and that we like efficiency. Historically, agricultural development happened around the world almost all at once with no communication between civilizations. Which may not sound like a huge accomplishment. However, this allowed humans to have specializations, such as blacksmithing, or tailoring and allowed free time for people to think even more and invent. Imagine what life would look like without being able to go to the grocery and get food. I submit to you that it is no mere coincidence, just human nature, that if given enough time, we will always (more often than not) arrive at the best solution for a problem.   

[MaxwellsDemon]

On 3/14/2017 at 3:16 PM, A_name said:

Rockets with huge fins on the bottom, a.k.a. I’ll be damned if my rocket flips again.

Yeah... There's an available niche for a small fin that has a steerable area.

ETA:  Hm.  Wonder if I can make one with a small fin and one of the control surface pieces.   (making a note)

Edited March 16, 2017 by MaxwellsDemon

[Norcalplanner]

I guess I'm an oddball.  I've never really used the Toroidal-8, and I only use Oscar-B's on unmanned probes.  My standard-ish one-man lander uses a Rockomax 8 or Rockomax 16 as a wider base for greater stability.  Here's one I've used on Iota and Ceti in my GPP game:

I think that the selection of mods used, as well as whether the player prefers career, science, or sandbox also has a big influence on the direction of craft design.

For example, one of my workhorse craft designs is the Bluebird, which was born out of a desire for a cheap rocket taxi to bring Kerbals to and from LKO. Since I was playing with Stage Recovery and KCT at the time (back in 1.1.3 when KCT still had a parts inventory), I designed it in such a way that no parts were discarded - everything was either recovered with SR, or was brought back to the surface as part of the reentry vehicle.  Here's a look:

[Long Finger]

Personally I really hate the designs I see where things are attached in such a way that they could never realistically function. For example you can stick things over the doors of a service bay (like some disgraceful youtubers) and still open those doors as if the things weren't there!

I have had to come to terms with the idea of Kerbals living inside of a full fuel tank with a docking port stuck on it however, and I quite like the magic strutting feature because attaching struts nicely is sooo hard.

[Spacetraindriver]

My Design Process:

Step 1: Think up a mission idea.

Step 2: Search craft designed (or planned to anyways) for the real-life counterpart of my mission.

Step 3: Build!

[KerikBalm]

12 hours ago, eloquentJane said:

The tank setup is popular because it's the best way to fit a huge amount of delta-v into a Mk3 cargo bay. There's fairly obvious reasons why the design works, I just find it quite interesting that it seems to be a fairly unanimous decision that a setup like this is the best.

Who cares if the lander fits in a mk3 bay? I dont need the lander to be enclosed on an orbital mothership. External docking is fine. I've never done this sort of design. I use 1.25 or 2.5m tanks on my landers to keep part count down. 

[AeroGav]

From an aircraft point of view,  Big S wings and Strakes are the min-maxers choice, since they give "free" fuel capacity.  But they don't fit together as neatly as modular segments, being of an irregular shape,  and you end up overlapping and clipping stuff just to make it look less awkward.   BTW, none of the stock wings are big enough for my needs.

Standard design crutch #1

Back to back Big S wings is the starting point.   CoL too far aft? clip a strake on.    Awkward transition to the nose? More strakes to smooth the angle.   Oh noes, CoL too far forward.  Put another strake on the trailing edge.   OK but now our Wet CoM is too far forward ! Let's build a tailfin out of back to back strakes so we can get more fuel in the back too ! Perfect !  Oh wait, we have too much liquid fuel and too much lift.   Actually, there is no such thing as too much LF or too much lift, who cares.

Strakes for tail fins is for smaller craft.  Monstrosities just use more Big S wings - 

Canard designs are of course ubiquitous.   It's one crutch for the  aft CG and short lever arm for the tail controls.  But the vertical stabilizer still needs to go at the back - so hang them off the trailing edge..

Finally, we need talk about your weight.   Well, to be more precise, mass.     All the mass of the jet engine is concentrated in a ridiculously short part which  is just basically the nozzle.   Which you can put anywhere , like on the back of a crew cabin,  implying that the turbomachinery is in the bit with the windows , which must get noisy,  and the intakes can be anywhere else on the craft.

IRL military jet engines are , at minimum, the length of a shock cone, pre-cooler and nozzle, and the mass is in the front half, where the compressors and turbines live.  The back half is just afterburner duct, an empty tube.  

The extreme aft CG caused by Kerbal jet engines means Skylon-like wing mounted engines are the rule, everything shoved as far forward as i can visually bear.  This means attaching your wings to the outside of the nacelle, which looks wierd, or wing tip mounting them, perhaps on forward swept wings.    

 

 

[septemberWaves]

6 hours ago, AeroGav said:

From an aircraft point of view,  Big S wings and Strakes are the min-maxers choice, since they give "free" fuel capacity.  But they don't fit together as neatly as modular segments, being of an irregular shape,  and you end up overlapping and clipping stuff just to make it look less awkward.

The obvious alternative is making biplane-like designs where you have two layers of the same wing shape because one wasn't lifty enough (I've gotten into that habit recently and I have noticed other people doing it as well).

[Rune]

Mmmm... it's been ages since I built a lander. Why do so, if I can use a kerbin SSTO that launches itself, and has huge tankage space by definition? Plus, I get to recover and/or reuse everything, so it's the most efficient way, fund-wise. Besides, with a proper fuel depot network, a Kerbin-capable SSTO can get itself almost anywhere, so there's hardly a need for a mothership. Maybe some help for the really difficult ejections, like Moho, but that's about it.

You must watch a lot of challenges, where the design priority is usually mass, not cost, and it's usually done in a single launch.

 

Rune. And even then, I would use a SSTA, because I hate throwing stuff away.

 

[netbumbler]

This is a good topic and one I've thought about a lot.  My philosophy right now is to go for fairly realistic ships in that I use TAC life support and always include about 4 or 5 tons of living space per Kerbal.  I basically look at the size of different crewed modules (I also use Planetary Base Systems) and incorporate what I think looks like enough living space for hard-core astronaut types that are totally motivated for their mission.  So of course that is way less space than a wimp like me would be able to deal with on a 5 or 6 year round trip to Jool, but probably realistic for hardened veterans of space flight. 

Though my ships are very functional and I can pretty much do any mission (though maybe not as efficiently as the real wunderkinds on here) I have yet to get a good feel for aesthetics.  That's my next big challenge is to 'have it all' as it were: realistic functionality combined with nice looks.  Plenty of food, living space and making the ships not too 'Kerbal' (by that I mean using multiple launches and docking realistic-sized sections instead of just making one 10,000 ton ship) and of course having it look nice and maybe even innovative. 

The other facet of this philosophy, however, is that part of my idea of 'realism' is to deal with the actual game limitations as I would any real life situation.  This isn't a perfectly realistic space/flight simulation so sometimes you have to do stuff that is a little unlikely to happen or be designed that way in real life.  It took me a long time to come to terms with that fact that I can't always solve problems like NASA or ESA would because of little quirks of the game, but in real life you deal with the actual situation confronting you and not always holding to some abstract ideal.  That may sound a bit overly philosophical, but when I do design something a little on the 'Kerbal' side versus a more realistic interpretation, I don't worry about it as much as I used to because being effective in the game is slightly more important to me than being 'realistic.' 

My first Eve Mission ship in 1.05 (one launch and NO refueling of any kind):

 

 

IIRC there is only a MK 1 Lander can and an ion powered Mercury capsule for return...not a lot of living space for a multi-year mission and of course no life support mods.  Actually I'm pretty sure this was totally stock except for Kerbal Engineer and maybe one or two other minor mods. 

 

And my last Eve mission in 1.2.2:

 

Separate launch for the transfer vehicle and the lander and a refueling mission to top them off in Kerbin orbit before departure.  And though there were escape pods built into the design for safety, when the ship returned to Kerbin orbit I actually sent up a crew return vehicle to retrieve the Kerbonauts instead of using the escape pods or some built in reentry vehicle. 

 

And the final assembled vehicle (a 'mere' 1300 tons):

 

Oh and I forgot to add that I build every crewed launch vehicle with a built-in escape system (this is probably my specialty in this game...no matter how crazy the ship is I always have a way to save the crew...except for my big shuttle launches...but NASA didn't really have a decent escape system for the real shuttle either).  And I go for perfectly stable flying vehicles throughout the mission profile: no wobbly ships and good, clean separations of stages.  I guess the one thing I don't do is worry too much about debris: can't recall how many times I've destroyed that poor launch pad...

Edited March 16, 2017 by netbumbler

[nascarlaser1]

I have only a few things I ever care about in my ships.

1: They keep the Kerbal alive. This is a must. Whether or not I'm playing with LS installed (which I do a lot) the Kerbals must NOT DIE. This can range from having elaborate and overpowered escape pods on my stations that will never be used, or having a simple yet effective ejector seat on my test shuttles that will always be used .

2: I build everything for purpose. I really don't care how stuff looks. This lets me have almost total freedom building, since it doesn't matter what it looks like so long as it gets my guys to where they are going alive and well.

3: Must be prepared for everything. This one ties in with number 1. All my ships can have the cockpit and any extra crew modules eject from the main shuttle and land on Kerbin, the Mun and Minimus independent of the rest of the shuttle (I don't have the ability to leave that area yet, so no worry about crash landing on Duna ), and include up to  300 days of supplies for each Kerbal (LS thing) for my Kerbals to survive on the planet till rescue arrives.

3 Continued: If I happen to need to eject the crew into space, they also have RCS included in order to help rendezvous with the rescue ship, and let the Kerbals EVA out and over to the rescue ship (I play sandbox so I don't need to worry about tourists). Finally, in the event that my ship somehow gets stuck somewhere that I don't have the experience to rescue from (like orbit around the sun) then..... well... actually, I haven't figured out a contingency plan for that one yet.... eep! back to the drawing board!!

Edited March 17, 2017 by nascarlaser1

[suicidejunkie]

54 minutes ago, nascarlaser1 said:

I have only a few things I ever care about in my ships

You may be interested in the idea of deploying rest stations in key areas:

Solar powered, self-cleaning restrooms with a sterilizing UV light.  Includes a drill and slightly modded low efficiency convertotron to slowly restock the greenhouse between visits.  Upgrade from Stayputnik to Hecs + docking port for skycrane based deployments.

 

[nascarlaser1]

6 minutes ago, suicidejunkie said:

You may be interested in the idea of deploying rest stations in key areas:

Solar powered, self-cleaning restrooms with a sterilizing UV light.  Includes a drill and slightly modded low efficiency convertotron to slowly restock the greenhouse between visits.  Upgrade from Stayputnik to Hecs + docking port for skycrane based deployments.

 

wait... a converatron can refill greenhouses?? YAAYYYY!!!

[NSEP]

I always make my rockets not unnecessarily bulky. With that, i mean that i wont use the 2,5m decoupler because it looks really cartoonish. Its like having a cool and realistic looking painting with a childish house drawing in the backround, it just pops out and does not fit. I use fairing-interstages instead and-or stack decouplers.

When it comes to Spacecraft/Probes, i try to make it simple and quick, and actually look like a real spacecraft at the same time. Not like a cylinder with a satelite dish and a solar panel.

I always put a battery and a reaction wheel system in my crafts too. The sad thing is: in my opinion they look slightly off style too, so i have to regulate those things too with my design.

[A_name]

On 3/14/2017 at 2:35 PM, MaxwellsDemon said:

Yeah... There's an available niche for a small fin that has a steerable area.

ETA:  Hm.  Wonder if I can make one with a small fin and one of the control surface pieces.   (making a note)

Or just go for the elevon 1's