Discussion: what are the best ways to reduce launch cost?

[blakemw]

Tall rockets: autostrut. All problems gone.

You're probably the one lucky person who finds a plane ascent profile easier than a gravity turn. I mean I'm very good at both now, but I got good at gravity turns much much earlier in my KSP playing than plane ascent profiles and it's easier to understand what makes a Rocket work (i.e. fly properly) than what makes a Plane work. Even a bad gravity turn will normally spend no more than 100m/s extra than an ideal gravity turn and as a player who has played A LOT of hard career I learned to value steerable tail fins to keep on course - this allows doing an aggressive gravity turn and if the rocket is tipping over too fast just use the fins to pitch up: you lose a bit to extra drag but only 50-100m/s compared with having the perfect gravity turn. These days I'm also confident using spaceplanes in hard career though it took probably 10x the playing time to understand how to make a spaceplane that will work every time without testing and I'm still much less confident about how much deltaV I'll have once in orbit - for an untested rocket I know within 100m/s, for an untested spaceplane my margin for error would be at least 500m/s since small adjustments to the trajectory make a large difference in performance (this assumes trying to maximize efficiency, rather than maximize predictability). My main point here is that in my experience gravity turns are more reliable with fewer variables to optimize for - altough there is certainly overlap between rockets with fins and spaceplanes with as much thrust as rockets (i.e. part of what makes a plane reliable is aerodynamic control surfaces, and part of what makes a rocket reliable is high thrust freeing up a reliance on lift, so rockets with control surfaces and spaceplanes with high thrust are both pretty reliable)

A Triple-Boar is a piece of cake to make recoverable, add two chutes per booster at spread angle of 10 and deploy them in advance - literally everything else is optional. It probably wouldn't land back at the KSC without quickloads (altough the glide distance is not to be underestimated with a probe core and tail fins for pitch control) but it probably could land in the 90% recovery rate zone. There's a trick I use with Triple-boars in particular: the Triple-Boar forms a "wall", it'll always fall over in one of two directions (or one of one direction if you bias the parachute placement), then I place small wings or landing gear to catch it when it tips ensuring no damage for a land landing. If it lands in water I do ninja-recovery though even if if tips you'll still definitely recover the boars which are the expensive part. Now the recovery isn't as good for a spaceplane but the whole thing can be done at 4x speed and best of all it if RUDs no big loss (getting the ~40k back is nice but in career it's still merely equivalent to a piddly contract reward).

Edited February 6, 2018 by blakemw

[Guest]

40 minutes ago, blakemw said:

the whole thing can be done at 4x speed

So can a plane. I fly mine at 4x most of the time.

40 minutes ago, blakemw said:

the one lucky person who finds a plane ascent profile easier than a gravity turn

Funny, I was thinking that you might be the one lucky person who finds gravity turns easier. Strange how we generalise from our personal experience, isn't it?

41 minutes ago, blakemw said:

to make a spaceplane that will work every time without testing

If you figure that out, tell me too. Developing a plane is hard work and it takes a lot of testing, and I would never even attempt it in an ironman career. The good news is, once it's developed it's infinitely reusable, and it's very easy to develop variants if you keep the basic layout the same. 

...

I could go on, but basically I get the impression that you don't particularly enjoy designing or tuning designs, and certainly don't enjoy tuning spaceplanes. You just want to get stuff in orbit with the minimum of effort. Of course making big fat rockets with big fat fairings is the way to go with that. Nobody's disputing that!

But that's not the topic under discussion. The topic is "what are the best ways to reduce launch cost?" And big fat rockets are not as good in this respect as spaceplanes: they burn more fuel per ton of payload even if they're recoverable, and they are extremely difficult to land on the launchpad or runway reliably for 100% recovery. 

So I was taking issue with some specific statements about spaceplanes here that, from my own experience, are just plain incorrect: in particular your assertion that they scale poorly (to mass and volume).

This just is not the case: I lift 300-ton payloads to LKO by plane, and I just posted a picture of such a typical heavy-lift wide load. I could show more -- there are tons of ways to place payloads on a plane without using cargo bays. In my current career, I've got a flatbed for wide roll-on roll-off cargo (currently in the field on the Kosmodrome on Laythe), my early career workhorse had four possible nodes for payloads plus tall landing gear so each node can take a 4+ metre payload, I've got variants of the big one I showed which carries heavy loads underslung, I've got designs which carry the payload on the back, and so on and so forth. You just have to think outside the box... or the cargo bay, if you will. 

And once again: yes it takes more thought, more testing, more planning, more experimentation, and more creativity than just building big fat rockets. But that's what KSP is all about, no?

[Warzouz]

Back in 1.0 / 1.1, I worked a lot on SSTO rockets systems.

I dig quite a lot in tweaking the design, and find that SSTO rockets are very easy to use. They are scalable very scalable. I build the 600T from the 300T in only 15min. It worked at the first try. SSTO planes are more efficient but aren't scalable. You spend a lot of time is you want to increase it's payload.

Even if I never got 100% recovery, I usually got around 97% return value. I used them a lot and don't remember recovering less than 90% even once. But as in 1.1, re-entry was changed and heating was increased, the design wasn't as good as it used to be. That's a good point because, SSTo rockets were way too easy to create, to fly and to land.

In the end, I use regular rockets because it's more fun to stage during launch. I don't care much now about cost effectiveness.

Here are some stats of my design (before 1.1)

 

[Guest]

40 minutes ago, Warzouz said:

SSTO planes are more efficient but aren't scalable

Kerbus, not this again.

If you're talking 300 tons to 600 tons payload then yeah, I believe you, not scalable. I've never done that anyway, and when I've pushed the limits past 300 tons the kraken starts biting. (I have theories about that though.) 

But they do scale perfectly well up to 300 tons or so, and if your base design is good and you have RAPIERs, scaling them up isn't hard either. It does take tweaking and testing but it's not... hard hard. 

For example, here's how I scaled my design. I started with a basic Boeing 707 type layout -- fuselage in the middle, tail at one end, long wings to the sides, a pair of engines on pods on each wing. Only, to fit the payload, I made the nose big and fat and the rest of a body a thin rail, so I can hang the payload inside fairings underneath. You can first build this design with a single Rapier on each pod. It'll lift a light satellite launch -- say, three or four small sats built around 1.25 m parts, with some bits protruding.

Then... scale it up. Put two RAPIERs on each pod, and scale everything else up by 50% in volume. Voilà, double the payload.

Then put four RAPIERs on each pod, and scale everything else up. You've doubled the payload again.

Then put two pods on each wing, and scale everything else to match. You've doubled it again.

Then put three pods on each wing, for a total of 24 RAPIERs, and scale up everything else to match. There you go, 300-ton lifter.

 Basically: If you can't do a 300 ton spaceplane which lifts wide loads, then all it means that you can't do it. It doesn't mean that spaceplanes qua spaceplanes don't scale.

[Brainlord Mesomorph]

5 hours ago, Brikoleur said:

Not true at all. Just use an external payload inside the standard fairings. As I said, I'm routinely lifting 4 m wide modules (2.5 m core with bits poking out), and was long before I even had access to Mk3 parts. All those fuel tanks on the payload are full BTW.

Where is (was) your CoL on that?

 CoM looks right, but have you already jettisoned a second wing or something? 

[Guest]

14 minutes ago, Brainlord Mesomorph said:

Where is (was) your CoL on that?

 CoM looks right, but have you already jettisoned a second wing or something? 

I haven't jettisoned anything other than the payload fairings. CoL/CoM are where they usually are, CoM a bit forward of CoL. 

The twist with this is, the heavy nose counterbalances the heavy payload. When fully fuelled but without a payload, it's very nose-heavy, so I place the payload a bit aft of CoM to balance it out. I also set the fuel flow priority on the nose tanks to -1 so they drain last: this means that aerodynamic instability won't be an issue as it'll be so far up when the nose tanks start draining that it won't go all flippy. 

With the payload released and the tanks mostly empty, CoM shifts right back to where it was when we started, and the craft is passively safe on re-entry -- it'll settle to somewhere around 60 degrees pitch when re-entering, and then nose down to coordinated flight once it hits thick air, somewhere around 20k altitude. 

So, basically, CoM fully fuelled and with the payload is just in front of CoL, and dry without the payload, in exactly the same place.

Edited February 6, 2018 by Guest

[Tex_NL]

18 hours ago, RizzoTheRat said:

I usually get about 97% recovery, very rare it's below 95%, and that's using chutes and airbrakes with only a short burst of throttle for a soft landing, couild probably do better if I had a bit more fuel (and therefore less payload) to play with on the way down.  I lost a few in development but I class that as simulation and reload.  Now I've got the hang of the entry profile I never lose one of my 45 tonne payload launchers, and as I also use them to recover Hitchiker cans full of crew I don't waste money on re-rentry capsules either.

Never landed on on the runway but I've got close.  I usually intend to land a bit short of the KSC though as that's a lot safer, they won't survive landing in the sea. 

Think it works out at about 550 kredits per tonne which isn't the cheapest, but way less than I can do it with disposables.  Not managed to get a heavier system to work though, everything I've tried so far has too much momentum for the drag and burns it's engines on the way down.

 

 

This exactly makes my point. If you are able to consistently land that close to the runway sooner or later you will land ON the runway. Either through practice or sheer luck, eventually it WILL happen.
So a statement it can never happen is, was and always will be pure BS !!!

 

 

Edited February 6, 2018 by Tex_NL

[RizzoTheRat]

2 hours ago, Warzouz said:

Here are some stats of my design (before 1.1)

 

How much difference have you found the latest versions makes to them?  I'd not played for a year or so until recently so a bit out of date on changes.  Quite prepared to believe my SSTO could have some improvements, but I'm using 4 Mainsails to launch 45 tonnes to 80 km compared to your 3 mainsails for 50 tonnes to 75km, and 150 kredits per tonne cheaper than me, which is a big difference.

[blakemw]

1 hour ago, RizzoTheRat said:

How much difference have you found the latest versions makes to them?  I'd not played for a year or so until recently so a bit out of date on changes.  Quite prepared to believe my SSTO could have some improvements, but I'm using 4 Mainsails to launch 45 tonnes to 80 km compared to your 3 mainsails for 50 tonnes to 75km, and 150 kredits per tonne cheaper than me, which is a big difference.

By my eyes the payload numbers look a little low. I don't use Mainsails because Twin-Boar is a much better engine (33% more thrust for less money and same weight, yes please). A recoverable Twin-Boar SSTO can deliver roughly 30.5t to orbit (almost as much as two Mainsails in that chart) at roughly $370/t @ 90% recovery. Actually if you go crazy a recoverable Twin-Boar can deliver 40.5t to orbit (more than a full orange tank) but that requires using wings to generate lift and burning way more fuel and is an awkward technique of dubious economic benefit.

Anyway as I said those numbers look a little low, though not outrageously so and maybe includes a comfortable margin for error. If you can't match them then you're probably doing something inefficient. One important change was that in 1.2 supersonic drag was changed: basically making blunt surfaces more draggy but making the drag more effectively mitigatable by streamlining. A sufficiently streamlined rocket can do an aggressive gravity turn and more-or-less ignore the atmosphere, the advantage being that more of the thrust is directed horizontally from early in the launch resulting in building up orbital velocity with less fuel burned.

You can also do a "lifting ascent", this is when instead of a strict gravity turn (zero lift turn) you initially tip over too fast then pitch up creating body lift. Strangely enough this seems to in some cases allow a slightly higher payload to orbit than a zero lift turn, probably because the lift/drag ratio can be as high as 2 for a basic long cylinder in KSP. It's hard to get a consistent advantage out of this technique because of the sensitivity of lift to angle of attack but the high amounts of lift does help to mitigate the drag and makes overly-aggressive gravity turns often about as optimal as ideal gravity turns.

Edited February 6, 2018 by blakemw

[Warzouz]

Use over 95% recovery because that what you'll really get. The worst recover I did was 93%. and rarely has been under 96%.

Now the issue with this design is overheating. The re-entry profile is more complex (including spinning like madness). But as I said, I don't use such device any more. I prefer building a non recoverable staged rocket. It's more fun.

[Brainlord Mesomorph]

1 hour ago, Warzouz said:

But as I said, I don't use such device any more. I prefer building a non recoverable staged rocket. It's more fun.

That seems to be the consensus of the more experienced players.

 

3 hours ago, Brikoleur said:

I haven't jettisoned anything other than the payload fairings. CoL/CoM are where they usually are, CoM a bit forward of CoL. 

The twist with this is, the heavy nose counterbalances the heavy payload.

I love to see some other angles on that. I just can't see it.

2 hours ago, Tex_NL said:

So a statement it can never happen is, was and always will be pure BS !!!

1. pardon my hyperbole (will never happen... to most players without several repeated attempts)

2. still, bit of an over reaction.

Edited February 6, 2018 by Brainlord Mesomorph

[septemberWaves]

Best way to reduce launch costs? Sandbox mode.

Though, in all seriousness, I find that designing a launch vehicle for a specific payload capacity and then designing payloads to use as much of that payload capacity as possible is a good way to go; it doesn't make the rockets any cheaper but it gets the most mass up there with the fewest overall launches. Of course, it doesn't always work (depends how many launch vehicles you want to design, and how much time you want to spend tweaking the payload masses, and whether you even want to use this approach) but I like to use it as a general guideline.

[Eric S]

On 2/5/2018 at 6:06 AM, Tex_NL said:

Agreed, landing back on the runway/launchpad can be difficult. But stating it will never happen is complete and utter BS ! With a bit of practice and proper design you can land back on the runway/launchpad each and every time.

I kind of have to agree that 75% recovery rate is pretty low.  I'm disappointed if I don't land close enough to get 97% recovery, and that's with rockets landed via parachute, not even spaceplanes.  Admittedly, I'm burning a lot of fuel so that's not 97% of the actual launch cost.

 

[Guest]

1 hour ago, Brainlord Mesomorph said:

I love to see some other angles on that. I just can't see it.

What exactly would you like to see?

 

[HalcyonSon]

23 hours ago, RizzoTheRat said:

I usually get about 97% recovery, very rare it's below 95%, and that's using chutes and airbrakes with only a short burst of throttle for a soft landing, couild probably do better if I had a bit more fuel (and therefore less payload) to play with on the way down.  I lost a few in development but I class that as simulation and reload.  Now I've got the hang of the entry profile I never lose one of my 45 tonne payload launchers, and as I also use them to recover Hitchiker cans full of crew I don't waste money on re-rentry capsules either.

Never landed on on the runway but I've got close.  I usually intend to land a bit short of the KSC though as that's a lot safer, they won't survive landing in the sea. 

Think it works out at about 550 kredits per tonne which isn't the cheapest, but way less than I can do it with disposables.  Not managed to get a heavier system to work though, everything I've tried so far has too much momentum for the drag and burns it's engines on the way down.

10 hours ago, Brikoleur said:

Suit yourself. I classify it as an elegant solution to a hard problem rather than brute-forcing it with a really big rocket. (You're shifting the goalposts by the way. You claimed that spaceplanes scale terribly to volume. I just demonstrated the contrary, and now you're going "but it's awkward.")

Also, fitting it in a 3.75 m fairing isn't trivial either -- for one thing, you'd end up with a tremendously tall rocket. That poses its own challenges. Making that rocket reusable would be even harder. Making it reusable and so that you can reliably land it at the KSC would be even harder. And even so, it would burn much, much more fuel, and would be harder to get to orbit efficiently and reliably because the gravity turn is trickier than than a plane's ascent profile.

I'll have to post a picture when I get home of my latest Mun Base launcher.  A Twin Boar with a couple 3.75m tanks and four Mainsails with accompanying 2.5m tanks.  I simply pile parachutes into a 2.5m fairing on top of each stack.  The payload is in a 3.75m monster fairing with no adapters, and docking ports instead of disconnects because they were already part of the design for on-Mun connections.  Saves on aero drag and on throw-away parts.  I had roughly 150 m/s of dV in the tanks after letting go of the payload in orbit.  I didn't even bother with landing legs or a powered landing because the whole booster landed about 50 km away from KSC in the ocean at about 4.5 m/s. 

Unfortunately, I don't think I grabbed a screenshot of the recovery value, but I can reverse engineer that.

I've found that it's often easier to place booster stacks radially to large/heavy payloads in fairings, rather than below the payload.  Each stack can then land separately.  If I have something that big (several hundred ton payload) to put in orbit, the additional cost and weight of a few parachutes and probe cores is minuscule compared to the benefit of returning three or four Twin Boar stacks.  The gravity turn remains the same as a start-of-game probe as long as the TWR is in the usual range (1.25 - 1.5 on the pad) because all of the large engines have gimbal.

[NewtSoup]

Just felt the need to exercise my smug gland

Parachuting back to the KSC is difficult.  But it's possible.  I can usually drop something in the ball park though not often on home base I'll give you that.  Why?  Because I get a kick out of it when it happens  I've never built a rocket SSTO.  My first plane spaceplane was today.

IIRC there is a mod out there that will recoup the salvage money from jettisoned stages if they have parachutes on.  The main reason disposables are disposable is because of the physics range  deleting stuff that's too far away when in atmospheric flight.  Otherwise I'd salvage every last little bit I could.

Here it is:

 

Edited February 7, 2018 by NewtSoup

[Allocthonous]

11 hours ago, septemberWaves said:

Though, in all seriousness, I find that designing a launch vehicle for a specific payload capacity and then designing payloads to use as much of that payload capacity as possible is a good way to go; it doesn't make the rockets any cheaper but it gets the most mass up there with the fewest overall launches. Of course, it doesn't always work (depends how many launch vehicles you want to design, and how much time you want to spend tweaking the payload masses, and whether you even want to use this approach) but I like to use it as a general guideline.

I like this approach too! I had a lot of fun designing a series of recoverable SSTO rockets that could reliably launch payloads based on the standard fuel tank sizes. It won't say it actually saved me any time, given how much of it I spent tweaking them, but it did make things faster later on when I could just merge designs in the VAB based on payload weight. I think the largest payload size I got up to was an orange tank, but it's been a while since I worked on that save.

As for launch costs, re-usables are probably my favorite method. Collecting multiple contracts that can be completed by the same spacecraft helps a bunch too.

[wumpus]

Yes, recoverable systems will keep most of your kerbucks intact.  I got really into this and burned out before making it to Duna.  I've since come back and avoid most recovery issues.

Kickers (big SRBs in general) are your friends.  They provide plenty of thrust and a good bit of delta-v for dirt cheap.  Drop a bunch of kickers before even thinking about recovering the rest.

There was a challenge for the cheapest kg to LKO called the "cheap and cheerful contest".  Most of the links are dead, but you should get the basic idea of what the winners looked like :

 

[Brainlord Mesomorph]

16 hours ago, Brikoleur said:

What exactly would you like to see?

 

How about top front and side from the SPH with CoL and CoM?

It looks to me (is that the bottom of the plane?) like the engine and payload are all hanging below the wing. That would put the CoL above the CoM.  (unless the wings are angled or something)  I've been doing large payload SSTOs and just want to see exactly what you did there.

[Guest]

4 minutes ago, Brainlord Mesomorph said:

It looks to me (is that the bottom of the plane?) like the engine and payload are all hanging below the wing. That would put the CoL above the CoM.  (unless the wings are angled or something)  I've been doing large payload SSTOs and just want to see exactly what you did there.

Sure, I'll post some once I'm home. Or if you like I can upload this -- or a later, more refined version -- to KerbalX so you can play with it yourself. (This one is mid-career and fairly rough.)

Yes the CoL is above the CoM. I put it there whenever I can, it makes them less twitchy, and in this case there's really not much choice as I want to leave as much room as I can for the payload under the wing.

In this version I have some vernors at the tip of the tail to stabilise it in space the same way the tailplane stabilises it in the atmosphere, but after making this one I discovered that putting a gimballed rocket engine there instead works better to compensate for thrust torque. Once in space I also switch off all or most of the lower row of engines to bring the CoT closer to the CoM. I key this to the same action that toggles mode on the RAPIERs.

[Brainlord Mesomorph]

Oh, you're burning fuel to compensate for the CoL being above the CoM. That would never occur to me. Now I understand.

An alternative might be to put a second, perhaps disposable wing under the payload.

My solution has been to strap two spaceplanes to a payload. CoM, CoL, and CoT line up by themselves with no need for directional tail thrusters.

thanks.

[Guest]

1 hour ago, Brainlord Mesomorph said:

Oh, you're burning fuel to compensate for the CoL being above the CoM. That would never occur to me. Now I understand.

An alternative might be to put a second, perhaps disposable wing under the payload.

My solution has been to strap two spaceplanes to a payload. CoM, CoL, and CoT line up by themselves with no need for directional tail thrusters.

 

Yep, I saw your spaceplane sandwiches. It's a cool and clever solution to a thorny problem, and in fact  one of my main inspirations for breaking out of the cargo bay mindset!

(A winged payload would not work in this case: the tail is sufficient to stabilise the craft in the atmosphere, and it won't do anything out of the atmosphere.)

(Edit: in case you're wondering why I didn't go with the sandwiches beyond an experiment or two, it's because of convenience: each half of the sandwich needs to re-enter separately, which means 50% more flying and 100% more pilots per launch.)

Edited February 7, 2018 by Guest

[KerikBalm]

On 2/5/2018 at 11:01 AM, blakemw said:

Also spaceplanes scale extremely poorly for hauling bulky payloads, payloads almost have to be designed for the spaceplane, whereas with rockets you just size the rocket and fairing to the payload.

Nonesense

*full disclosure, the rear 3.75m tank section was docked to the mk3 LF fuel section, so that payload comes from a total of 2 launches. This was long ago before I moved up to a 3x rescale.

On 2/5/2018 at 4:06 PM, Brainlord Mesomorph said:

I suggested that a couple of years ago and someone told me that ore is actually more expensive than refined fuel. So if you want to go that route, just mine huge quantities of ore and then recover it for cash, then buy fuel.

Don't believe everything that you're told. This is completely wrong.

[suicidejunkie]

On 2/6/2018 at 9:58 AM, Tex_NL said:

This exactly makes my point. If you are able to consistently land that close to the runway sooner or later you will land ON the runway. Either through practice or sheer luck, eventually it WILL happen.
So a statement it can never happen is, was and always will be pure BS !!!

I managed to land a capsule stack of suborbital tourists on the painted line down the middle of the runway twice, with FAR allowing the radial parachute boxes to act as miniature wings and provide terminal control before popping the cloth out.

The rockety bits all dropped into booster bay or the desert however.

[Brainlord Mesomorph]

52 minutes ago, KerikBalm said:

On 2/5/2018 at 10:06 AM, Brainlord Mesomorph said:

I suggested that a couple of years ago and someone told me that ore is actually more expensive than refined fuel. So if you want to go that route, just mine huge quantities of ore and then recover it for cash, then buy fuel.

Don't believe everything that you're told. This is completely wrong.

Well this was when ISRU was first released in stock, I mean within the first couple of days.

And I think it was true then but I'm sure they fixed it in later builds.