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

[blakemw]

3 hours ago, KerikBalm said:

Nonesense

Well maybe I should rephrase it. It scales complexly. There's nothing wrong with the cost scaling but to be honest so far the examples of "it scales just fine" are the kind of payloads that'd be a piece of cake to launch with a single Mammoth or triple-Twin-Boar - basically dead simple rockets that take like 2-5 minutes to build and will very likely work perfectly first time.

Complexity doesn't appear as a career financial cost unless you're playing with reverts and quickloads disabled, in which cases complexity is directly proportional to RUDs which are expensive. But even with reverts and quickloads it comes at a very great play-time cost, which becomes an important factor if you want to spend time completing contracts, rather than building spaceplanes (if your goal is to maximize the amount of time spent building spaceplanes, then by all means, use spaceplanes).

My experience with career (hard career reward level) is that using disposable rockets you can usually get around 70% profit margin on contacts: 10% payload, 20% launcher, 70% profit. Using recoverable rockets roughly halves the launcher costs: 10% payload, 10% launcher, 80% profit. Spaceplanes cost about 1/3rd as much as rockets per flight (assuming 100% recovery and no RUDs), so it becomes 10% payload, 3% launcher, 87% profit. If you cheese ISRU you can eliminate the fuel costs: Rockets become 10% payload, 1% launcher (for missing the runway), 89% profit and spaceplanes become 10% payload, 0% launcher, 90% profit.

• Disposable: 70% profit
• Rocket recovery: 80% profit
• Spaceplane: 87% profit
• Rocket + ISRU: 89% profit
• Spaceplane + ISRU: 90% profit

Now the question is, how much play time are you willing to "pay" for the increase in profit per contract? Doing rocket recovery is an easy +14% profit. Lets say a contract takes 30 minutes to complete (build, launch, deploy), that would mean recovery should add no more than 4.2 minutes to the build and launch time in order to be worth doing, which is probably not going to happen, but it's not impossible, particularly as SSTOs don't require any staging so the build is simplified in certain respects and there is a 0% chance of staging mishaps (though that's kind of a positive aspect of rocket SSTOs, not their recovery per-se). Using ISRU would add another 11% profit, but you'd have to do it in under 4 minutes for it to be actually worthwhile, which is improbable, it's fiddly. Technically most payloads can usually be recovered on Kerbin thanks to the mild reentry heating experienced by un-streamlined things but you're highly unlikely to be able to do this in 4 minutes, hence most payloads get abandoned.

Here is where spaceplanes get well and truly wrecked, as using a spaceplane can double, triple, quintuple or more the time to complete a contract. In fact I'm sure contracts can take ten or twenty times longer when using a spaceplane when the payload is tricky. And it should be noted that in those cases where a contract can be completed by the spaceplane itself: say launching a 5 man station, then even in this case by recovering the payload you're only increasing profits from 80% to 97% and it's probably still not going to be worthwhile compared with abandoning the payload and moving on.

Imagine a question: "What is a good way to save money?". One possible answer would be "don't go to work, you'll save money on transit". But that's not a good way to save money, it's a stupid way to save money. This is like spaceplanes, they are a stupid way to reduce launch costs, the only good reason to use them is because you enjoy designing and flying spaceplanes. Rocket recovery is a marginal way to reduce launch costs, general consensus of experienced players is it's not really worthwhile but I think it comes close when using cheap SSTOs 4x speed and eating the loss if it RUDs (the big giant time waster is quickloads).

[Guest]

17 hours ago, Brainlord Mesomorph said:

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.

It just occurred to me that you may have gotten one thing wrong, which may be crimping your style when designing your craft. Apologies if I've misunderstood and am telling you something you already know.

Having the CoL above the CoM is not a problem, on the contrary, it makes your craft more stable in the atmosphere. That's why big jetliners camber the wings upward a bit. Having the CoL below the CoM will make the craft more agile, which is why some jet fighters etc. have a downward camber. 

However, having the CoT not line up longitudinally with the CoM produces thrust torque. In the atmosphere this isn't a problem as the tailplane (or canards) will stabilise it away the same way as the longitudinal difference between CoM and CoL. Your spaceplane sandwich resolves this by making the craft vertically symmetrical: the CoM and CoT always line up.

Out of the atmosphere, however, it becomes a problem as aerodynamic stabilisation no longer happens. Here's where switching off some engines to bring the CoM closer to the CoT axis, the gimbaled tail engine, or attitude control RCS thrusters placed in the tail come in: they now counter the thrust torque the way the tailplane does in the atmosphere. If the plane has a long tail, it gives it a long lever arm which relatively little force is needed to do the job, making the fuel expenditure minimal.

 

5 hours ago, blakemw said:

Well maybe I should rephrase it. It scales complexly.

You could have rephrased it even more simply: planes are harder than rockets.

I don't think you'd find anyone here to dispute that. Nor is anyone disputing that developing planes tends to involve a lot of crashing, so playing with them with quickloads and reverts disabled would certainly be more expensive than building big fat rockets that always work the first time and are sometimes recoverable for a part of the cost.

5 hours ago, blakemw said:

Imagine a question: "What is a good way to save money?". One possible answer would be "don't go to work, you'll save money on transit". But that's not a good way to save money, it's a stupid way to save money. This is like spaceplanes, they are a stupid way to reduce launch costs, the only good reason to use them is because you enjoy designing and flying spaceplanes. Rocket recovery is a marginal way to reduce launch costs, general consensus of experienced players is it's not really worthwhile but I think it comes close when using cheap SSTOs 4x speed and eating the loss if it RUDs (the big giant time waster is quickloads).

Oh come on, now you're being just silly. We play this game for entertainment, nobody's giving us an hourly salary. So it's obvious we're discussing in-game costs here.

In any case, you keep on answering a different question than the OP posed -- if the OP's question had been, "What's the quickest way to make most money in KSP, per minute spent at the keyboard?" or "what's the least risky way of playing KSP, for example when you're playing with quickloads and reverts disabled?" then your answers would make some kind of sense.

But that's not what he asked, is it now?

[KerikBalm]

20 hours ago, Brikoleur said:

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.)

I had a similar but earlier concept of strap-on SSTOs. These were a bunch of mini SSTOs saved as subassemblies, that I just radially attached to a payload as needed. I used them mainly(but not exclusively) for low mass but large dimension payloads that didn't justify launching my heavy cargo SSTOs. In particular I used them for winged payloads that couldn't SSTO on their own. Often they were essentially air breathing engine pods with just enough wing area that they could fly themselves back and land with mostly empty tanks. Unlike a spaceplane sandwhich, these oftern involved 3 or 4 independent re-entries. I had a 2 and 4 engine version, because I wasn't going to do 8 re-entries.

Example payloads:

Spoiler

Winged Duna lander, for precision landings near a surface base:

Moded electric engine, solar powered science plane:

Duna cargo plane, for moving base modules'/rovers around Duna/to and from orbit:

In some cases, like the tail sitting winged duna lander, I had to do a vertical takeoff, so I deviated from a strict SSTO, and had strap on SRBs as well that just burn for a couple seconds  to allow the craft to pitch over and transition to vertical flight. The SRBs would parachute down to the runway before leaving the physics bubble, so it was still 100% recoverable.

^this bottom picture was before a redesign of the strap-ons, with the canards that far forward, they were a bit too unstable during reentry.

If you just want to lift a heavy payload without wings, then the strap-on SSTOs need more ing area than these have, obviously.

More pics of strap-ons, the 2 engine version I made a while back:

Spoiler

Launching an earlier version of an electric-motor duna plane

And the tail sitting winged duna crew transfer vehicle in flight with strap on SSTOs over kerbin v

and the SRBs required to launch the thing, since its large X profile wings precluded a horizontal takeoff. Note that they were 100% recoverable, even if its not strictly SSTO.

With Strap-On designs, its quite easy to scale with payload, just attach more. Its not really so much more complex, its just the same sub-assembly repeatedly symmetrically attached near the CoM (If not symmetric, like that 3 strap on electric motor plane, the offset tool must be used to ensure thust is inline with CoM). It does get part count up quite high though because multiple small tanks and wings are used - so it doesn't scale well just like part count doesn't scale well. The solution is to make bigger "strap-ons", which obviously would lead to Mesomorph's "spaceplane sandwich".

I do 3x rescale kerbin with stock parts, and so far my attempts at strap on spaceplanes have not gone well, they are all "chemical only", and barely make orbit, yet I recently got my "twin boom" cargo spaceplane (which includes nukes) to a 17% payload fraction. I think the problem is using the smaller parts, as I got a "chemical only" 2.5m and mk3 cargobay design to reach a 5% payload fraction, so in theory I could use a "spaceplane sandwhich" system in 3x rescaled kerbin... but in 3x rescale the SSTOs are so large that I'd worry about them collapsing under their own weight when placed on the runway (its better once they are supported by wings, which distributed the load more evenly)

Another method I played around with early on was airbreathing vertical launch, horizontal landing SSTOs, somewhat similar to the space-shuttle.

Their payload fraction wasn't as good as a "normal" SSTO spaceplane, but they had many of the advantages of vertical launch SSTO rockets, with a greater capability for 100% recovery

^ I think this was before a nerf to the KR-2L

And as for other launch cost reducing measures... I found shuttle style semi-reusable designs to not really be competitive with disposable launchers (in this case I'd recover the ET, not the boosters like the shuttle). This was about 700 funds per ton to orbit, and it barely made orbit with maximum payload:

(you can see the DV readout of 3171... maybe I should have entered it for one of those challenges to get to orbit with the lowest dV)

I also tried a Musk-esque flyback booster design, merged with a shuttle design (because KSP's simple control scheme doesn't allow for pinpoint returns to the pad with just rockets and a ballistic trajectory:

Its flight profile was very complicated because of gameplay engine limitations and the need to get an Ap over 70km, and switch between vessels. I also tried flying it without detaching the booster for a flyback, but continuing to orbit with the booster attached (transferring fuel and tweaking thurst as needed to keep it under control). I got about the same payload fraction whether I flew the booster back after a suborbital decouple or if I took the booster to orbit. It had airbreathing panthers, but most of the impulse was not from airbreathers, and it just did a shallow gravity turn (had to get the booster Ap above 70km, didn't want too much horizontal velocity at 70km because of the retroburn of the mostly empty booster).

So I would advise against trying flyback boosters within KSP for reducing launch cost (KSP parameters deviate from real life, this is no comment of the falcon flyback booster).

Vertical launch, horizontally landing SSTO rockets work fine, IMO, but also don't bother with partially reusable shuttle designs.

Edited February 8, 2018 by KerikBalm

[Guest]

1 hour ago, KerikBalm said:

Vertical launch, horizontally landing SSTO rockets work fine, IMO, but also don't bother with partially reusable shuttle designs.

On the other hand, if they're able to land horizontally, why not also take off horizontally? Point 'em up at 45 degrees and watch them go.

(Okay, I can see how a VTO/HL rocket would be easier, as you don't have to worry about CoM shift so much: it's perfectly fine to have a really nose-heavy rocket, so you'll just have to make sure they line up when it's empty.)

[KerikBalm]

Exactly the CoM issue... you can put a heavy payload on the nose, have lots of fuel ahead of the engine, etc. The CoL is so far behind the CoM when loaded that its just going to follow prograde, not maintain enough AoA for aerodynamic flight. Those designs ascend with a gravity turn, and can't deviate much from a gravity turn... just steep or shallow gravity turns. If using turboramjets (not rapiers, because of static thrust/TWR), a shallower gravity turn will get better results.

Also note the heavy use of rocket thrust and LFO, horizontally flying rocketplanes are pretty inefficient. You could spam jet engines for a vertical launch "rocket" without using oxidizer in the early part of flight as well, I suppose. But if you go for horizontal flight, then you lose the advantage of the vertical launch, mainly that you just balance it for empty tanks and no payload for the horizontal landing, for a vertical launch you don't care how far forward the CoM is shifted by fuel and payload.

[Guest]

How much do airbreathers help for a gravity turn ascent? You're not spending all that much time in the atmosphere where they help.

I'm currently in a "planes only [almost]" career, which meant rocket planes for a looong time, and you might be surprised at how efficient a HTOL rocket plane is. Once off the runway you point it up at 45 degrees until 10k/500 m/s, then follow prograde like a rocket. You lose some to drag but you win part of it back by having a lower starting TWR and building velocity instead of altitude. 

TBF I didn't do a comparison between VTO/HL and HTOL, but I did get to 40-ton payloads with Skipper-based rocket planes which was sufficient for the time. 

(And also I "planes only [almost]" was a limitation I set on myself because I wanted to, not because I thought it was the most efficient way to play early on. So I would expect that VTO/HL rocket planes are more efficient than HTOL rocket planes. It's just that HTOL rocket planes aren't maybe quite as inefficient as you might think.)

Edited February 8, 2018 by Guest

[Brainlord Mesomorph]

4 hours ago, KerikBalm said:

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.)

1. drone cores (everything I make can fly manned or unmanned)

2. To me the fun part of SSTO planes is putting them back on the runway.  With the sandwich I get two fun landings for each boring takeoff

[Hotel26]

7 hours ago, Brikoleur said:

That's why big jetliners camber the wings upward a bit.

Note also dihedral angle.  Causes wings to tend to level themselves.  The wing that is down has a more orthogonal angle of attack and generates slightly more lift, canceling the wing down state.

Edited February 8, 2018 by Hotel26

[Guest]

36 minutes ago, Brainlord Mesomorph said:

drone cores (everything I make can fly manned or unmanned)

I'll see your drone cores and raise you a CommNet blackout.

37 minutes ago, Brainlord Mesomorph said:

To me the fun part of SSTO planes is putting them back on the runway.

I can relate to that.

[KerikBalm]

49 minutes ago, Brainlord Mesomorph said:

1. drone cores (everything I make can fly manned or unmanned)

2. To me the fun part of SSTO planes is putting them back on the runway.  With the sandwich I get two fun landings for each boring takeoff

Ummm I didn't say what you quoted, and what this is in response to, Brikoleur did. I quoted that section to respond to it though.

I use probe cores a lot too, but I often play with plasma blackout on though... But I have mixed feelings on that because the probe cores should be able to do more than what they currently can without a connection - such as hold a fixed AoA and keep wings level.

Edited February 8, 2018 by KerikBalm

[Guest]

3 hours ago, KerikBalm said:

But I have mixed feelings on that because the probe cores should be able to do more than what they currently can without a connection - such as hold a fixed AoA and keep wings level.

I use Atmosphere Autopilot which lets probe cores do just this even in blackout. I think it's an unintended side effect but I don't feel bad about it because I agree with you.

[Nich]

On 2/8/2018 at 6:25 AM, Brikoleur said:

How much do airbreathers help for a gravity turn ascent? You're not spending all that much time in the atmosphere where they help.

I'm currently in a "planes only [almost]" career, which meant rocket planes for a looong time, and you might be surprised at how efficient a HTOL rocket plane is. Once off the runway you point it up at 45 degrees until 10k/500 m/s, then follow prograde like a rocket. You lose some to drag but you win part of it back by having a lower starting TWR and building velocity instead of altitude. 

TBF I didn't do a comparison between VTO/HL and HTOL, but I did get to 40-ton payloads with Skipper-based rocket planes which was sufficient for the time. 

(And also I "planes only [almost]" was a limitation I set on myself because I wanted to, not because I thought it was the most efficient way to play early on. So I would expect that VTO/HL rocket planes are more efficient than HTOL rocket planes. It's just that HTOL rocket planes aren't maybe quite as inefficient as you might think.)

I have tried this again and again but I can never surpass 1100 kredits/ton.  I think KSP tanks are too heavy to make it doable.  I should try this in RSS but there engine burn times would probably limit you from doing anything useful.  I think there is one soviet sustainer that might be usable.