Is asparagus the best staging system? (might contain science)

[DerekL1963]

Too bad everyone except Korolev thought he was crazy. Korolev took on his idea of "packet rocket" (but not the cross feeding) and hence the now familiar shape of the R-7 / Soyuz rocket family's first stage.

Not quite. Korolev designed the R7 with droppable strapons for the same reason we put droppable engines on the Atlas, not because it was particularly efficient but because nobody was quite sure how to ignite the engines in flight, or quite how [tandem] staging would work. (Plus it kept the size of the individual units down for easy transport.)

[DerekL1963]

Not knowing a lot about rocket design myself, I have been struggling with creating proper efficient rockets. It always bothered me that, with 'normal' SaturnV type staging, you throw away engines each time and need to carry new engines in each stage.

Real rockets do that because you need different engine bell designs at different altitudes for maximum efficiency. (There's some other factors too, but that's the big one and one of the reasons for the ongoing interest in aerospike engines.) This is not (AFAIK) well simulated in KSP.

[numerobis]

KSP rockets are basically modeled as having bells that work at all altitudes and work best in vacuum. They also have mass flow dependent on air pressure (rather than having thrust vary). KSP doesn't model having a bell that will self destruct at the wrong altitude.

[capi3101]

<cast_spell_of_lesser_thread_necromancy />

Hey Temstar, quick question: say I wanted to employ asparagus staging for a launch from a world other than Kerbin. How would that adjust the formulas you use for asparagus staging? Would it adjust your formulas?

Thinking Eve here. Lander "payload" (the part I want to return to space) is 2.05 tonnes.

[Whackjob]

Great minds, thinking alike, etc.

[arq]

I use onion staging as a simpler alternative to asparagus with only slightly less efficiency.

As far as how to keep a high TWR in the later stages? One trick here is to make your outer stages carry more fuel. For example, for a 400t payload I put up I use 2 x32 tanks above my center stack and first ring (of 6). I then use 3 x32 tanks above the other rings. This gives you a lower TWR (I design for roughly 1.7 at the launchpad) earlier but can keep it from falling off as badly as you lose engines. If you were really trying to maximize TWR, you could use even fewer tanks on the inner stages.

Also, I sometimes design my innermost ring (around the core) to *not* feed into the center, so that I get to keep all 7 engines for their full burntime, but this means you need to feed the center stack from the 2nd ring.

The key to asparagus (or onion) staging is managing TWR. If all your stages start with at least 1.7 TWR and you can get ~4700 delta-v on there, it'll take you to orbit. Well, assuming it holds together...

[NoMrBond]

The other thing to consider is not every tank in your asp setup needs to have an engine under it, as long as the TWR for each stage in your setup is acceptable

[Tiron]

I made a rocket once that used giders for its primary structure, with tanks hanging off the sides and the engines on the girders at the bottom. Tanks would get kicked off the sides when they were expended, in asparagus pairs, with the engines not getting dropped at all. It was part of an attempt to make a supermassive fuel-to-orbit hauler, and that was only supposed to be the core, with the tanks in question not getting expended, at least not mostly.

The first test with just that core worked, arriving on orbit with almost no fuel left. All attempts to add additional tankage and SRBs in order to increase the mass to orbit resulted in horrible crashing, mostly caused by wobble in the girder-to-girder joints. The design was eventually abandoned because stabilizing it was requiring so many struts it was defeating the original purpose of getting the part count down versus a more traditional design.

Part of the core concept could work very well though: Keep the engines, and just drop tanks. It worked absurdly well, with the TWR constantly increasing as it climbed, and never a loss of thrust.

As for why asparagus isn't used in real life... Rather than angular momentum issues (Consider that the flow rate drops off at each engine, so most of the fuel flow is on the outside of the craft anyway, with much less reaching the inner portions), I suspect it's far more simple: Complexity, Reliability, and Expense.

Even with our click+click fuel pipes an asparagus setup is complicated and delicate, easy to screw up by doing one thing just a bit wrong. Imagine how much more so it would be in real life, where there isn't a simple GUI to build it with, and drag and drop off the shelf parts that snap together. Not to mention fuel pipes with unlimited flow capacity!

Imagine how hard it would be to build a fuel plumbing system with the capacity and capability to asparagus stage. How many pumps you'd need, how fast those pumps would have to push fuel on the first few stages. How massive the pipes on the first stage would have to be to supply 2 and a half engines, let alone more as we sometimes build. Consider how you'd cope with the fact that the burn rates would never be even, since individual rocket engines vary in thrust unpredictably.

When you think of it in terms of what it'd take to build it becomes pretty apparent that it'd be enormously complicated, and an incredible feat of engineering. And finicky as hell, because if even one part of the system failed to work properly, you're not going to space today.

It'd be difficult, unreliable, and expensive. So much so that the gains might not be worth it. At least not without a ton of development. Rather like the cluster engines the Russians are so fond of.

Edited July 4, 2013 by Tiron

[Whackjob]

The design was eventually abandoned because stabilizing it was requiring so many struts it was defeating the original purpose of getting the part count down versus a more traditional design.

MOAR STRUTS

[AndreyATGB]

^even my pretty good PC cries just seeing that..

As for asparagus: due to the smaller TWR of KSP engines, I find it necessary for any significant payload. It might be overkill for light payloads or low gravity landers, but it works very well regardless. It definitely doesn't look realistic at all (hell, they're all flying boxes to be honest) but they all get the job done.

I find asparagus very rollable, this ability increases with stages. My 250t lifter probably spins at 20-30RPM by the time the last stage is jettisoned. Unlike real-life, I just time warp to kill it, but the rolling (+lag) make manual control impossible.

It also requires insane amounts of struts to keep it together, my rule of thumb would be:

-all stages must be connected to payload

-all stages must be contented to the previous stage

-"complete the ring", basically the last stage strutted to itself on the other end, making a circle or square around the lifter

-connect them to each other as much as possible

-winglets on last stage helps with the spinning in the atmosphere, it still isn't enough though but at least it doesn't go helicopter mode by 50km.

[AustralianFries]

<cast_spell_of_lesser_thread_necromancy />

Greeeaaatttt necro.

(17 days later).

[Beerkeg]

Is the max payload what you have after the orbit or your whole ship?

[luchelibre]

Is the max payload what you have after the orbit or your whole ship?

For the purposes of the thread, it's what you're trying to leave in orbit. This doesn't count whatever final stage of the lifter might still be attached after circularization.

[annallia]

Say what you will about asparagus being better, there are times where raw power is the answer.

At least for me. Then again I say whatever gets the job done is good enough..

[luchelibre]

Say what you will about asparagus being better, there are times where raw power is the answer.

At least for me. Then again I say whatever gets the job done is good enough..

Jeremy? Is that you?

[Bothersome]

There is a place for asparagus staging (the Kerbal X) and then sometimes it would be more headache than it's worth (my SST-18 for example, see sig).

[SkyRender]

My own policy is, if it needs more than 7 orange tanks in asparagus formation, split it off into multiple launches. This has by and large resulted in most of my designs being perfectly suited to a 2-, 4- or 6-outer-tank asparagus staging setup.

[CaptainCrunch]

The above process is what I followed when I designed the Zenith rocket family:

Hands down my favorite lifters. When I'm goofing off I just slap something together. When I'm working on bigger projects that need an efficient lifter for my payload; I always fall back to these bad boys.

[Kerbart]

Great minds, thinking alike, etc.

HOLY F##K!!! Sir, I salute you!

More seriously, I think that compared to Kerbal, NASA and SpaceX are a bunch of pansies who put a disproportionate amount of value on boring things like "getting the payload into orbit" and undervalue "thar she blows!"

7 Engines = 7 points of failure. I'm sure someone can work out the calculus needed for n-th grade polynomals but it's late and I went to highschool when most of you young whippersnappers weren't even born yet. So I plug numbers in the only calculator one should use: the HP-15C (a 42 is accepted, too).

Say that the chance of failure of an engine for it's projected burn time is 5%

That means that with one engine, you have a success rate of 95%

With 7 engines, you have a success rate of .95**7 = .6983 = roughly 70% and that's assuming the failure rate PER ENGINE is the same even though complicated crossfeeding has to take place!!

With a total launch cost of say $10M, the single engine expected failure cost is 5%×10M = $500k

Of course your asparagus setup is more efficient, so it's costing less (is it? with all those engines?) Let's say 10% savings.

With a total launch cost of say $9M, the 7 engine expected failure is 30%×9M = $2700k

So you're saving 1M on your launch cost but you're increasing your expected losses with 2.2M per launch.

Of course I don't know what the real numbers are but I bet that the increased chances of failure make, despite increasing efficiency if, IF, a launch is successful make asparagus staging in the real world economically not as desirable as it would seem.

Kerbal is a different story of course. The bigger the fireball the better the launch!

[SkyRender]

-stuff-

Except that there is not and (according to the developers) never will be random failure of parts in KSP. So yes, in real life it'd be a somewhat bad idea to use asparagus staging (not that this stopped NASA for the Saturn V's design; 5 engines on the launch stage, 5 on the boost stage!), but in KSP it can only ever be beneficial.

[LameLefty]

HOLY F##K!!! Sir, I salute you!

More seriously, I think that compared to Kerbal, NASA and SpaceX are a bunch of pansies who put a disproportionate amount of value on boring things like "getting the payload into orbit" and undervalue "thar she blows!"

7 Engines = 7 points of failure. !

The Saturn I had 8 H-1 engines in the first stage and it's S-IV upper stage had 6 RL-10's. The current SpaceX Falcon-9 has 9 Merlin 1C engines and has made four successful flights, including one in which it survived a dramatic engine failure on ascent. The next launch will use uprated Merlin 1D engines and stretched tanks. That core configuration will be used in the three-core 27 engine Heavy variant; first flight scheduled for next year.

In short, yes, more engines equals more points of failure. But in the real world it also equals redundancy and engine-out capability without necessarily loss of mission. So it's a trade off.

[Kerbart]

I'm not saying that multiple engines are not used in real life; obviously they are. But with an asparagus setup the engines are seperated out so much that failure of one engine means an extremely uneven thrust distribution (I suggest turning one of your asparagus engines off for giggles during a launch) where the closely packed engines on the Falcons and the Saturns can be gimballed to correct for a failed engine (assuming there is enough thrust available).

[Eric S]

of failure make, despite increasing efficiency if, IF, a launch is successful make asparagus staging in the real world economically not as desirable as it would seem.

It's even worse than that, really. I get the impression that of launch failures that aren't attributed to human error are usually turbopump failures, especially when you realize that a lot of launches that are blamed on engine failures were actually turbopump failures, since the turbopumps are really part of the engine in most rockets.

In KSP, you might be able to get an asparagus-staged rocket into orbit if one engine fails. If one fuel line fails? Forget it, unless you get really lucky.

To make matters worse, the difference in efficiency between asparagus-staged rockets and traditional stacks, even assuming no launch failures, would be much less in reality than in KSP, because KSP has very low TWR engines and high-dry-mass fuel tanks.

I usually sum it up as: Asparagus staging is the natural and inevitable result of the combination of low TWR engines, fuel tanks with a high dry mass, and high throughput, high reliability fuel transfer ability. If those three conditions applied to reality, we'd already have asparagus staging in real life.

[LameLefty]

Generally good summary, Eric S. I agree with most of what you say.

The first step toward asparagus staging in the real world (clustering the tanks themselves) has been on the books, so to speak, since the 60's. The Saturn I was essentially created by clustering Redstone-type tanks into one thrust structure and interstage. The next real step was multiple boosters surrounding a sustainer core. That was accomplished with the Soviet hypergolic boosters (which have now have operated continuously for almost 50 years). The American Delta IV Heavy has three cores (two boosters and a central core that generally runs through a good portion of the ascent throttled down until after the boosters separate). That's how SpaceX will run the F9 Heavy as well. ULA (the operator of the Delta IV) has offered to develop cross-feed if their U.S. government customer is willing to pay for it. By contrast, SpaceX has publicly stated they WILL develop and implement cross feed for the F9H, but it won't be in the first versions.

Now, having said that, the F9 suffered an engine failure on a previous flight and it wasn't a turbo pump failure, but rather (apparently) a manufacturing or metallurgical problem in the fuel dome above the combustion chamber. Given the scale at which they have to produce Merlin engines (10 for each flight - 9 first stage, 1 second; 28 for each F9H) they are building at a scale not seen since the height of the Space Race. They are more likely to have manufacturing issues as they ramp up, as opposed to say PW/Rocketdyne who essentially hand-build their products at a much slower pace.

[wilt57]

One thing to be mindful of when using asparagus-style staging is that if all the stacks are the same, the payload becomes proportionally heavier as you drop stages. So the TWR will go down. That is not what you want at, say, around 10Km where you clear the thickest of the atmo and want to start hammering it east. This can be fixed if you make the stacks smaller (less fuel, same engine, more TWR) as you progress the staging sequence.

Or... just don't use asparagus-style staging. I do not. Sure it's more efficient, but more complicated and harder to build robust large craft.

Sorry to quote from so far back.

This is true, but with the right ascent profile this can be negated. Many players use LV-Ns with total ship TWRs of around .2 once in space because our orbital speed has reduced the effect of gravity on our ship. Even up in the thinner atmosphere, if we get our orbital speed up, our launcher can have a lower TWR than the initial stage. My Jumbo-64 fuel launcher has a payload of 40.42 tons. The core launch stage is a Jumbo-64 with a TVR-400L at the base with four LV-T45s for thrust vectoring. This stage only has a TWR of .98. The rest of my launcher consists of four Jumbo-64s with TVR-400Ls using LV-T30s and is asparagus style dropping 2 at a time. Total vehicle mass is 248.02, for a payload fraction of 16.3%. I use MechJeb to launch it to 80k, starting the gravity turn at 10k, ending at 70k, and using a 40% curve. There may be a better launch profile, but this one works.