Maximum Delta-V rocket

[GoSlash27]

6 hours ago, Tatonf said:

So much empiric tests while the maths are quite simple. Just take the rocket equation :

Delta V = v_e * ln (M_initial / M_final )

If you were adding an infinit amount of fuel tank, that would push M_initial to infinity. If you were to calculate the limit of this equation with M_initial --> +oo ; knowing that M_final and v_e  are positive finite numbers, then you would get an infinit delta V.

Tatonf,

 Remember that tanks have an absolute maximum Rwd of 9:1. As payload and engine mass approach zero, Rwd approaches 9 and therefore DV approaches 21.6*Isp.

 But of course, this is for a single stage only.

Best,

-Slashy

[Streetwind]

13 hours ago, GoSlash27 said:

tanks have an absolute maximum Rwd of 8:1.

(Fixing your typo so people don't snap up the wrong thing.)

[Jhawk1099]

On 5/25/2013 at 6:23 PM, Arganth said:

Woah ^^ thx for all that input

just landed on duna and had one hell of a time constructing a rocket for it

time to get back to the rocking drawing board

This is exactly me. Right now im getting ready to build a minmus refueling station while i wait for the phase angle to get my guys back

[GoSlash27]

4 hours ago, Streetwind said:

(Fixing your typo so people don't snap up the wrong thing.)

Streetwind,

 Actually, it's not a typo. The Rwd of a tank is 9:1. The tank itself is 1/8 the mass of the fuel it contains, but the full mass is the fuel plus the empty tank.

Best,
-Slashy

[Streetwind]

Then I misunderstood you, @GoSlash27. Apologies.

[MarvinKitFox]

Max I've ever managed was 62835m/s

That is ion engine with 5 stages of drop tanks, on top of 3-stage nuke, on top of the biggest thing I could get off the ground.

Swing that baby on a very low orbit around the sun for ultimate oberth effect boost.

(this was under KSP 0.24, so solar heat was not a problem)

 

Top speed was in excess of 110000m/s, it left the system at more than 83000m/s

 

The image below is NOT one of this specific vehicle, this one is from a "how fast can you impact Duna" quest.

But it shows a similar maneuver as described above, only at 100000 instead of 110000, and with no final ion stages, just the nuclear stage.

It is *easy* to tack on another 40K m/s of ion propulsion on top of this thing.

http://i.imgur.com/agpBxjL.png

Edited January 25, 2016 by MarvinKitFox
looking for matching pic

[wumpus]

On 3/7/2014 at 4:36 PM, Xaiier said:

Now let's come back to reality, and realize that DV is UTTERLY DOWNRIGHT USELESS without an equivalent TWR stat. The two are entirely interchangeable, so giving one without the other is pointless! This craft I've made, while having a downright monstrous DV, has a TWR so useless that it could never ever be flown to anywhere. (and the sad part is, I've seen ships posted on these forums that have a TWR like this!)

If we are going to talk about the maximum DV, we need to include TWR into our discussion, because otherwise delta-v means nothing.

There is nothing interchangeable about delta-V and TWR.  First of all, TWR only applies when you are in a gravity field and fighting against it (just thrusting prograde/retrograde won't change anything).  Use TMR in such rants.

Obviously, a full "Voyager" probe (with full capture everywhere) would be painful with even the low TMR you would get with a 10-20km/s delta-v (asparagus plenty of NERVs, please) (better yet, refuel on the way back from Moho).  For absolute theoretical flights, it would be presumably for interstellar flight.  You would have to make sure your craft could handle 4x physics time acceleration and let the thing burn overnight.  Unfortunately, I couldn't find a mod with interstellar objects to discover, and at .0002C your craft isn't going to get anywhere (interstellar) fast.

One thing that *does* matter with high deltaVs and low TMR, you won't get anywhere near that "official" delta-v that kerbal engineer lists.  You can only kick in perigee until you hit escape velocity, then you are ejected out of the SOI, and you can't make your burns long enough to matter at such wimpy TMRs.  Your best bet is calculating delta-v manually, or possibly using Gilly or something as "location".

[GoSlash27]

4 hours ago, Streetwind said:

Then I misunderstood you, @GoSlash27. Apologies.

Oh, no worries

Best,

-Slashy

[Teutooni]

I remember seeing this thread a few months back. I had a few hours to waste in matlab trying to figure out what is the theoretical and achievable maximum delta-v stock KSP can achieve (you know, for excrements and giggles).

Single stage

Pretty straightforward, obtain fuel tank full mass to empty mass ratio (9 to 1 for most tanks in KSP), and plug it into Tsiolkovsky rocket equation. As your fuel tank mass approaches infinity, the total vessel mass ratio approaches this dominant fuel tank mass ratio:

dV = Ve * ln(m0 / m1)

For LV-N, it is around 17km/s.

Serial staging

Now it gets a bit more complicated. I figured there needs to be an optimal point for each stage that maximizes delta-v and minimizes mass. Pretty straightforward, divide delta-v by vessel mass and calculate maximum.

https://www.dropbox.com/s/6baik93mnluxl8w/dv.png?dl=0

This shows the total delta-v as a function of number of fuel tanks. In this example I used mk1 liquid fuel fuselages as fuel tanks, with total payload + engine mass of 3.3 tons, up to 100 fuel tanks.

https://www.dropbox.com/s/tgz702mriymk84p/mratio.png?dl=0

This plot shows the delta-v divided by mass. There is a clear tipping point at 2 fuel tanks, after which it goes downhill. In fact, I discovered that for all subsequent stages, the delta-v-to-mass ratio tipped at around 5.98 km/s, no matter how far down the staging. Obviously this number depends on the engine and fuel tank mass ratio. This resulted in some interesting simplifications, namely that to get an optimal serial staging, each stage added roughly 6 km/s to the total delta-v and multiplied the total mass by about 2.5.

Now we can calculate some large numbers to get really high delta-v. Let's say interstellar  travel would require 200km/s delta-v. 200 / 6 = 33.33 stages. This would multiply the first stage mass (8.3 tons) by 2.5^32, roughly. Which is about 45 trillion tons. Speed of light would be possible in theory (newtonian physics), but would *still* require a rocket with a mass of roughly 10^20000 tons. I think you'd have a hard time coming up with a multiverse theory that had anything in that kind of scale. Note that there is no true theoretical limit of delta-v, but the exponential mass curve will quickly shed any ties to reality.

In practice, serial staged rockets with LV-N can reach around 65 km/s with part count in the high 2000s. Obviously this massive ship would need to be assembled in orbit and would have an utterly useless TWR, requiring probably thousands of years to burn. But it's in the range that *could* probably be built in game with a monster PC.

Btw while ion engines have decent ISP, they have truly horrible fuel tank mass ratio of a bit over 2, so they are actually only about 2 to 3 times as good as LV-N for total delta-v. So for example for 200km/s, you'd need 15-22 trillion tons instead of 45.

An actually useful figure from all this could be the optimal stage delta-v for each engine type in serial staging. This can give you a rough idea what to aim for for each stage.

ISP      optimum stage dV
300      2240 m/s    
350      2616 m/s
800      5980 m/s
4200    15200 m/s

There is a possibility I missed something crucial, so any comments are welcome! The numbers do seem reasonable in my experience.

EDIT: Pffft how do I embed images?

Edited April 7, 2016 by Teutooni

[Frank_G]

My Eve ascent vehicle had 12.300 m/s after touching down on Eve (TWR 1.1x) back in .90. Together with the docked transfer and separate return stage it must have had something around 15.000 m/s +. But as said, the last 3.000 m/s came due to orbital assembly.

The craft is still available on my craftfile page. Havent checked its numbers in 1.0.5. though, as launching from Eve became a lot easier (which is good in respect for players who dont fiddle on craft all night long).

Edited April 7, 2016 by Frank_G

[theend3r]

I had some really interesting designs before the OP ARM parts came, but I never computed their dV and MJ/KER couldn't since I was manually shedding fuel tanks in pairs and was too lazy to stage that for one mission.

Spoiler

 

 

Spoiler

 

The last one took an unflippable rover and lander to Eeloo and back. It also required manual staging by docking. Ah, good old times.

Edited April 7, 2016 by theend3r

[KerbonautInTraining]

This guy made a rocket with 75000 m/s and launched it.

ION ENGINES FTW

 

[Archgeek]

2 hours ago, Teutooni said:

I remember seeing this thread a few months back. I had a few hours to waste in matlab trying to figure out what is the theoretical and achievable maximum delta-v stock KSP can achieve (you know, for excrements and giggles).

[...]

An actually useful figure from all this could be the optimal stage delta-v for each engine type in serial staging. This can give you a rough idea what to aim for for each stage.

By serial staging, do mean just sans cross-feed?  My crazy ion experiments could stand to benefit from better droptank staging.

 

[Dman979]

3 hours ago, Teutooni said:

There is a possibility I missed something crucial, so any comments are welcome! The numbers do seem reasonable in my experience.

What about asparagus or onion (Each inner stage is fed by an outer stage, which is fed by another, and so on. As apposed to aparagus, where usually 2 stages feed all the other ones, and then get dropped away.)?

Have you factored in the weight of the decoupler, too?

[Teutooni]

36 minutes ago, Dman979 said:

What about asparagus or onion (Each inner stage is fed by an outer stage, which is fed by another, and so on. As apposed to aparagus, where usually 2 stages feed all the other ones, and then get dropped away.)?

Have you factored in the weight of the decoupler, too?

The way I understand staging is, parallel or onion staging is less efficient than serial (stages on top of each other, i.e. the normal staging method), but they benefit from all kinds of TWR considerations. Asparagus is one of the most efficient staging methods when considering delta-v, mass and TWR. TWR (or TMR) was left out of this consideration on purpose. In any onion or asparagus staging that preserves centre of thrust and mass, two fuel tanks need to be dropped at each stage. In serial staging only the dead weight of one near empty fuel tank needs to be dragged along until each separation, thus making it slightly more efficient in terms of delta-v.

Actually come to think of it I might be wrong on this, needs checking.

Anyway, in my calculations I used ~3 tons extra weight for each stage consisting of the engine, decoupler and any other stuff I might want to slap on, such as struts. This 3 tons is vanishingly small in large stages though and the actual final payload has a high impact on just the first few stages. Remember I considered rockets potentially trillions of tons in mass driven by a single 60kN nerva, so yeah...

Edited April 7, 2016 by Teutooni

[Jhawk1099]

On 5/25/2013 at 6:23 PM, Arganth said:

Woah ^^ thx for all that input

just landed on duna and had one hell of a time constructing a rocket for it

time to get back to the rocking drawing board

Was it a one way trip? If so the  Δv should be less than a round trip from Minmus which is an easy trip even for most beginners. 

[Nich]

My best was a max dv challange.  it had 19.5km in LKO with only 2 hours burn time

[wumpus]

7 hours ago, Dman979 said:

What about asparagus or onion (Each inner stage is fed by an outer stage, which is fed by another, and so on. As apposed to aparagus, where usually 2 stages feed all the other ones, and then get dropped away.)?

Have you factored in the weight of the decoupler, too?

Before 1.0, traditional onion made zero sense.  By "traditional onion" I mean drawing from multiple fuel tanks (more than one pair) and feeding multiple engines.  You would always get more delta-v (and higher TWR) by grouping the engines together and dropping each pair of fuel tanks as drop tanks, and then finally dropping all the engines along with the final pair of drop tanks.

Post 1.0, this whole thing *might* work better aerodynamically with the classical setup, but I'd be extremely  surprised.  I'd still expect a poodle+multiple drop tanks to beat a terrier-based onion setup.

Note: for any asparagus-inspired design, always try to connect each pair of tanks to the next pair of tanks to be dropped.  This means that any extra decoupler parts are jettesioned in the next stage.

[zarakon]

There was a challenge a while ago:

• Build a rocket to put a 1-man pod into orbit
• Build a rocket that could put THAT rocket into orbit
• Build a rocket that could put THAT rocket into orbit
• etc

I ended up taking that to recursive level 5, which resulted in a nearly 7000-ton monster with 23km/s on only chemical rockets with no asparagus or droptank stuff, just straight vertical staging

I actually had a version before that with slightly more, but it was uglier

Edited April 8, 2016 by zarakon

[suicidejunkie]

19 hours ago, Teutooni said:

Serial staging

...Obviously this number depends on the engine and fuel tank mass ratio...

There is a possibility I missed something crucial, so any comments are welcome! The numbers do seem reasonable in my experience.

You can make it notably better by having just a single engine set on the top of the rocket (a pair of thuds for example, or LV-Ns on arms), and then dropping only fuel tanks and separators, never engines.

If you throw in an engineer with a KAS tool, you don't even need the separators!

[Nich]

Oh and FYI oberth is less and less important the more and more dv a burn takes.  If you want to get .5c and have infinite TWR your gravity losses out of the gravity well would be nearly 0.000000 dv.  If you spent 1000 years accelerating your gravity losses would be escape velocity (about 1000 m/s for kerbin) However this is only .000000000000001% of your total burn and becomes insignificant

Edited April 8, 2016 by Nich

[Archgeek]

Welp, this happened in light of this thread: (KER was failing at the staging for some reason, but each stage has been tuned for around 15.2km/s, save for the ship up front, which carries 8.2 or so, making for 84.2km/s total.)

Considering how little the decouplers weigh, I've my doubts that the 15.2km/s stage rule quite works out here, especially for the very large tank stacks on S6.  However, since the little KER bug means I have to lock all the tanks in the later stages to get an accurate read on a given stage's dv, I'll have to save that tedium for later.

[Aozora7]

I made a 17000 m/s rocket with parts from KW rocketry for my manned Moho mission. 

I never built things only to get most dV out of them. When I bother making something, I want it to be useful. 

Edited April 9, 2016 by Aozora7

[Teutooni]

7 hours ago, Archgeek said:

Considering how little the decouplers weigh, I've my doubts that the 15.2km/s stage rule quite works out here, especially for the very large tank stacks on S6.  However, since the little KER bug means I have to lock all the tanks in the later stages to get an accurate read on a given stage's dv, I'll have to save that tedium for later.

Yeah I think my initial assumption that maximizing delta-v to mass ratio for each stage does not get optimum total delta-v. I reckon truly optimal staging would be 1 engine, fed by drop tanks arranged in an asparagus-esque pattern, with somehow optimizing tank weight vs decoulper weight. So potentially having a decoupler for each tank and thousands of stages. Like someone pointed out, theoretically even better would be to have a single kerbal detatch each fuel tank manually with KIS/KAS, so every single fuel tank could be jettisoned when empty with no decoupler mass cost.

[GoSlash27]

Teutooni,

Quote

 

An actually useful figure from all this could be the optimal stage delta-v for each engine type in serial staging. This can give you a rough idea what to aim for for each stage.

ISP      optimum stage dV
300      2240 m/s    
350      2616 m/s
800      5980 m/s
4200    15200 m/s

There is a possibility I missed something crucial, so any comments are welcome! The numbers do seem reasonable in my experience.

 

Hmm... definitely food for thought...

Best,
-Slashy