Face to Face with The Tyranny

[Archgeek]

2 hours ago, GarrisonChisholm said:

I can talk about wanting a 100,000 m/sec probe, but that is a Lot of hobby time spent watching fuel tanks empty for accurate staging purposes.  I only burned a nuke last night and I spent over 3 hours on 2 burns- amounting to only about 3km/sec, as more than half the dV in my nuke stage was directed in a Normal or Radial direction so-as to maintain my velocity on my correct intercept.

Lastly, If I successfully plot an intercept of Valentine, as 9.5 Trillion km, "node-massaging" the intercept to maintain the contact but add velocity would be Hellacious!  "30m/sec Prograde, now 40m/sec Radial, now another 30 Pro- oops too far,..."   ...there are a lot more issues than "just" engineering in this endeavor!

Watching fuel tanks drain?  Normal and radial?  Okay, I smell two problems in execution here.  

1:  Crazy long burns like this pretty much demand you use Kerbal Alarm Clock -- just set alarms to pause the game at around each staging event, leave it running for an hour or however long the stage takes to burn out, come back, stage, check/correct vector if needed, repeat until near manuever completion, where you'll want alarms more frequently.  I like to do the last few minutes manually myself.

2: You're not trying for Hohnman transfer, are you?  When burning extra gas to cut time, you're likely to catch your target sooner than you otherwise would.  So, instead of chasing it around with a node, you aim for an early intercept, set a node that goes ripping past it, missing entirely, then set a braking node for slowing down, that actually makes the intercept happen.  For bonus points, try to hit objects in the target system with your spent stages during turnaround manuvers.

[GarrisonChisholm]

52 minutes ago, Archgeek said:

Watching fuel tanks drain?  Normal and radial?  Okay, I smell two problems in execution here.  

1:  Crazy long burns like this pretty much demand you use Kerbal Alarm Clock -- just set alarms to pause the game at around each staging event, leave it running for an hour or however long the stage takes to burn out, come back, stage, check/correct vector if needed, repeat until near manuever completion, where you'll want alarms more frequently.  I like to do the last few minutes manually myself.

2: You're not trying for Hohnman transfer, are you?  When burning extra gas to cut time, you're likely to catch your target sooner than you otherwise would.  So, instead of chasing it around with a node, you aim for an early intercept, set a node that goes ripping past it, missing entirely, then set a braking node for slowing down, that actually makes the intercept happen.  For bonus points, try to hit objects in the target system with your spent stages during turnaround manuvers.

I use KAC, but never thought of using it *this* way.  So, if the "stage" burn is 90 minutes, I'd have to figure out how long it takes my PC to run 90 minutes (at X3 Physics warp (x4 makes badness happen), best guess would be 2 hours for a 90 minute burn), then I could not pay attention for a while.  Ok, I'll give that a try. 

And, you know I think I was.  I matched planes, then tried to push the Ap of my orbit to the target's asc/decending node, which I guess is technially a transfer.  The way I ended up getting my intercept was (after an hour of faffing) "guessing" where End would be, and then just increasing my acceleration vector to draw it nearer, which I think is the second way you mentioned.  Then, getting my "accidental" intercept, I micro-massaged it to "full speed" on the current stage.

Next time I try this I'll do it your way, by just throwing out a high speed pass and then drawing the node retrograde to get the intercept. :]

That of course implies a high degree of confidence in where the target will be, and I find that far distant targets don't always support orbit elipses.

[Terwin]

15 hours ago, GarrisonChisholm said:

I use KAC, but never thought of using it *this* way.  So, if the "stage" burn is 90 minutes, I'd have to figure out how long it takes my PC to run 90 minutes (at X3 Physics warp (x4 makes badness happen), best guess would be 2 hours for a 90 minute burn), then I could not pay attention for a while.  Ok, I'll give that a try. 

I'd be careful of that, I believe KAC uses in-game time, so a 90 minute burn should only need a 90 minute alarm(perhaps more if your node is near a planet but most of the burn will not benefit from Oberth)

[Archgeek]

47 minutes ago, Terwin said:

I'd be careful of that, I believe KAC uses in-game time, so a 90 minute burn should only need a 90 minute alarm(perhaps more if your node is near a planet but most of the burn will not benefit from Oberth)

'Pretty sure 'e means guessing how long it'll take his computer to run a 90-minute burn so as to return to the device near to burn (or stage) completion.

[GarrisonChisholm]

So, as a faintly interesting follow-up to these discussions, my first launch aimed at an extra-Kerbol target bore fruit, and I have a probe en-route with an actual intercept of Valentine, 9.5 trillion km distant!  At 15.5 km/sec, Valentine SOI is only 50,000 years away.  So!  Let's see!;

At x7 Warp PC manages 2 days/sec

1 Kerbin year in 213 seconds

So, 50,000 years will only take... (math) ... about 4 months of PC time.  -.-

*looks up Better Time Warp, and vows to learn more from @Archgeek...*

[Archgeek]

Welp, I've completed the design of the b variant -- 62 stages wouning up with a little over 1km/s extra, ~85 tonnes lighter, but for 208 more parts.

KER is still being buggy about having ion engines drop away, so it's dv numbers are plenty off.  The design does, however, assume false things about the fuel flow, and so probably only really has around 88km/s like the 'a' variant.  As you can see, starting with the smallest tank I could resulted in a very long, noodly top section, and the stages under the plate were compact enough that I had to intentionally stretch the crap out of stage 49 so it'd have room for all the panels to power those 14 extra engines.  This piece of insanity isn't worth a .craft file.  I'll try again with c-variant that aims for something between 1.5 and 9 km/s/stage, and attempts to take advantage of the strange flow behaviour that actually exists.

[wumpus]

11 hours ago, Archgeek said:

Welp, I've completed the design of the b variant -- 62 stages wouning up with a little over 1km/s extra, ~85 tonnes lighter, but for 208 more parts.

60 hour total burn?  That seems low.  But if it is sufficiently low you might get away with using solar panels.

I'm pretty sure that having all those extra engines and decouplers aren't helping.  I'd recommend going with something much more asparagused (for lower burn time) and try to get the last bit (tall thin part) to increase fuel size quadratically.  The ion engine weighs twice as much as a full [tiny] xenon tank, so after the first few stages you need many tanks per engine (or better yet don't lose the engine in a stage: fire off drop tanks sideways (with your near-massless-decoupler of course) and use a single engine.

[Archgeek]

4 hours ago, wumpus said:

I'm pretty sure that having all those extra engines and decouplers aren't helping.  I'd recommend going with something much more asparagused (for lower burn time) and try to get the last bit (tall thin part) to increase fuel size quadratically.  The ion engine weighs twice as much as a full [tiny] xenon tank, so after the first few stages you need many tanks per engine (or better yet don't lose the engine in a stage: fire off drop tanks sideways (with your near-massless-decoupler of course) and use a single engine.

Hehe, you'd be surprised.  Decouplers make surprisingly little difference for xenon tanks, as one is lighter than a dry tank by quite a bit...you gain more from dropping a tank that you lose from having a decoupler.  It turns out this craziness works out best when each droptank stage has as close to the same mass ratio as possible.  This design is kinda the extremis of that -- just one little tank (dv is directly proportional to mass ratio, so I'll use that) at around 1500m/s, followed by gradually increasing tankage staying as close to the curve as possible.    I'm guessing they aren't visible enough, but two engines are actually at the top under that z4k -- this thing's a puller with a few extra stageable engines to give token consideration to TWR (keep it going below .02, pretty much)  That plate's got 6, and there are 14 more towards the bottom.  A single engine would have an insane burn time 190.2 hours, it looks like.  Even if using a lot less decouplers and engines cut the fuel needed in half, one engine would still take 95.1 hours burn time, which is a bit much.

Also, in practice much of the initial stage time would be periapsis kicks around kerbin, then a transfer to Jool for a sun-slingshot, followed by a hard, long, interstellar transfer burn around the sun.  Plenty of insolation everywhere but Jool, there.

[Archgeek]

Hmm, that reply might've been a bit rambling and seemed somewhat defensive.  This thread does a better job of explaining things: 

Significant investigation wound up concluding that best mass efficiency for stage distribution was equal stages (if all of 'em have the same Isp).  It's a little more complex than that but the xenon droptank problem turns out to be a rather convenient special case.  9km/s stages were sub-optimal due to lack of resolution resulting in a lot of fudging, piling on the mass too quickly(also, I think I miscounted the engines which may've thrown things far out of whack), while the smallest possible in stock turns out sub-optimal due to part count (which was in part due to needing to stretch a stage a lot to have room for the panels for its engines).  4.5km/s seems to be working out nicely, but I'll see when I get a chance to finish it up.

[wumpus]

18 hours ago, Archgeek said:

Hehe, you'd be surprised.  Decouplers make surprisingly little difference for xenon tanks, as one is lighter than a dry tank by quite a bit...you gain more from dropping a tank that you lose from having a decoupler.

I'm more concerned with the engines (which weigh more than the tanks).  If there is an ion engine with the decouplers (because I can't see them above it), that is where you are hauling way too much extra mass.  Also since a decoupler has 1/4 the mass of an empty stage, having more tanks make sense.  Looking closely, this was done in the example rocket (after 4 or so stages down, you see two decouplers per stage).

[Archgeek]

45 minutes ago, wumpus said:

I'm more concerned with the engines (which weigh more than the tanks).  If there is an ion engine with the decouplers (because I can't see them above it), that is where you are hauling way too much extra mass.  Also since a decoupler has 1/4 the mass of an empty stage, having more tanks make sense.  Looking closely, this was done in the example rocket (after 4 or so stages down, you see two decouplers per stage).

They do, which is why the probe's main body just has the two.  The next six on that plate are there to keep TWR from getting below .02, as are the 14 on the lower stages.  They don't stick around though -- the only point they really make up considerable fraction of the thing's mass is the last few stages, the very last of which still pulls a solid 9km/s.  Plus, their very presence, to my utter surprise, proved to only cost a few km/s for the same amount of fuel.

What's more the curve (which can only be followed so closely) is roughly quadratic.  The thing goes from single 700 tanks to 12 5220s per stage.  And what even is meant by aspargus here?  Xenon tanks drain from the outside stages in based on the count of decouplers between them and the root.  Mass-optimally, you'd have a long goldfish-poop train of droptanks.

[Gordon Fecyk]

On ‎7‎/‎7‎/‎2016 at 6:12 PM, GarrisonChisholm said:

That being said, if anyone has a craft-file for something approaching 100,000 dv, I'd love to see it-  if only to pick it apart and have its engineering inform my own decisions.

Turbo Pumped did something like that last year, and hit 103 km/s at Escape. Only had 75 km/s delta-v, but he used Jool and the sun to slingshot in, and back out.

(Never mind... I see OP referred to Turbo Pumped originally... a mod may delete this post if needed...)

--

Edited August 1, 2016 by Gordon Fecyk
(sorry, only found Turbo Pumped reference after the fact)

[Archgeek]

At last I have done it -- a feasible stock-part interstellar probe.  100,270m/s in 22 stages for only 170.4 tonnes, in a not-too-evil 466 parts.  The XenonBreeze Extra-Solar Explorer variant c:

Spoiler

 

It's even 10 meters shorter than the b-variant was, and a lot less noodly.

 

This thing is staged in a manner actually consistent with current 1.1.3 fuel flow, meaning it really does have 100km/s on board.  In fact, s11 and s12 actually take advantage of a quirk it possesses, dumping empty 700 unit tanks on the outside while a center column of 5200 unit tanks remains mostly full.  This saved enough mass to get away with only 12 extra engines in the last 5 stages, allowing for a quartet of 1x6 panels to be used instead of a 4th pair of gigantors.

Since this thing is actually flyable if you can get it upstairs (sectioning and orbital assembly may prove necessary), here's the breakdown of dv, burn time, and TWR per stage, plus some other stuff from the main design document:

Spoiler

And here's the .craft file:

https://dl.dropboxusercontent.com/u/59091477/Monstrosities/XenonBreeze Extra-Solar Explorer - c.craft

[teag2]

You could always try asparagus ions, with full usage of the Oberth effect you could probably get fast enough.

[Sharpy]

I don't recall the exact values, but I made a craft to save a kerbal from retrograde Sun orbit.

The first part of the journey - roughly to bringing the craft to a stop somewhere beyond Jool orbit - was on nukes. Then came a stand-up guy asparagus of ion engines with fuel cells. Each stage was a ion engine, a small xenon tank, and a small fuel tank, with every third stage containing the fuel cell battery (capable of sustaining 3 engines). The ugly part was no crossfeed: every time I'd burn 2 tanks worth of resources, before dropping a stage I had to move resources by hand, filling up all remaining stages from the one dropped. Still, good 30k across 4 stages, and under 2 tons or so.

Edited August 8, 2016 by Sharpy

[magnemoe]

1 hour ago, Sharpy said:

I don't recall the exact values, but I made a craft to save a kerbal from retrograde Sun orbit.

The first part of the journey - roughly to bringing the craft to a stop somewhere beyond Jool orbit - was on nukes. Then came a stand-up guy asparagus of ion engines with fuel cells. Each stage was a ion engine, a small xenon tank, and a small fuel tank, with every third stage containing the fuel cell battery (capable of sustaining 3 engines). The ugly part was no crossfeed: every time I'd burn 2 tanks worth of resources, before dropping a stage I had to move resources by hand, filling up all remaining stages from the one dropped. Still, good 30k across 4 stages, and under 2 tons or so.

You could have used Jool gravity to turn your orbit. I think I would set it up to use nukes until the low solar stage where ion makes more sense as solar panels would be effective. 
Now kerbin intercept and landing would be interesting
 

[Sharpy]

2 hours ago, magnemoe said:

You could have used Jool gravity to turn your orbit. I think I would set it up to use nukes until the low solar stage where ion makes more sense as solar panels would be effective. 
Now kerbin intercept and landing would be interesting
 

Jool gravity saved some delta-V, but increased the travel time to nearly a century, as the astronaut was on a pretty high orbit (around Eloo) and on the opposite side of the system.

Funny thing with tyranny of rocket equation: there's a number of maneuvers that allow you to save massive amounts of delta-V, providing you have several thousand years to execute them.

[magnemoe]

Just now, Sharpy said:

Jool gravity saved some delta-V, but increased the travel time to nearly a century, as the astronaut was on a pretty high orbit (around Eloo) and on the opposite side of the system.

Funny thing with tyranny of rocket equation: there's a number of maneuvers that allow you to save massive amounts of delta-V, providing you have several thousand years to execute them.

I don't get it, yes you would have to wait for an jool window but I thought an Jool flyby would let you turn around cheaper as you could use jool to help you, has not tried this myself so not sure how much you save, 

[Sharpy]

5 minutes ago, magnemoe said:

I don't get it, yes you would have to wait for an jool window but I thought an Jool flyby would let you turn around cheaper as you could use jool to help you, has not tried this myself so not sure how much you save, 

Cheaper yes. But how long does a Jool year take? Wiki says: 4 845 d 2 h 37 m 12.1

That's about 13 years. Eloo is 19 years. Middle between the two - about 16. Half of that - eight years from Jool to that kerbal alone. Never mind getting to Jool, then the return trip.

By going directly I was able to reduce the mission to something like 6 or 8 years. The gravity turn would surely save some delta-V but at cost of quite massive mission duration increase.

[magnemoe]

25 minutes ago, Sharpy said:

Cheaper yes. But how long does a Jool year take? Wiki says: 4 845 d 2 h 37 m 12.1

That's about 13 years. Eloo is 19 years. Middle between the two - about 16. Half of that - eight years from Jool to that kerbal alone. Never mind getting to Jool, then the return trip.

By going directly I was able to reduce the mission to something like 6 or 8 years. The gravity turn would surely save some delta-V but at cost of quite massive mission duration increase.

My thought was to enter Jool on something like an free return trajectory. Then do an radial burn to push your Pe so you come out retrograde with an fitting Pe, not tested it but used Mun for this around Kerbin. 

Your though was probably to enter an epileptic orbit and wait half an Jool year for the eject burn, should work. Standard on Laythe. Around Jool it would be an long wait as you say and you have to dodge moons all the time 

[Archgeek]

3 hours ago, magnemoe said:

I don't get it, yes you would have to wait for an jool window but I thought an Jool flyby would let you turn around cheaper as you could use jool to help you, has not tried this myself so not sure how much you save, 

Do keep in mind, the OP has no need for 100km/s to reach their targeted extra-solar system -- that's only there to cut transit time by going a lot faster than needed, then doing a turn-around burn to slow down for intercept, like an under-powered torch ship.  One can sac a few years to save a few km/s getting to the sun if they like, but the point of bringing all that gas in the first place is to get there faster.

Turnaround's going to be kinda slow going at first, since the target system's star will still be pretty far off, leaving the burn to have to throttle down enough to run largely on RTG power, but it'll crank up as the star looms closer.

[Sharpy]

8 hours ago, magnemoe said:

My thought was to enter Jool on something like an free return trajectory. Then do an radial burn to push your Pe so you come out retrograde with an fitting Pe, not tested it but used Mun for this around Kerbin. 

Your though was probably to enter an epileptic orbit and wait half an Jool year for the eject burn, should work. Standard on Laythe. Around Jool it would be an long wait as you say and you have to dodge moons all the time 

Not at all. I was planning to do a braking gravity assist combined with Oberth maneuver sending me onto retrograde Sun orbit.

First, gravity assist alone, is nowhere near to send one into a retrograde solar trajectory. You can at best modify your speed by half the orbital speed of the body you're doing the assist against, and in that case it would still mean prograde orbit, although with a pretty low periapsis (below Moho?)

Next, any roughly circular orbit at altitude of Jool takes roughly the same time to circle the Sun as Jool. So no, even not staying at Jool, I'd be still moving at about the same speed as it does. And to depart I'd have to burn twice Jool's orbital speed to enter the retrograde solar orbit. So, while Jool goes half its year "counterclockwise", I'd need to go half its year "clockwise". Never mind the target would have moved by quite a bit, so more like 3/4 Jool's year, passing Jool "head on" on the opposite side of the Sun.

The gravity assist would be a pretty good option if Jool and the target were in the right alignment - exitting Jool SOI on Sun's retrograde trajectory I'd be rather close to the target. This wasn't the case; the target was very distant and it would take a long time to catch up with it.

[Sharpy]

6 hours ago, Archgeek said:

 

Turnaround's going to be kinda slow going at first, since the target system's star will still be pretty far off, leaving the burn to have to throttle down enough to run largely on RTG power, but it'll crank up as the star looms closer.

Probably a much better idea is to utilize Oberth effect at the Sun to both speed up and slow down, That way not only you have massive fuel savings, you can run on very moderate amount of solar panels that grow immensely powerful near the Sun (and the flyby still takes weeks, so even 60-hour burns are still perfectly viable).

[Archgeek]

4 minutes ago, Sharpy said:

Probably a much better idea is to utilize Oberth effect at the Sun to both speed up and slow down, That way not only you have massive fuel savings, you can run on very moderate amount of solar panels that grow immensely powerful near the Sun (and the flyby still takes weeks, so even 60-hour burns are still perfectly viable).

That depends a lot on how massive the target system's star is.  If we're coming out of the Kerbol system like a bat outta heck with however much speed we manage to pull by burning ~46km/s worth of gas as close to the star's surface as we can, either the target system's star needs to be as heavy or moreso than Kerbol, or you're going to to want to start burning a bit early, like when coming in towards Moho.

[Sharpy]

15 minutes ago, Archgeek said:

That depends a lot on how massive the target system's star is.  If we're coming out of the Kerbol system like a bat outta heck with however much speed we manage to pull by burning ~46km/s worth of gas as close to the star's surface as we can, either the target system's star needs to be as heavy or moreso than Kerbol, or you're going to to want to start burning a bit early, like when coming in towards Moho.

a) it will still be more massive than Jool, so there will be *some* Oberth effect.

b) you're still saving a lot of mass in electric sources.

Also, you're coming in with a much lighter vehicle, so unless you dropped as much weight in engines as in fuel tanks, you'll have a better TWR.