Atomic Motors Not Worth It

[Frederf]

Nothing's debris which is released to collide with the Sun!

As far as optimization math goes. I'm very curious where the maximum efficiency point comes in with regards to TWR vs dV. Long duration burns have an associated cost with respect to Oberth (assuming one-pass constraint). There's a minimum TWR to even capture on the first pass but I guess that TWR results in lower overall efficiency than utilizing Oberth more. Obviously very high TWR is wasteful by the nature of how much engine must be carried (and fuel to move that engine).

If I said that every 100 units of thrust cost 0.1 tons. What would be the ideal TWR for a Mun capture with 1000m/s escape energy? Too little TWR and you have to burn at Oberth-poor distances. Too much TWR and the engine weight penalty is too great.

[Specialist290]

By the way, the fact that fuel tanks do have mass also means there is a maximum delta-v for any engine:...

Hurrah for drop tanks!

Boo for interplanetary debris!

Well, now you people have done it. I'm blaming all of you for my latest silly idea. Specifically: What if we made the drop tanks reusable?

Properly, I suppose, they wouldn't really be "drop tanks" anymore, but the principle is the same: I'm picturing a large nuclear "booster" in orbit which other ships can dock to. The booster pushes the fully-fueled craft that's acting as its payload past escape velocity and onwards to its destination, then undocks, drops itself back under escape velocity, and aerobrakes back into LKO, where it's refueled for the next mission.

[loknar]

Well, now you people have done it. I'm blaming all of you for my latest silly idea. Specifically: What if we made the drop tanks reusable?

Properly, I suppose, they wouldn't really be "drop tanks" anymore, but the principle is the same: I'm picturing a large nuclear "booster" in orbit which other ships can dock to. The booster pushes the fully-fueled craft that's acting as its payload past escape velocity and onwards to its destination, then undocks, drops itself back under escape velocity, and aerobrakes back into LKO, where it's refueled for the next mission.

What is the benefit of this? You would need to use almost as much fuel to fall back into Kerban as you would making a return trip. Maybe I missed something. I bring the thug with me on other planets. Something needs to bring back the mission crew anyway, right?

[SpaceSphereOfDeath]

^ because it looks cool.

And more importantly, you can set it up so that you can propel up to 19 tonnes to Duna with a burn time of 28 seconds!

(At least mine can)

[ComradeGoat]

Boo for interplanetary debris!

I have, on occasion, dropped them at strategic points so that they will clean themselves up. After completing the plane change burn to Jool, for example: set Joolian periapsis up to hit Jool, drop empty tanks, fix periapsis.

But mostly I just litter interplanetary space with empty orange tanks.

[tavert]

What is the benefit of this? You would need to use almost as much fuel to fall back into Kerban as you would making a return trip. Maybe I missed something. I bring the thug with me on other planets. Something needs to bring back the mission crew anyway, right?

You can aerobrake it back down nearly for free, as long as you don't go past the edge of the SoI with the reusable part of your transfer stage. This concept was proposed for Project Troy by Reaction Engines http://en.wikipedia.org/wiki/Reaction_Engines_Limited#Reaction_Engines_TROY, and probably a few people before them. You use an Earth (or Kerbin) departure stage to get up just below escape, separate your remaining interplanetary stage and finish the transfer just with it. The final interplanetary burn probably needs to happen the next time around at periapsis, for best Oberth benefit. The Earth/Kerbin departure stage never leaves the SoI, you bring it back nearly empty with aerobraking, and refuel it in low orbit. For extra reusability, make your refueler an SSTO (Skylon style).

Edited October 2, 2013 by tavert

[maltesh]

There's a minimum TWR to even capture on the first pass

To make a minor point, if you've got the fuel for it, there isn't a minimum acceleration required to capture on the first pass. Nothing prevents you from starting your capture burn outside the SOI of your target, if necessary. Though crossing an SOI boundary while thrusting does kill the throttle.

(Well, except boredom, the fact that the game doesn't give relative-velocity information on Celestial Body Targets, and the difficulty of plotting maneuver nodes on future Conic Patches if you haven't changed the nearly-useless default CONIC_PATCH_DRAW_MODE)

[Technical Ben]

Well, now you people have done it. I'm blaming all of you for my latest silly idea. Specifically: What if we made the drop tanks reusable?

Properly, I suppose, they wouldn't really be "drop tanks" anymore, but the principle is the same: I'm picturing a large nuclear "booster" in orbit which other ships can dock to. The booster pushes the fully-fueled craft that's acting as its payload past escape velocity and onwards to its destination, then undocks, drops itself back under escape velocity, and aerobrakes back into LKO, where it's refueled for the next mission.

Re-usable drop tanks? I already have a large array of Orange dockable tanks I'm slowly expanding out of Kerbin orbit. I say slowly, I have 5 tankers set to deliver to 4 or 5 planets. During the wait for the burn (50 days off for the closest) I'm moving my tanks out from orbit, to the mun then to a high orbit. Filling them up with replacements as I burn through the fuel.

[loknar]

You can aerobrake it back down nearly for free, as long as you don't go past the edge of the SoI with the reusable part of your transfer stage. This concept was proposed for Project Troy by Reaction Engines http://en.wikipedia.org/wiki/Reaction_Engines_Limited#Reaction_Engines_TROY, and probably a few people before them. You use an Earth (or Kerbin) departure stage to get up just below escape, separate your remaining interplanetary stage and finish the transfer just with it. The final interplanetary burn probably needs to happen the next time around at periapsis, for best Oberth benefit. The Earth/Kerbin departure stage never leaves the SoI, you bring it back nearly empty with aerobraking, and refuel it in low orbit. For extra reusability, make your refueler an SSTO (Skylon style).

OK I get it now. Wow that is insane ! lol

[magnemoe]

Well, now you people have done it. I'm blaming all of you for my latest silly idea. Specifically: What if we made the drop tanks reusable?

Properly, I suppose, they wouldn't really be "drop tanks" anymore, but the principle is the same: I'm picturing a large nuclear "booster" in orbit which other ships can dock to. The booster pushes the fully-fueled craft that's acting as its payload past escape velocity and onwards to its destination, then undocks, drops itself back under escape velocity, and aerobrakes back into LKO, where it's refueled for the next mission.

You can have reusable drop tanks, has used it on an lander who needed extra fuel to land on Vall. Drop tanks on fighter jets are mostly used like this anyway.

For an transfer ship you can have an extra fuel module for longer burn.

This requires kethane mining, without it you have to launch new tanks anyway.

Space junk outside of LKO is not an issue anyway.

[ComradeGoat]

Re-usable drop tanks? I already have a large array of Orange dockable tanks I'm slowly expanding out of Kerbin orbit. I say slowly, I have 5 tankers set to deliver to 4 or 5 planets. During the wait for the burn (50 days off for the closest) I'm moving my tanks out from orbit, to the mun then to a high orbit. Filling them up with replacements as I burn through the fuel.

When I've tried setting off from the edge of Kerbin's SOI to Jool, I've noticed that even fully fuelled, it's still less efficient than burning direct from LKO because of Oberth. YMMV, I guess.

[Kasuha]

When I've tried setting off from the edge of Kerbin's SOI to Jool, I've noticed that even fully fuelled, it's still less efficient than burning direct from LKO because of Oberth. YMMV, I guess.

Theoretically, you don't need much dv in high Kerbin orbit to drop to an elliptical orbit with low PA. And there you could burn towards the destination making the most from Oberth. I think I should try that.

[Specialist290]

Huh. Not only has my silly idea been tried, but people have done studies on something similar in real life. Looks like I have some more interesting reading ahead of me.

[Kronikor]

The nuclear engine is hands down the most efficient engine for vacuum travel, and one of the worst for atmospheric thrust.

[Meithan]

As far as optimization math goes. I'm very curious where the maximum efficiency point comes in with regards to TWR vs dV.

tavert already solved this problem in the most general way using optimization software. His analysis blows mine out of the water completely:

tavert's fantastic (and very complete) engine efficiency analysis

The first half of the graphs are for vacuum applications, and each graph has a different TWR restrictions. Just pick your TWR restriction, look up the payload mass and your desired delta-v on the axes, and the color of the intersection point on the graph will tell you which engine is more efficient for those parameters. His analysis was done in full detail so he's including as many engines of each type to meet the TWR restriction and he's considering the fact that fuel comes in discrete lumps (individual fuel tanks). In the graphs I've shown I'm assuming fuel amount can be increased continuously, which is unrealistic.

[aeronaut]

Well I was about to post a short analysis but ^^this^^ blows mines out of the water!

[ComradeGoat]

Theoretically, you don't need much dv in high Kerbin orbit to drop to an elliptical orbit with low PA. And there you could burn towards the destination making the most from Oberth. I think I should try that.

The problem with that is that you will likely miss your launch window while waiting until the right point in your (lengthy) orbit round Kerbin to drop your periapsis.

[Cygnus]

Hurrah for drop tanks!

There is a feasibility limit to this as well, and you can figure out the maximum possible dV with regards to the number of drop tanks you have, by using a summation equation for orange tanks plus 1 LV-N engine: 9.81*800*sum{ln[(36x+2.25)/(36x-29.75)]}

With x=1 to 10 the dV is 29538

with x=1 to 100 the dV is 45527 The dV didn't even double from increasing the number of fuel tanks by 10 times.

going on to 1000 orange tanks only gets you to 61582 dV, so while it can go on to infinity, you will need ridiculous amounts of fuel tanks to get you a higher dV.

Yuo can do this yourself at http://www.wolframalpha.com/input/?i=sums and input the equation as ln((36x+2.25)/(36x-29.75))*800*9.81 and start at 1 and end at whatever number of drop tanks you want.

[loknar]

There is a feasibility limit to this as well, and you can figure out the maximum possible dV with regards to the number of drop tanks you have, by using a summation equation for orange tanks plus 1 LV-N engine: 9.81*800*sum{ln[(36x+2.25)/(36x-29.75)]}

With x=1 to 10 the dV is 29538

with x=1 to 100 the dV is 45527 The dV didn't even double from increasing the number of fuel tanks by 10 times.

going on to 1000 orange tanks only gets you to 61582 dV, so while it can go on to infinity, you will need ridiculous amounts of fuel tanks to get you a higher dV.

Yuo can do this yourself at http://www.wolframalpha.com/input/?i=sums and input the equation as ln((36x+2.25)/(36x-29.75))*800*9.81 and start at 1 and end at whatever number of drop tanks you want.

Really? The limit is mathematical and not feasibility?

Whackjob, accept this challenge.. we can start by building a liquid Nitrogen cooled 12GHz i7 Extreme with 1 Terabyte of RAM and 6-way Raid 0 Solid State disks, for start.

[Cygnus]

Really? The limit is mathematical and not feasibility?

Whackjob, accept this challenge.. we can start by building a liquid Nitrogen cooled 12GHz i7 Extreme with 1 Terabyte of RAM and 6-way Raid 0 Solid State disks, for start.

As I said, the equation goes on infinitely, so the only real speed limit is c, the speed of light. And just for fun, I calculated the number of tanks that are need to reach 1% the speed of light. I couldn't use the website that I linked because it's servers couldn't handle sums above 1E7 but I was able to extrapolate the number to 1E186 orange fuel tanks. That would mean the starting mass is at 3.6E190 kg, which is more than the mass of the universe... And I worry that I made a mistake somewhere so feel free to correct me.

[ComradeGoat]

There is a feasibility limit to this as well, and you can figure out the maximum possible dV with regards to the number of drop tanks you have, by using a summation equation for orange tanks plus 1 LV-N engine: 9.81*800*sum{ln[(36x+2.25)/(36x-29.75)]}

With x=1 to 10 the dV is 29538

with x=1 to 100 the dV is 45527 The dV didn't even double from increasing the number of fuel tanks by 10 times.

going on to 1000 orange tanks only gets you to 61582 dV, so while it can go on to infinity, you will need ridiculous amounts of fuel tanks to get you a higher dV.

Scott Kerman to James T. Kerman, "dammit Jim, she cannae take much more without implementing resources and mining!"

[Frederf]

tavert already solved this problem in the most general way using optimization software. His analysis blows mine out of the water completely:

tavert's fantastic (and very complete) engine efficiency analysis

The first half of the graphs are for vacuum applications, and each graph has a different TWR restrictions. Just pick your TWR restriction, look up the payload mass and your desired delta-v on the axes, and the color of the intersection point on the graph will tell you which engine is more efficient for those parameters. His analysis was done in full detail so he's including as many engines of each type to meet the TWR restriction and he's considering the fact that fuel comes in discrete lumps (individual fuel tanks). In the graphs I've shown I'm assuming fuel amount can be increased continuously, which is unrealistic.

As impressive as this analysis is, it misses the mark on my specific question. DeltaV is given as an independent variable. In my optimization problem I'm considering that TWR affects the total dV required for a maneuver by requiring that impulse to be made at non-optimal times in the orbit. For example if a 100km flyby capture might take 1000dV if done with an impulse burn, more relaxed (non-infinite) TWR requirements will require >1000dV. The problem boils down to a fixed maneuver and the cheapest (say, start mass) method of achieving it. I guess what I'd really need is the fractional inefficiency of non-impulse burns compared with fractional inefficiency of more engine mass.

[aeronaut]

As impressive as this analysis is, it misses the mark on my specific question. DeltaV is given as an independent variable. In my optimization problem I'm considering that TWR affects the total dV required for a maneuver by requiring that impulse to be made at non-optimal times in the orbit. For example if a 100km flyby capture might take 1000dV if done with an impulse burn, more relaxed (non-infinite) TWR requirements will require >1000dV. The problem boils down to a fixed maneuver and the cheapest (say, start mass) method of achieving it. I guess what I'd really need is the fractional inefficiency of non-impulse burns compared with fractional inefficiency of more engine mass.

As long as you keep the burn relatively short you aren't going to see a noticeable efficiency drop. If you relax the single burn constraint from your previous post you could shorten your burn duration and achieve a pseudo impulsive burn by completing several smaller burns over a series of orbits; this is probably the easiest way to achieve the greatest efficiency.

[tavert]

As impressive as this analysis is, it misses the mark on my specific question. DeltaV is given as an independent variable. In my optimization problem I'm considering that TWR affects the total dV required for a maneuver by requiring that impulse to be made at non-optimal times in the orbit. For example if a 100km flyby capture might take 1000dV if done with an impulse burn, more relaxed (non-infinite) TWR requirements will require >1000dV. The problem boils down to a fixed maneuver and the cheapest (say, start mass) method of achieving it. I guess what I'd really need is the fractional inefficiency of non-impulse burns compared with fractional inefficiency of more engine mass.

Name a maneuver, you could calculate something to the effect of these http://forum.kerbalspaceprogram.com/threads/39812-Landing-and-Takeoff-Delta-V-vs-TWR-and-specific-impulse - it'll require numerically solving a nonlinear differential equation.

[Frederf]

The point of my issue is that the burn is not short enough to be well-approximated by the instantaneous impulse model. I'm choosing single-event as a given to model such needs as capture, gravity assist, rendezvous, limited life support, etc. where multi-burn isn't an option. I'm interested in multi-burn maneuvers especially scheduling them to have the final burn coincide with an exact time, but that's another day. Eventually I'd like to expand "extended burn first order corrections" to plan low-thrust spiral burns with ion engines for example.

I'm not too afraid of ODEs but I was hoping someone with more current experience would provide easier insight. I think that looking at it from a orbital energy viewpoint might be clearer.