Efficient atmospheric launches: More staging or less drag?

[antbin]

I've been playing around with designing atmospheric landers (that parachute to a landing on Kerbin, then relaunch to orbit), and one big question I've had is: how large should the fuel tanks be?

Many small tanks means they can be staged and dropped quickly once empty. Few large tanks means fewer parts and less total drag (at least according to the part selector window). Which is more fuel efficient? Should I ditch small draggy tanks early and carry the big one to orbit, or ditch the huge big tank early on?

To narrow the question a bit, assume:

* ships are dropping fuel tanks only - the same engines are used the whole launch

* orbit is 100km

* launch speed is controlled by mechjeb to be 'optimal'

* ship might have less than 2 TWR on launch

I ran a test with the same rocket, different staging order, and found the difference small. Dropping the large tank first saved 40L of fuel to orbit (out of 5600L on the pad). So is it a moot point?

Edited September 16, 2012 by antbin

[Nadrek]

To do these tests, you'll want to set Mechjeb Auto-throttle off and your throttle to 100%, go back to 0.15.2, or just wait for 0.17, since the 0.16 fuel bug (25% fuel at 50% thrust, 1% fuel at 10% thrust, etc.) will really do a number on any attempts at efficiency tests.

[Tommygun]

Drag seems to be very forgiving in the game, but weight isn't.

I would go with shedding weight as fast as possible.

I also think aerospikes are OK for stages up to 100k at Kerbin.

[UmbralRaptor]

The main reason for frequent staging would be a better mass ratio. 0.16 parts don't have the limited TWR or poor single stage mass ratios of 0.7.3 - 0.15.2 ones.

So, I made a spreadsheet to look at various drop tank configurations. The 'lander' is assumed to have a 1-kerbal pod, 1 ASAS, 7 parachutes, 8 landing legs, and an aerospike. The nominal 16 FL-T200 tanks is equivalent to one FL-T32, or 3200 fuel units. Each staging is assumed to require 0.05 tonnes in decouplers. The results are useful, but not overwhelming:

(This is the most efficient staging, though it's not always possible while keeping the ship symmetrical)

Bigger rockets benefit more -- with 1600 fuel units, drop tanks were rather pointless.

[Tommygun]

(This is the most efficient staging, though it's not always possible while keeping the ship symmetrical)

Bigger rockets benefit more -- with 1600 fuel units, drop tanks were rather pointless.

With your 4 and 5 stage rockets, are you using some stages as side strap on booster arrangements?

[UmbralRaptor]

They're all side mounts. Let me throw together some mockups. Like I said, the exact tankage amounts aren't always achievable. Also, some of the numbers were just to see how bad the diminishing returns for more/smaller droptanks got.

These might not be the best way of going about this.

edit: ...and I have some of the parts (eg: the engine) in the wrong stages. That'll need to be changed...

[Apotheosist]

The drag values for the parts to not add up if you use more parts. The drag values shown are relative to the mass of the part. This is especially shown when you notice that both the big and small parachutes have the same drag values. However, the large parachute has three times the mass of the small one, and so it has three times the drag.

Edited September 18, 2012 by Apotheosist

[EndlessWaves]

Yeah, so the advantage of larger tanks is additional stiffness from fewer connections and a smaller overall part count.

[antbin]

the 0.16 fuel bug (25% fuel at 50% thrust, 1% fuel at 10% thrust, etc.) will really do a number on any attempts at efficiency tests.

I am using the hacked fuel bug fix, which I hope will work

The drag values for the parts to not add up if you use more parts. The drag values shown are relative to the mass of the part.

That is... totally not obvious. Thank you. Based on this understanding, then I guess the added decouplers/fuel lines are the only drag difference between tiny tank-fests and big hulking drums.