Optimal airless body lander research

[MBobrik]

Looking at the KSP dV maps, I've been always wondering from where they've got the numbers. Whether they are exact, or too big or too small.

Judging by how much dV I usually needed to go up and down, or travel between nearby planets, they were more or less correct ( except for moho ).

But were my landers optimal ? could one do much better ? After all, the theoretical limit of infinite TWR is hohmann to the surface + killing all the orbital velocity.

which is significantly less less than the given numbers.

Then, watching this tutorial

I realized that this landing maneuver is simple enough to compute the exact dv needed as an piloting skill-independent benchmark.

The initial short program was quickly done, taking thrust, ISP, dry mass and fuel mass and returning whether the ship ran out of fuel, crashed because of insufficient thrust, or what percentage of fuel it burned during landing.

Then I tried inverse approach - computing what amount of fuel were needed to land successfully, then I tried to vary the engines,

and soon realized that the continuous approach would not give any reasonable answers, because the real constraints were not some continuous trade-off function between thrust, mass, and ISP, but

by a discrete number of engines of a few predefined types.So I went to implement a full discrete optimizer that takes just parameters of the given celestial body, ( mass, diameter, minimum safe orbit ),and payload mass and spits out the optimal number and type of engines and fuel tanks, and real dv needed,

that gives the smallest spaceship mass, both for landing and ascent. and computes the real dV needed.

This way one can see that while it would be possible for example to land on the mun for the 640 dV according to the map. it is more mass-efficient to use less engines and saving weight, even though the dv spent will be higher.

Of course there are some limitations. first, this approach works only for airless worlds, second, my optimizer does compute only single stage landers/ascent vehicles and also does not try to combine different engine types together.

But nonetheless, the results look nice just as they are, especially as a graph of payload mass -> optimal ship for a given celestial body that shows how radically the optimal solution sometimes changes with changing payload mass.

.

So there are the outputs of my optimizer. enjoy.

.

Description :

Left strip is engine type and count. bar length is engine count and color engine type.

Right strip is fuel amount and is a little more complicated to interpret - each orange tank takes up 64 units X200-32 32 units of length and is colored differently and so on till torus and oscar which both take exactly one unit and differ from FL-T100 only by color.

On the vertical axis each bar represents increments of payload mass and so each horizontal line represents one solution - optimal amount and type of engines and fuel tanks giving the smallest vehicle mass for a given payload mass.

For example in the Eeloo Asc 1.0 - 10.0 Ton the exact 5.00 ton row says 1x 24-7S engine and 1x FL-T100 + 1x toroidal + 1x oscar fuel tank.

.

.

And now the raw glory of the data.

.

Bop http://imgur.com/a/XPbRg#0

.

Dres http://imgur.com/a/T7YNf#0

.

Eeloo http://imgur.com/a/wdIh4#0

.

Ike http://imgur.com/a/DEWqj#0

.

Minmus http://imgur.com/a/2xn9n#0

.

Moho http://imgur.com/a/PxfZp#0

.

Mun http://imgur.com/a/mSWgT#0

.

Pol http://imgur.com/a/TvdrZ#0

.

Tylo http://imgur.com/a/D3hp7#0

.

Vall http://imgur.com/a/lyY8u#0

Edited September 7, 2013 by MBobrik

[Johnno]

First, top marks for doing tables/calculations.

However, a simple delta-v map and your personal preference of lander fuel/engine (with TWR calculated by engineer, mechjeb or yourself) really is more than enough. You can easily overengineer a lander (as I show here) to have enough delta-v with few parts/mass changes.

As much as I applaud your contribution of going through all the effort of getting all this published, I shamelessly scrolled through all of your graphics knowing that it's too easy to throw together a optimized/overengineered lander to bother with any complex calculations or tables.

I guess what I'm trying to say is I appreciate your dedication, but at the same time, picking an engine, fuel tank, payload, matching them all up and sending them off is far easier than looking up a premade table (or doing the math) to see what would be optimal, for a computer game we play for entertainment.

[tavert]

Johnno, a lot of us enjoy making the absolute minimalist craft possible for any given mission.

MBobrik, I get what I'm seeing here, but you should really clean up the presentation. Put the images in an album that can be scrolled through more easily than a way-too-long forum post, make them a consistent size, etc. The payload axis should really be labeled with numerical values, as well.

Basically it seems like what you've done here is a combination of this: http://forum.kerbalspaceprogram.com/showthread.php/39812-Landing-and-Takeoff-Delta-V-vs-TWR-and-specific-impulse

and this: http://forum.kerbalspaceprogram.com/showthread.php/45155-Mass-optimal-engine-type-vs-delta-V-payload-and-min-TWR

It would be nice to see your source code, in a way that could be run interactively.

[codepoet]

Surely what matters in a lander isn't that you have minimized the delta-v that you use to land / take off, but the wetmass of the lander before it begins the landing. As has been mentioned, depending on the choice of engines etc it is possible to have a more massive lander with the same delta-v. What matters to me is that until we start the landing, the lander+fuel+tanking+engines are just cargo, and that cargo needs to be hauled into orbit and transfered to another body by earlier stages, all of which can be smaller if the cargo is smaller.

[Skorpychan]

Interesting stuff. It's certainly helped show me what needs to be tweaked with my tiny lander design.

Namely, a little more fuel.