Optimizing for cost

[Pecan]

By the way, I'm sorry I'm not actually contributing anything here yet but I'm trying to update my tutorial for 0.25.

[PB666]

1.5 g is wasteful coming off the pad (only one third of the rockets thrust is being applied to acceleration), one is wasting E = Mass*4.5m/s*T until one reaches a slower terminal velocity. Ideally acceleration is at structural limit and comes down to 2.0 by the time g-force = drag force (~100 m/s for .2 CoD, and then increases again slowly after 2400 alt.

I use SepratronI on horizontally mounted TT38 and scale-mod of RT-10 to make sure my launches get quickly to term, you can always take down liquid engines until boosters are expended. The RT-10 mods can be scaled in size and duration (without exploiting) in order to keep thrust. Alternatively, one can stagger their firing and because they are only used for a minute, they can be batteried on a single decoupler.

Most of the weight of a rocket is shed during the first minute of flight, inefficiencies in early flight are the most costly in terms of fuel and success. There is a plateau for pre-orbital ascent, g-forces between 2.7 and 3.0 can add efficiency but not by much, over 3.0 efficiency declines.

In addition one has to look at the ISP. For example the Main-sail is very powerful but has a low ISP, plowing this thing into space is not wise, however it is more efficient than a booster. Switiching to an Aerospike over 15,000 feet takes advantage of power (If you can find a way to mount the rocket into a stack). Sheding low ISP weight in low atmosphere and switching to high ISP after making gravity turn (with some speed boosts) gets one into orbit best.

There is also an advantage in slower launches to traveling 090 to about 25 m/s horizontal velocity before reaching gravity turn.

[Thanny]

I decided to try a few tanker designs out to see whether or not SRB's really do beat out SSTO for efficiency. The last tanker I used to load up my fuel depot contributed about 340t of fuel at a cost of about 10.5 funds/unit of LF (plus matching oxidizer). That's the cost of the burned fuel and the discarded SRB's, as the rest of the ship is recovered at 100%. I loaded up MechJeb with my sandbox save just to check if my ascent were reasonably efficient, and it put the same ship in orbit with about 500 units less fuel than I had with my last manual ascent (about a 1.5% difference).

I tried a handful of SSTO designs around the same basic shape - a central stack of Kerbodyne tanks on top of a KR-2L, surround by four or eight truss-attached additional Kerbodyne stacks on top of KS-25x4 engines. The number and size of tanks in each stack were varied, which changed total fuel amount and TWR. The best efficiency I got was about 11.9 funds/unit of delivered fuel, with a design that started at a TWR of 1.2. The same design with more or less fuel did worse. That most efficient SSTO design delivered only 120t of fuel, and had about a 35% higher launch cost.

I also took one of the designs that ended up at about 15.2 funds/unit (for about 120t fuel) and added a bunch of SRB's. Total deliverable fuel was 410t, at a cost of 10.4 funds/unit.

So I'm pretty convinced that chucking SRB's at the problem of large payloads is far more cost efficient than any SSTO. I don't know if I could even build a SSTO big enough to get the same fuel payload. And if I could, it'd have so many more parts that docking would be more painful than it already is.

[Wanderfound]

This thread may be of interest: http://forum.kerbalspaceprogram.com/threads/90337-Economic-Fuel-to-Oribit

[Norcalplanner]

This thread may be of interest: http://forum.kerbalspaceprogram.com/threads/90337-Economic-Fuel-to-Oribit

Agreed. My SSTO tanker plane was able to deliver 1.5 OTEs (orange tank equivalents) to orbit for a cost of less than 0.6 funds per unit (5514 funds for 9600 units of fuel and oxidizer).

http://forum.kerbalspaceprogram.com/threads/90337-Economic-Fuel-to-Oribit?p=1428382&viewfull=1#post1428382

[Progressm]

Reading this with much interest and I have to say (in the absence of actual facts to add to the debate), this discussion and the fact that there is no single right answer revealed yet is a tribute to the game by itself.

[Wanderfound]

Agreed. My SSTO tanker plane was able to deliver 1.5 OTEs (orange tank equivalents) to orbit for a cost of less than 0.6 funds per unit (5514 funds for 9600 units of fuel and oxidizer).

http://forum.kerbalspaceprogram.com/threads/90337-Economic-Fuel-to-Oribit?p=1428382&viewfull=1#post1428382

Similar for me: http://forum.kerbalspaceprogram.com/threads/90337-Economic-Fuel-to-Oribit?p=1361984&viewfull=1#post1361984

Wet cost: √149322

Recovered: √143321.7

LF delivered: 3727.9

O delivered: 4186.4

Cost: √6000.3

Splitting the bill evenly between the LF and O portions:

Price per unit of LF: √0.8

Price per unit of O: √0.7

[Norcalplanner]

Yep, once you have an orbital fuel depot that can be topped off cheaply, a lot of options open up. Upper stages can be launched with nearly empty tanks, so rocket lifter stages can be smaller and cheaper.

[sal_vager]

Interesting read Sivert3, stickied for November

[John FX]

I did something similar recently, SSTO to orbit with the aim of all the craft becoming system wide shuttles (recovering stages is hard with FAR) with standard fittings, modular etc

http://forum.kerbalspaceprogram.com/threads/97475-Standard-Utility-Construction-Kit

With the supplied craft you can build this easily.

[numerobis]

The largest SRB produces a bit more impulse than a Mainsail with 200X-16 and 200X-8 tanks, for less than 20% the cost. That SRB costs about 160% of the amount of LF+O that does the same job with a pre-existing engine, but when you factor in the need for more tanks and engines, as well as even more LF+O to lift the LF+O you're adding, disposable SRB's become quite cost effective.

The mass-optimal thing generally is that if you've got an engine on board, you're best off having it fire on the ascent if you can -- otherwise lower stages need more engines. And if on top of that, LF+O is cheaper than solid fuel, then for sure you'll want to be firing your bipropellant rocket engines from launch. Using the LF+O rockets all the way also means you can tweak your thrust in flight, as the SRBs empty out.

You'll need additional fuel to fire the rockets through the first stage. Will it be cheaper to take those extra fuel tanks all the way to orbit and back (remember to pack enough extra parachutes), or will it be cheaper to put the extra fuel on top of the SRBs use fuel lines, and dump them?

[Ruedii]

I've lately found that making the first stage solid rockets is often the most cost effective. Can you modify the script to handle that (with both types of boosters, of course.)

[Sivert3]

@sal_vager: Oh, thanks. I guess I'll have to keep up the work.

@Ruedii: If you've read the thread I've shown calculation using a pure solid fuel lower stage. However, as numerobis points out, if you have a liquid engine on board and you're not fireing it, that is wasting funds as the liquid fuel is cheaper than solid fuel. Calculating the best configuration of a mixed solid liquid stage through atmosphere is something I haven't tackled yet. But it's definetively on the plan.

Liquid Boosters

Dropping liquid engines in a stage on the way to orbit will never be cost effective as long as it can't be recovered. The cost of the engines alone will be more than the total cost of getting the payload to orbit on the efficient rockets presented here. Dropping fuel tanks on the way to orbit is not any better. For the FL-T200, dropping it costs 425 funds while bringing it back again costs about 91 funds. That over 335 funds for a gain of having 0.125 tonne less weight to bring up. With a cost efficency of 4.5 the cost of bring up 0.125 tonnes to orbit is 125 funds, so dropping it is 200 funds down the drain.

Efficency of various crafts people have posted

Please use comparable numbers when you post your craft. It's worth nothing to say that the rocket is pretty cost effective or state a number in combined oxygen fuel units to funds when the other guy uses funds to tonnes. The most sensible unit for cost efficency for lifting a payload to orbit is tonnes payload per unit funds. The most sensible unit for costs is funds per tonne of payload. There are of course many variation to this. The only advantage my unit has in this is that the resulting number is usually in the range 2-10 for a typical rocket.

Slugy

less than 1000 funds/tonne = Cost Efficency of above 4.5

Thanny

SRB assisted: 10.5 funds/liquid fuel = Cost Efficiency of 4.3

SSTO rocket: 11.9 funds/liquid fuel = Cost Efficiency of 3.8

Norcalplanner

SSTO plane: 0.6 funds/units of fuel = Cost Efficiency of 37.5

Wanderfound

SSTO plane: ~7900 units of fuel / 6000 funds = Cost Efficiency of 29.6

LKO payload conversion table

Conversion table from various units to to my Cost Efficency unit. See the first post in this thread for an explanation of the Cost Efficiency unit. Note that all of these are for a craft taking a payload to low Kerbin orbit using about 4500 delta v.

[TABLE=width: 500]

[TR]

[TD]From[/TD]

[TD]To[/TD]

[TD]Formula[/TD]

[/TR]

[TR]

[TD]funds/unit of fuel^{[1], [2]}[/TD]

[TD]Cost Efficency[/TD]

[TD]22.5 / (funds/unit of fuel)[/TD]

[/TR]

[TR]

[TD]funds/liquid fuel^[1][/TD]

[TD]Cost Efficency[/TD]

[TD]46 / (funds/liquid fuel)[/TD]

[/TR]

[TR]

[TD]funds/tonne[/TD]

[TD]Cost Efficency[/TD]

[TD]4500 / (funds/tonne)[/TD]

[/TR]

[TR]

[TD]tonnes/funds[/TD]

[TD]Cost Efficency[/TD]

[TD]4500 * tonnes/funds[/TD]

[/TR]

[TR]

[TD]Cost Efficency[/TD]

[TD]funds/tonne[/TD]

[TD]4500 / (Cost Efficency)[/TD]

[/TR]

[TR]

[TD]Cost Efficency[/TD]

[TD]tonnes/funds[/TD]

[TD](Cost Efficency) / 4500[/TD]

[/TR]

[/TABLE]

Notes

1. Only valid if the ratio between liquid fuel and oxygen is 110/90. The weight of the fuel tank is counted as part of the payload.
   
2. Units of fuel is the sum of liquid fuel units and liquid oxygen units added together.
   

Edited November 4, 2014 by Sivert3

[numerobis]

Nitpitcking: "tonne" not "tone"

Looking back I couldn't find a link to your script. Am I blind, or if not, could you share it?

[Sivert3]

@numerobis: *looks in a dictionary*, ehe. Oops. I've always had problems when the same letter is repeated in a word. Thanks for pointing it out, I've fixed all the "sounds" in my posts to tonnes. Also you're not blind as I have not yet shared the script.

Regarding the script. There are a lot of issues with it, which is the reason I have not posted it yet. It depends on mathplotlib, which took me over half an hour just to set up on my computer. It uses lots of hardcoded values. It only calculates staging for 2 stage rocket with a pure solid bottom stage and a liquid upper stage. It doesn't support anything else, i.e no aspargous staging, mixed solid and liquid stages or more than 2 stages. And the calculations are all based on isp at sea level. Until I can fix at least half of these issues it's not going to be very useful.

[numerobis]

Feel free to snatch the KSP-scripts to get accurate physics.

[MarvinKitFox]

Two comments on your rockets.

Firstly:

You want to leave the launchpad at a LOT more than TWR 1.5

You want to get up to something like terminal velocity as soon as possible, then trim your TWR to a steady 2.1 or so.

The way to do this is not by increasing primary engines, but by using a cluster of the smaller SRB's to kick your initial ascent, then eject them and sustain only on the primary lifters.

And secondly:

I spent around 10 complete rockets assemblies, and a few launch pads trying to get the SRBs from colliding into the core. The most troubleling being the Sepratrons destroying the core fuel tank. Ouch! Proper design and a bit of experience mitigates these risks greatly.

Seperatrons? to eject SRB's? how... wasteful!

Simply set up a moderate rate of spin in the 5 seconds before staging, and *fling* those empties off using centrifugal 'force'.

Virtually free, and very VERY reliable.

[Pecan]

You want to leave the launchpad at a LOT more than TWR 1.5

When optimising for mass, maybe, but not generally when optimising for cost. TWR of 1.2 takes more cheap fuel rather than expensive engines. I still haven't seen a moderate-sized rocket (>20t payload) that really benefits much from SRBs because it takes so many of them to give useful thrust, but then I'm optimising for part-count too.

[Dispatcher]

I've had good success in changing my designs so that I have the mission stage and a launch stage that puts it into orbit. The launch stage is jettisoned after achieving orbit, has a probe core and enough fuel/ o2 left for a de-orbit burn. Drogues and parachutes enable it to be returned intact and the recovery value is high. Its somewhat challenging and not always the most fuel-efficient way to fly, but no hardware is lost (except for the occasional SRBs used to augment the launch stage).

[John FX]

Lately I have been using a pure solid first stage staggering the thrusts so that I get a more level TWR over the burn.

For example I set the centre SRB at 100%, the next ring at 75-66% and the last ring at 50% (to make a TWR of 2.0 at launch)

Then my TWR rises to about 3.5 then drops to 3 with the centre engine cutout. Rises again to 3.75 then drops to 2 when the first ring cuts out. It saves some of the fuel used to push through the air and allows the later first stage burn to be more controllable with solid boosters as the first stage. I use a stage recovery mod so I fit them with chutes and recover 75% of the costs (excluding fuel)

Often I am left with less than 350Dv to burn for orbital insertion. With uncontrolled burns sometimes I have needed to use up to 1000Dv to circularise.

Very much like the proposed system from NASA for cheap launches that use a solid first stage. I can`t find the image now but it was basically a bunch of generic solid boosters with a set thrust and burn time where you just put as many as you need in a bunch for the desired TWR, recover them later and fill them up again. Very few parts so very reliable and cheap.

Also very heavy but for a first stage where the only cost is fuel, they work out very cost effective if the stage can be recovered.

IIRC they were to be made from steel to reduce costs...

[AmpsterMan]

I'm not sure if this has been suggested, but instead of a cost/dv formula, you can multiply the dv with the mass of the rocket, I.e. cost/dv-mass. This would normalize the different tiers of your rockets so they can all be directly compared.

[NBZ]

What I am seeing in this very interesting thread, is a bunch of RLV's being circularized before deorbiting.

Regardless of propulsion, wouldn't it be best to stage while coasting to apoapsis? This way, no dead weight needs to be circularized, which is especially important in KSP where engines and empty tanks are heavy compared to fuel.

Of course, one would have to act quickly to save the plummeting RLV after circularizing the payload. This in turn would require higher apoapsis or higher TWR on the last stage, but I am sure a balance could be struck.

[ola]

This is a very fascinating thread, full of good conclusions.

One thing which strikes me when I read this thread, however, is that many of the carefully crafted and well-balanced points become completely moot if you use a Proper drag model like FAR, spiced with a bit of DRE.

[Pecan]

What I am seeing in this very interesting thread, is a bunch of RLV's being circularized before deorbiting.

Regardless of propulsion, wouldn't it be best to stage while coasting to apoapsis? This way, no dead weight needs to be circularized, which is especially important in KSP where engines and empty tanks are heavy compared to fuel.

Of course, one would have to act quickly to save the plummeting RLV after circularizing the payload. This in turn would require higher apoapsis or higher TWR on the last stage, but I am sure a balance could be struck.

Your points are good, but generally the second one (got to circularise the payload before the sub-orbital LV falls too far back into atmosphere) is the critical one. It's a lot easier to fly if everything gets into orbit then you de-orbit what needs to be recovered. Secondly; a sub-orbital launch will leave whatever is recovered a long way from KSC, reducing its recovered value and the point of the whole exercise. If you want to land the recoverable launch-vehicle back at KSC (preferably on the pad/runway) it has to complete at least a single orbit anyway (almost) so there's no great extra expense/fuel required to place it in a stable orbit rather than 'not quite'.

A bit more arcane is where that circularisation fuel comes from - and by definition where the partly-empty tank ends up. If the LV carries a bit more fuel than neccessary just to make full orbit then so what? All it has to do later is a de-orbit/landing burn and its job is over. If the payload has to carry circularisation fuel instead then the excess tank mass has to be carried throughout its mission, which might be a long time and incur significant additional deltaV penalties.

So only circularising the (lighter) payload would be more efficient if it weren't for the fact that you often want to get the LV into orbit as well, for a better landing, and you want the payload 'full' for its mission. *Grin* In practice it'll be sufficiently marginal that there's lots of room for personal preference and play-style.