Are Single Large Payloads Always More Efficient Than Multiple Smaller Payloads

[wisnoskij]

Are larger payloads always more efficient per mass than smaller ones? Is it always better to do a single assent (assuming a reasonable profile), or can it be more cost effective to do multiple ascents and then join in orbit? Is the only real limit the part count your computer can handle and your ability to keep the a rocket from wobbliness because of its large length, or is their a mathematical point where splitting a payload into multiple launches is more efficient?

[James M]

Two things you need to decide. Are you going to use a rocket or a shuttle? And if you're going to use a rocket, are you going to save the rocket after expending its fuel or not? If you can afford to take everything up and not waste building materials then by all means do it. Just make sure that the fuel you use to get your payloads up there doesn't end up costing more than the payload will be worth. 

[Aegolius13]

47 minutes ago, wisnoskij said:

Are larger payloads always more efficient per mass than smaller ones? Is it always better to do a single assent (assuming a reasonable profile), or can it be more cost effective to do multiple ascents and then join in orbit? Is the only real limit the part count your computer can handle and your ability to keep the a rocket from wobbliness because of its large length, or is their a mathematical point where splitting a payload into multiple launches is more efficient?

There are two ways to look at this one; purely mathematically and from a practical KSP design perspective.  

First, the math - you may have seen info on here about the Tsiolvovksy Rocket Equation.  If not you'll definitely want to check it out if you're interested in these kind of issues.   Basically, a rocket's delta-v (cumulative ability to change speed, and hence, go from one orbit to another) is a function of wet mass (fuel), dry mass (everything that's not fuel) and specific impulse (engine efficiency).  Staging is also a factor, since it allows a rocket to ditch dry mass once it's no longer needed, and to move to higher specific impulse engines.  

So, assuming hypothetically that you can scale everything equally, if you want to double the payload mass but keep the same delta-v, you'd end up doubling the total mass of the rocket.  So there is no inherent greater or lesser efficiency from splitting a payload.  It should be fairly easy to picture this - imagine one rocket with half your total payload.  You can send two of these up by themselves, or you could staple two of those rockets together, but they'd perform essentially the same. 

 

Now from a practical perspective, you're not really holding everything else equal when you scale up a rocket to increase a payload.  You're probably using different engines, you're staging differently, and so forth.  In general, these differences mean that it's generally more efficient to send up one big payload.  A few reasons:

• In KSP, engines generally get more efficient (both in terms of thrust-to-weight ratio and specific impulse) as they get bigger.  So using one big engine will likely end up better than several smaller engines.  
• You don't have to double up on certain one-per-rocket parts like probe cores, so you may get some mass savings here. 
• If you can stack your two payloads vertically, this will probably improve your aerodynamic performance and reduce drag.  Basically, the more mass you have versus a given surface area, the better your rocket will cut through the atmosphere (the technical term to look at is "ballistic coeffificent").  Note that. conversely, this also makes bigger stuff harder to deborbit or aerobrake.  
• Finally, you won't have to pack whatever hardware you were going to need to dock your various payloads. 

 

All that said, there are a few cases when splitting up payloads may be better.  If your full craft is really un-aerodynamic (think something like the International Space Station), you may be better splitting it up into different pieces.  Another reason might be if your bigger rocket hits some really awkward breakpoints, like it's too heavy for one Mammoth but two is overkill.  But you can generally design around this issue by adding or subtracting fuel so suit each stage's thrust.   

[Reusables]

1 minute ago, Aegolius13 said:

In KSP, engines generally get more efficient (both in terms of thrust-to-weight ratio and specific impulse) as they get bigger.  So using one big engine will likely end up better than several smaller engines.  

This, and bigger fuel tanks are cheaper as well. (Nearly every parts in ksp obey this 'economy of scale'.

[Wanderfound]

The rocket stuff above is good and accurate, but in KSP you might also want to consider the influence of SSTO lifters.

A good SSTO Mk3 spaceplane of moderate size can lift about fifty ton to LKO for around √130/ton. Similarly, SSTO tankers can deliver fuel to orbit at trivial expense, opening the option of launching your spacecraft unfuelled (or using the transfer fuel to lift and then refuelling before departure), extending your lift abilities massively.

A compact lander and nuke transfer ship with all the science trimmings can be sent on a Moho return trip for about √80,000 (√70,000 of which is the disposable payload) if you lift it to LKO via one of those spaceplanes, refuel the lifter with a tanker and then use the refuelled lifter to boost the payload on its way.

With SSTO lifters, your limiting factor is usually bulk rather than mass. But you can build compact if you try...

About 11,000m/s ΔV in that nuke+lander payload. Putting the same cargo up on a rocket would cost more than ten times as much.

This is an area where KSP sharply diverges from reality, BTW. Kerbin's roughly 1/3 orbital speed relative to Earth makes SSTO spaceplanes a lot more practical.

On Kerbin, after your jets take you to 1,500m/s, you only need enough rocketry for another 2,000m/s. On Earth, even if you can jet up to 1,500, you've still got another 6,500 to go. Carrying enough fuel for 6,500m/s makes getting to 1,500m/s on jets very difficult.

Edited March 27, 2017 by Wanderfound

[bewing]

I think one of the most important things that didn't get mentioned exactly is single stack vs. multiple stack. (It kind of goes into the category of "inconvenient scaling".) A rocket built all in a single stack is going to be much more efficient to launch than one with some boosters hanging off the sides. So the moment that increasing the payload will cause you to start adding boosters -- you may be better off stopping there, and launching what you have.

 

[FancyMouse]

Another reason for splitting rocket payload is that it enables you to design different trajectories for each piece. It might turn out to be useful for interplanetary aerocapture (where different pieces have different heat tolerance).

[tomf]

I'm going to disagree with everyone else here, the cost of engines goes up much faster that there thrust and for a large rocket can make up most of the cost. Firing in atmosphere as a first stage The lv 45 costs only 7.1 per kN, the mainsail 9.4 and the mamoth 10.4. That difference is considerably larger than the fuel tank difference and means that you can get a better cost to orbit using several launches. 

Ineeficiencies are gong to come in due to the higher proportional drag of smaller rockets, the difficulties of perfectly splitting payloads onto the appropriate sized launcher, the couplings required to put them back together, and by far the most important to me, the cost in time of making multiple launches.

[wisnoskij]

46 minutes ago, tomf said:

Inefficiencies are gong to come in due to the higher proportional drag of smaller rockets, the difficulties of perfectly splitting payloads onto the appropriate sized launcher, the couplings required to put them back together, and by far the most important to me, the cost in time of making multiple launches.

This question in general, I think, and definitely my intent, was mainly about payloads that approach what KSP/my computer seem able to handle. Rockets so tall/wide that stock parts, the vab, and the physics just barely seem capable of handling fairly well. Splitting the payloads that I am talking about up into multiple normal launches that are easy to steer, create, and run smoother, will only make the project easier and quicker.

[Kryxal]

There's also the "efficient in what sense?" question.  Multiple smaller launches may be relatively cheaper, but you'll end up spending more time launching stuff.  Is that time worth the possible extra cost?  Could you make up that extra by using the time you save to run a mission instead?  Conversely, would a single launch be too awkward?  Mind you, I've launched some pretty awkward stuff, sometimes by attaching TWO stacks of rockets...

[Freshmeat]

My personal experience is that every booster I have build has been cheaper per tonne to LKO the bigger it was, with the largest practical booster taking 330 t to orbit (220t to Minmus). I have yet to find a need to lift anything bigger, as I refuel everything in a Minmus orbit.

[wisnoskij]

Any idea on number for how less/more efficient a bigger vs smaller launch is. Any examples of projects you have done? I will post some before and after pictures as I convert my current project to a 2-4 launch one, and see how it turns out
Also, rate my design, I have been doing this weird hybrid asparagus staging that I think should be fairly efficient, but as far as I am aware I make up the design completely alone so I have never had any feedback on it and only have a few personal launches that I never tested alternatives to.

Here is my current design (1967t). It is a mining mission to the Mun (Fairing removed for image).
The payload (dry mass, with just enouhg fuel to land part of it and to start the mining operation): 115K kg, 312K cost.
From top to bottom: Science/surface explorer, Ore tank, Mono Tank, Rocket Fuel Tank, Miner, Lander, Station, The rocket to list it all to orbit
On paper it looks like it should work to get it to the Mun and Land it and some tested makes it seems like it is probably steerable.
The launch strategy is a little unique as well:
First off, I filled all the rocket fuel tanks in the payload, pretty much all it it will need to be transferred on route to the engines.
The plan was to get to into a 70K-69K orbit with the center rocket still intact and hopefully a decent amount of remaining fuel. Decouple the engine (and nothing else, it should eventually burn up in the atmosphere), decouple the lander, and redock it to the station (turned 90 degrees) and then redock that with the top bit of the payload and use the landers engines to blast for the Mun. with the stantion under it and the rest of the payload docked to its top.
Now back to the hybrid asparagus design. Well first off, I tweakscaled one of the kickback solid rocket boosters to be 3.75m, then I placed enough rocket fuel on top of it to fuel all of the engines for the time the solid rocket booster takes to burn out. So that is the first stage, then the orange tank stages will follow quickly after that, leaving me just the main rocket body.

Edited March 28, 2017 by wisnoskij

[CSE]

18 hours ago, wisnoskij said:

Is the only real limit the part count your computer can handle

There's also the question of what you, the player, are familiar with. If you've designed and flown something similar to your one-off large payload lifter, perhaps in sandbox practice or in previous games, then larger is better - most rockets will make orbit by a large margin in stock KSP so using a new design is not as daunting as it might be here on Earth.

But if you're playing a strict no-reverting-for-mistakes career and launching a payload you haven't tried before in an alternate planets pack, then it may be cheaper to split the payload so you can re-use one of your existing proven medium-lifter designs, rather than risk even one total-loss failed ascent. My own loss rate for scratch-built heavy lift rockets is, erm, ... not zero.

[Aegolius13]

10 hours ago, tomf said:

I'm going to disagree with everyone else here, the cost of engines goes up much faster that there thrust and for a large rocket can make up most of the cost. Firing in atmosphere as a first stage The lv 45 costs only 7.1 per kN, the mainsail 9.4 and the mamoth 10.4. That difference is considerably larger than the fuel tank difference and means that you can get a better cost to orbit using several launches. 

I thought OP was asking about efficiency in terms of mass fractions, so did not mention cost.  I've never crunched the numbers, but I still have to believe combining payloads should be cost competitive.  Sub-thought:

• Kickbacks seem to be just stupidly cheap for the amount of thrust they generate, even once you factor in the cost of the decoupler and nosecone.  But they don't work so well with little rockets. 
• TwinBoars are also just a ridiculously good deal for what you get.  
• The little LF engines are just... meh.  The Swivel is awful at sea level for a launch engine.  The Reliant, as you mention, is a good in terms of thrust per ton.  But since it doesn't thrust vector, unless you're a fantastic pilot, you'll probably need something additional to steer, which will add to the cost.  
• As you mention, the fuel does get a little cheaper per unit as you get bigger, which helps offset the cost increases of the bigger engines. 
• The Mainsail, Rhino, Mammoth and (especially) the Vector are not terribly cost-efficient.  But if you're a penny-pincher, you can do an awful lot with the TwinBoar and Kickback.

7 hours ago, wisnoskij said:

Also, rate my design, I have been doing this weird hybrid asparagus staging that I think should be fairly efficient, but as far as I am aware I make up the design completely alone so I have never had any feedback on it and only have a few personal launches that I never tested alternatives to.

Here is my current design (1967t). It is a mining mission to the Mun.

I may be doing something wrong, but can't see a high-res version of your screenshot.  That said, 2000 tons seems really large, but I imagine tweakscaled Kickbacks are quite heavy.  And first-stage mass is not that big of an issue.  

The top part looks draggy... you might consider putting a fairing around that for launch.  

[Reusables]

I did some math on cost efficiency of engines(including the fuel tanks), check it out.

KSP Engine Specifications

Generally bigger engines are better, with exception of the cheap Reliant.

[Wanderfound]

1 hour ago, Aegolius13 said:

The little LF engines are just... meh.  The Swivel is awful at sea level for a launch engine.  The Reliant, as you mention, is a good in terms of thrust per ton.  But since it doesn't thrust vector, unless you're a fantastic pilot, you'll probably need something additional to steer, which will add to the cost.  

The main feature of the Swivel and Reliant is that they're cheap.

Lifting a 30 ton Moho ship, my high-tech version uses a Mammoth and Kickbacks for the first stage, followed by a Rhino for circularisation, transfer and partial capture. That costs about √130,000 (not including payload) and is nice and simple and elegant:

The low-tech Mainsail/Thumper/Reliant->Poodle version, OTOH, is a mess of multi-layer asparagus:

 

...but the high-tech version is roughly double the price of the low-tech one.

OTOH, when I actually care about cost, I'll just lift and boost it with an SSTO (for about 1/10th the price of the low-tech rocket). And the high-tech rocket is much nicer to fly; Rhino-powered transfer burns are gloriously fast and precise.

Edited March 28, 2017 by Wanderfound

[wisnoskij]

56 minutes ago, Aegolius13 said:

I may be doing something wrong, but can't see a high-res version of your screenshot.  That said, 2000 tons seems really large, but I imagine tweakscaled Kickbacks are quite heavy.  And first-stage mass is not that big of an issue.  

The top part looks draggy... you might consider putting a fairing around that for launch.  

Fixed. The fairing was removed for picture taking.

57 minutes ago, Aegolius13 said:

• TwinBoars are also just a ridiculously good deal for what you get.

I see what you mean, I replaced a Mammoth with a tweakscaled TwinBoar and every metric was significantly improved. Unfortunately, I would consider that cheating (I only TS when their is no alternative).

And the Results are in.
Recap: A single 1967t $600K rocket (If anything I needed to add a little extra boost)
Payload split into two launches: 625t $466K (I am not sure If i am missing something or it really is that much more efficient)
I am really happy with these rockets, they were incredibly easy to build while the original was a huge Pain, I would add huge new stages and the dV would tick up a tiny amount.
I might add two kickbacks (or smaller) to the first one, but it probably does not need it. The second one needs fins and the booster's fuel tanks could actually be orange tanks, but that would be overkill, so I might try to balance that better. When the two payloads meet in orbit, I need the total equivalent of the payload of the second completely full. What we really need are procedural solid fuel boosters, balancing the thrust verses the burn length is really important, but we simply are given so few options. I am tweakscalling up the second solid booster because that results in a far shorter burn time, the kickback burns for far too long imho.

\